diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,135100 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "global_step": 22505,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.958579881656805e-08,
+      "loss": 5.2148,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.91715976331361e-08,
+      "loss": 5.7266,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.875739644970415e-08,
+      "loss": 5.5156,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.183431952662722e-07,
+      "loss": 5.5312,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4792899408284025e-07,
+      "loss": 5.5508,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.775147928994083e-07,
+      "loss": 5.8633,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0710059171597638e-07,
+      "loss": 5.7266,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.366863905325444e-07,
+      "loss": 5.2773,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.662721893491125e-07,
+      "loss": 5.0156,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.958579881656805e-07,
+      "loss": 5.3203,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.254437869822485e-07,
+      "loss": 5.5391,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.550295857988166e-07,
+      "loss": 5.5645,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 5.5312,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1420118343195276e-07,
+      "loss": 5.8086,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4378698224852073e-07,
+      "loss": 5.3516,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.733727810650888e-07,
+      "loss": 5.1328,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.029585798816569e-07,
+      "loss": 5.043,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.32544378698225e-07,
+      "loss": 6.4414,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.621301775147929e-07,
+      "loss": 4.8672,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.91715976331361e-07,
+      "loss": 4.9219,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.21301775147929e-07,
+      "loss": 5.4375,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.50887573964497e-07,
+      "loss": 5.5254,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.804733727810651e-07,
+      "loss": 5.0625,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.100591715976332e-07,
+      "loss": 4.7637,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.396449704142013e-07,
+      "loss": 5.1309,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 4.8906,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.988165680473374e-07,
+      "loss": 4.2148,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.284023668639055e-07,
+      "loss": 4.2578,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.579881656804734e-07,
+      "loss": 4.4004,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.875739644970415e-07,
+      "loss": 4.0938,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.171597633136095e-07,
+      "loss": 4.375,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.467455621301776e-07,
+      "loss": 4.2949,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.763313609467456e-07,
+      "loss": 4.1406,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0059171597633138e-06,
+      "loss": 4.0742,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0355029585798817e-06,
+      "loss": 3.8027,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06508875739645e-06,
+      "loss": 3.7188,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0946745562130179e-06,
+      "loss": 3.5684,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1242603550295859e-06,
+      "loss": 3.6914,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 3.5156,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.183431952662722e-06,
+      "loss": 3.4766,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.21301775147929e-06,
+      "loss": 3.3848,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.242603550295858e-06,
+      "loss": 3.2773,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2721893491124263e-06,
+      "loss": 3.2344,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.301775147928994e-06,
+      "loss": 3.3672,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3313609467455623e-06,
+      "loss": 3.2363,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3609467455621303e-06,
+      "loss": 3.2227,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3905325443786982e-06,
+      "loss": 3.1289,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4201183431952664e-06,
+      "loss": 3.0234,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4497041420118344e-06,
+      "loss": 2.9453,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4792899408284026e-06,
+      "loss": 2.873,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5088757396449705e-06,
+      "loss": 2.8359,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 2.7734,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5680473372781067e-06,
+      "loss": 2.6738,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5976331360946749e-06,
+      "loss": 2.5332,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6272189349112426e-06,
+      "loss": 2.5195,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.656804733727811e-06,
+      "loss": 2.3926,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6863905325443788e-06,
+      "loss": 2.4277,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7159763313609468e-06,
+      "loss": 2.3672,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.745562130177515e-06,
+      "loss": 2.3965,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.775147928994083e-06,
+      "loss": 2.3545,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.804733727810651e-06,
+      "loss": 2.2627,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.834319526627219e-06,
+      "loss": 2.3477,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8639053254437872e-06,
+      "loss": 2.2256,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8934911242603552e-06,
+      "loss": 2.0693,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 2.208,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.952662721893491e-06,
+      "loss": 2.1953,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9822485207100593e-06,
+      "loss": 2.1738,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0118343195266275e-06,
+      "loss": 2.252,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0414201183431953e-06,
+      "loss": 2.2041,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0710059171597635e-06,
+      "loss": 2.1445,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1005917159763316e-06,
+      "loss": 2.1279,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1301775147929e-06,
+      "loss": 2.1338,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1597633136094676e-06,
+      "loss": 2.1328,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1893491124260358e-06,
+      "loss": 2.248,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2189349112426035e-06,
+      "loss": 2.1953,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2485207100591717e-06,
+      "loss": 2.1436,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.27810650887574e-06,
+      "loss": 1.9707,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 2.1445,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.337278106508876e-06,
+      "loss": 2.123,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.366863905325444e-06,
+      "loss": 2.0977,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.396449704142012e-06,
+      "loss": 2.1123,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.42603550295858e-06,
+      "loss": 2.3203,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.455621301775148e-06,
+      "loss": 2.0078,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.485207100591716e-06,
+      "loss": 2.1699,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5147928994082845e-06,
+      "loss": 1.9863,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5443786982248527e-06,
+      "loss": 2.2988,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.57396449704142e-06,
+      "loss": 1.917,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.603550295857988e-06,
+      "loss": 1.9951,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6331360946745564e-06,
+      "loss": 2.0869,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6627218934911246e-06,
+      "loss": 2.0957,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 2.1035,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7218934911242605e-06,
+      "loss": 1.9639,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7514792899408287e-06,
+      "loss": 2.0146,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7810650887573965e-06,
+      "loss": 1.9951,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8106508875739646e-06,
+      "loss": 2.1045,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.840236686390533e-06,
+      "loss": 1.8779,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.869822485207101e-06,
+      "loss": 2.0713,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8994082840236688e-06,
+      "loss": 1.9453,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.928994082840237e-06,
+      "loss": 2.1064,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.958579881656805e-06,
+      "loss": 2.0098,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9881656804733733e-06,
+      "loss": 1.8564,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.017751479289941e-06,
+      "loss": 2.0098,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0473372781065093e-06,
+      "loss": 1.9814,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 2.0332,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.106508875739645e-06,
+      "loss": 1.8936,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1360946745562134e-06,
+      "loss": 1.9014,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1656804733727816e-06,
+      "loss": 1.8828,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1952662721893497e-06,
+      "loss": 2.0205,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.224852071005917e-06,
+      "loss": 1.8145,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2544378698224853e-06,
+      "loss": 1.9326,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2840236686390534e-06,
+      "loss": 1.9854,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.313609467455622e-06,
+      "loss": 1.9053,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3431952662721894e-06,
+      "loss": 2.0059,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3727810650887576e-06,
+      "loss": 1.8955,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4023668639053257e-06,
+      "loss": 1.8994,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4319526627218935e-06,
+      "loss": 1.8223,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.749,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.49112426035503e-06,
+      "loss": 1.833,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.520710059171598e-06,
+      "loss": 1.8135,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.550295857988166e-06,
+      "loss": 1.8057,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.579881656804734e-06,
+      "loss": 1.708,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.609467455621302e-06,
+      "loss": 1.7822,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6390532544378704e-06,
+      "loss": 1.7305,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.668639053254438e-06,
+      "loss": 1.9131,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6982248520710063e-06,
+      "loss": 1.7617,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7278106508875745e-06,
+      "loss": 1.8203,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7573964497041422e-06,
+      "loss": 1.7832,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7869822485207104e-06,
+      "loss": 1.7969,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.816568047337278e-06,
+      "loss": 1.8496,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.8203,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8757396449704146e-06,
+      "loss": 1.9053,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.905325443786982e-06,
+      "loss": 1.833,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.934911242603551e-06,
+      "loss": 1.8867,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.964497041420119e-06,
+      "loss": 1.8809,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9940828402366864e-06,
+      "loss": 1.7549,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.023668639053255e-06,
+      "loss": 1.6953,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.053254437869823e-06,
+      "loss": 1.8418,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0828402366863906e-06,
+      "loss": 1.8252,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.112426035502959e-06,
+      "loss": 1.8857,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.142011834319527e-06,
+      "loss": 1.8184,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1715976331360955e-06,
+      "loss": 1.9453,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.201183431952663e-06,
+      "loss": 1.7754,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.8105,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2603550295858e-06,
+      "loss": 1.7383,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.289940828402367e-06,
+      "loss": 1.9014,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.319526627218935e-06,
+      "loss": 1.7354,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.349112426035503e-06,
+      "loss": 1.7021,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3786982248520715e-06,
+      "loss": 1.7178,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.408284023668639e-06,
+      "loss": 1.6416,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.437869822485207e-06,
+      "loss": 1.8008,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.467455621301776e-06,
+      "loss": 1.668,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.497041420118343e-06,
+      "loss": 1.7744,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.526627218934911e-06,
+      "loss": 1.627,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.55621301775148e-06,
+      "loss": 1.7217,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5857988165680475e-06,
+      "loss": 1.7969,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.752,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.644970414201184e-06,
+      "loss": 1.6309,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.674556213017752e-06,
+      "loss": 1.8252,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.70414201183432e-06,
+      "loss": 1.6162,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.733727810650888e-06,
+      "loss": 1.5869,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.763313609467456e-06,
+      "loss": 1.707,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.792899408284024e-06,
+      "loss": 1.6377,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.822485207100592e-06,
+      "loss": 1.7451,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.85207100591716e-06,
+      "loss": 1.665,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8816568047337285e-06,
+      "loss": 1.7588,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.911242603550296e-06,
+      "loss": 1.6865,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940828402366865e-06,
+      "loss": 1.7344,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.970414201183432e-06,
+      "loss": 1.7158,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-06,
+      "loss": 1.7188,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.029585798816569e-06,
+      "loss": 1.7539,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.059171597633137e-06,
+      "loss": 1.6055,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.088757396449705e-06,
+      "loss": 1.7568,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.118343195266272e-06,
+      "loss": 1.5078,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.14792899408284e-06,
+      "loss": 1.627,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.177514792899409e-06,
+      "loss": 1.5303,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.207100591715976e-06,
+      "loss": 1.7061,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.236686390532545e-06,
+      "loss": 1.665,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.266272189349113e-06,
+      "loss": 1.6172,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.295857988165681e-06,
+      "loss": 1.543,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.325443786982249e-06,
+      "loss": 1.541,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.355029585798817e-06,
+      "loss": 1.6055,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.5156,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.414201183431953e-06,
+      "loss": 1.6582,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.443786982248521e-06,
+      "loss": 1.6035,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.47337278106509e-06,
+      "loss": 1.5908,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.502958579881657e-06,
+      "loss": 1.6494,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.532544378698226e-06,
+      "loss": 1.5,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.562130177514793e-06,
+      "loss": 1.5312,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.591715976331361e-06,
+      "loss": 1.6123,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.621301775147929e-06,
+      "loss": 1.6699,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.650887573964498e-06,
+      "loss": 1.5244,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.680473372781066e-06,
+      "loss": 1.5771,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.710059171597634e-06,
+      "loss": 1.5947,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.739644970414202e-06,
+      "loss": 1.6914,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 1.7334,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7988165680473375e-06,
+      "loss": 1.6094,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.828402366863905e-06,
+      "loss": 1.4404,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.857988165680474e-06,
+      "loss": 1.4766,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.887573964497042e-06,
+      "loss": 1.7256,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.91715976331361e-06,
+      "loss": 1.7578,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.946745562130178e-06,
+      "loss": 1.667,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.976331360946747e-06,
+      "loss": 1.7568,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0059171597633135e-06,
+      "loss": 1.46,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.035502958579882e-06,
+      "loss": 1.7207,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.06508875739645e-06,
+      "loss": 1.5801,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0946745562130185e-06,
+      "loss": 1.5488,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.124260355029586e-06,
+      "loss": 1.5586,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.5928,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.183431952662723e-06,
+      "loss": 1.6641,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.21301775147929e-06,
+      "loss": 1.5566,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.242603550295858e-06,
+      "loss": 1.6582,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.272189349112427e-06,
+      "loss": 1.4844,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.3017751479289945e-06,
+      "loss": 1.5586,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.331360946745563e-06,
+      "loss": 1.6123,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.360946745562131e-06,
+      "loss": 1.6123,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.3905325443786995e-06,
+      "loss": 1.5732,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.420118343195266e-06,
+      "loss": 1.5459,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.449704142011834e-06,
+      "loss": 1.5205,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.479289940828403e-06,
+      "loss": 1.6221,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.5088757396449705e-06,
+      "loss": 1.4902,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 1.5488,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.568047337278107e-06,
+      "loss": 1.6045,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.5976331360946755e-06,
+      "loss": 1.5156,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.627218934911244e-06,
+      "loss": 1.5664,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.656804733727811e-06,
+      "loss": 1.6494,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.686390532544379e-06,
+      "loss": 1.625,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.715976331360947e-06,
+      "loss": 1.6064,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.745562130177515e-06,
+      "loss": 1.5664,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.775147928994084e-06,
+      "loss": 1.8545,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.8047337278106515e-06,
+      "loss": 1.6992,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.83431952662722e-06,
+      "loss": 1.7705,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.863905325443787e-06,
+      "loss": 1.6211,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.893491124260356e-06,
+      "loss": 1.5176,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.6016,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.952662721893492e-06,
+      "loss": 1.5166,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.98224852071006e-06,
+      "loss": 1.5469,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.011834319526628e-06,
+      "loss": 1.4502,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.041420118343196e-06,
+      "loss": 1.5361,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.071005917159763e-06,
+      "loss": 1.5762,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.100591715976332e-06,
+      "loss": 1.5938,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.130177514792899e-06,
+      "loss": 1.6895,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.159763313609468e-06,
+      "loss": 1.5684,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.189349112426036e-06,
+      "loss": 1.6562,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.218934911242604e-06,
+      "loss": 1.6719,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.248520710059173e-06,
+      "loss": 1.3613,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.278106508875741e-06,
+      "loss": 1.4927,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 1.5107,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.337278106508876e-06,
+      "loss": 1.4814,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.366863905325444e-06,
+      "loss": 1.4287,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.396449704142013e-06,
+      "loss": 1.5391,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.42603550295858e-06,
+      "loss": 1.7549,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.455621301775149e-06,
+      "loss": 1.6387,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.485207100591717e-06,
+      "loss": 1.6338,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.5147928994082845e-06,
+      "loss": 1.5938,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.544378698224852e-06,
+      "loss": 1.7148,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.573964497041421e-06,
+      "loss": 1.5859,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.603550295857989e-06,
+      "loss": 1.4727,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.633136094674556e-06,
+      "loss": 1.5312,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.662721893491125e-06,
+      "loss": 1.4189,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.4961,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.72189349112426e-06,
+      "loss": 1.5864,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.751479289940829e-06,
+      "loss": 1.4893,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.781065088757396e-06,
+      "loss": 1.5254,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.810650887573965e-06,
+      "loss": 1.6113,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.840236686390533e-06,
+      "loss": 1.5293,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.869822485207102e-06,
+      "loss": 1.4521,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.89940828402367e-06,
+      "loss": 1.4727,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.928994082840237e-06,
+      "loss": 1.5479,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.958579881656804e-06,
+      "loss": 1.5889,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.988165680473373e-06,
+      "loss": 1.584,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.017751479289941e-06,
+      "loss": 1.5293,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.04733727810651e-06,
+      "loss": 1.5186,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 1.5186,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.106508875739646e-06,
+      "loss": 1.5225,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.136094674556214e-06,
+      "loss": 1.5879,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.165680473372781e-06,
+      "loss": 1.6123,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.19526627218935e-06,
+      "loss": 1.5498,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.224852071005918e-06,
+      "loss": 1.4648,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.254437869822485e-06,
+      "loss": 1.4814,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.284023668639054e-06,
+      "loss": 1.5215,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.313609467455622e-06,
+      "loss": 1.5801,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.343195266272191e-06,
+      "loss": 1.5,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.372781065088758e-06,
+      "loss": 1.8613,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.402366863905327e-06,
+      "loss": 1.5039,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.431952662721893e-06,
+      "loss": 1.5684,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.6436,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.49112426035503e-06,
+      "loss": 1.5127,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.5207100591716e-06,
+      "loss": 1.5469,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.550295857988166e-06,
+      "loss": 1.4844,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.579881656804735e-06,
+      "loss": 1.5186,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.609467455621302e-06,
+      "loss": 1.4521,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.63905325443787e-06,
+      "loss": 1.4111,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.668639053254439e-06,
+      "loss": 1.6289,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.698224852071006e-06,
+      "loss": 1.4814,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.727810650887574e-06,
+      "loss": 1.355,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.757396449704143e-06,
+      "loss": 1.5762,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.786982248520712e-06,
+      "loss": 1.4131,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.816568047337279e-06,
+      "loss": 1.4834,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 1.4229,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.875739644970414e-06,
+      "loss": 1.4316,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.905325443786983e-06,
+      "loss": 1.5088,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.934911242603551e-06,
+      "loss": 1.5713,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.96449704142012e-06,
+      "loss": 1.4863,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.994082840236687e-06,
+      "loss": 1.4795,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.023668639053255e-06,
+      "loss": 1.4941,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.053254437869822e-06,
+      "loss": 1.4941,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.082840236686391e-06,
+      "loss": 1.668,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.11242603550296e-06,
+      "loss": 1.5693,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.142011834319528e-06,
+      "loss": 1.5547,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.171597633136095e-06,
+      "loss": 1.4492,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.201183431952664e-06,
+      "loss": 1.5654,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.4277,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.2603550295858e-06,
+      "loss": 1.5742,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.289940828402368e-06,
+      "loss": 1.4395,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.319526627218935e-06,
+      "loss": 1.4419,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.349112426035503e-06,
+      "loss": 1.4229,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.378698224852072e-06,
+      "loss": 1.4067,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.40828402366864e-06,
+      "loss": 1.5234,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.43786982248521e-06,
+      "loss": 1.5205,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.467455621301776e-06,
+      "loss": 1.4736,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.497041420118343e-06,
+      "loss": 1.4814,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.526627218934912e-06,
+      "loss": 1.6455,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.55621301775148e-06,
+      "loss": 1.4922,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.585798816568049e-06,
+      "loss": 1.5811,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 1.6016,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.644970414201184e-06,
+      "loss": 1.4697,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.674556213017751e-06,
+      "loss": 1.4619,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.70414201183432e-06,
+      "loss": 1.4019,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.733727810650888e-06,
+      "loss": 1.6094,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.763313609467457e-06,
+      "loss": 1.4331,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.792899408284024e-06,
+      "loss": 1.4917,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.822485207100593e-06,
+      "loss": 1.4404,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.852071005917161e-06,
+      "loss": 1.3965,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88165680473373e-06,
+      "loss": 1.5469,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.911242603550297e-06,
+      "loss": 1.5693,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.940828402366864e-06,
+      "loss": 1.457,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.970414201183432e-06,
+      "loss": 1.4902,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4414,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0029585798816568e-05,
+      "loss": 1.4971,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0059171597633138e-05,
+      "loss": 1.4209,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0088757396449705e-05,
+      "loss": 1.4912,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0118343195266274e-05,
+      "loss": 1.5107,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.014792899408284e-05,
+      "loss": 1.3887,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.017751479289941e-05,
+      "loss": 1.4932,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0207100591715978e-05,
+      "loss": 1.3789,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0236686390532545e-05,
+      "loss": 1.4561,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0266272189349113e-05,
+      "loss": 1.5625,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.029585798816568e-05,
+      "loss": 1.5225,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.032544378698225e-05,
+      "loss": 1.5068,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0355029585798817e-05,
+      "loss": 1.4111,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 1.4058,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0414201183431953e-05,
+      "loss": 1.4287,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.044378698224852e-05,
+      "loss": 1.4922,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.047337278106509e-05,
+      "loss": 1.5693,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0502958579881657e-05,
+      "loss": 1.4014,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0532544378698226e-05,
+      "loss": 1.3779,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0562130177514792e-05,
+      "loss": 1.4746,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0591715976331363e-05,
+      "loss": 1.6504,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.062130177514793e-05,
+      "loss": 1.4229,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0650887573964498e-05,
+      "loss": 1.7363,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0680473372781067e-05,
+      "loss": 1.5078,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0710059171597634e-05,
+      "loss": 1.499,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0739644970414202e-05,
+      "loss": 1.375,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.3975,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.079881656804734e-05,
+      "loss": 1.2646,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0828402366863907e-05,
+      "loss": 1.4512,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0857988165680475e-05,
+      "loss": 1.4082,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0887573964497042e-05,
+      "loss": 1.5557,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0917159763313609e-05,
+      "loss": 1.5264,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.094674556213018e-05,
+      "loss": 1.5195,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0976331360946746e-05,
+      "loss": 1.5234,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1005917159763315e-05,
+      "loss": 1.5117,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1035502958579882e-05,
+      "loss": 1.5586,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1065088757396452e-05,
+      "loss": 1.354,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1094674556213019e-05,
+      "loss": 1.4805,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1124260355029586e-05,
+      "loss": 1.5449,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 1.4502,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1183431952662721e-05,
+      "loss": 1.4414,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1213017751479292e-05,
+      "loss": 1.4175,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1242603550295859e-05,
+      "loss": 1.4458,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1272189349112427e-05,
+      "loss": 1.3604,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1301775147928996e-05,
+      "loss": 1.5127,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1331360946745563e-05,
+      "loss": 1.4912,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1360946745562131e-05,
+      "loss": 1.3789,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1390532544378698e-05,
+      "loss": 1.4287,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1420118343195268e-05,
+      "loss": 1.4072,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1449704142011835e-05,
+      "loss": 1.439,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1479289940828404e-05,
+      "loss": 1.4678,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1508875739644971e-05,
+      "loss": 1.3574,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 1.5508,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1568047337278108e-05,
+      "loss": 1.374,
+      "step": 391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1597633136094675e-05,
+      "loss": 1.4766,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1627218934911244e-05,
+      "loss": 1.5215,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.165680473372781e-05,
+      "loss": 1.4629,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1686390532544381e-05,
+      "loss": 1.5859,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1715976331360948e-05,
+      "loss": 1.3955,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1745562130177515e-05,
+      "loss": 1.3926,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1775147928994083e-05,
+      "loss": 1.4307,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1804733727810652e-05,
+      "loss": 1.2856,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.183431952662722e-05,
+      "loss": 1.3896,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1863905325443787e-05,
+      "loss": 1.5312,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1893491124260356e-05,
+      "loss": 1.543,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 1.5684,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1952662721893493e-05,
+      "loss": 1.3174,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.198224852071006e-05,
+      "loss": 1.4199,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2011834319526627e-05,
+      "loss": 1.4014,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2041420118343197e-05,
+      "loss": 1.5234,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2071005917159764e-05,
+      "loss": 1.4717,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2100591715976333e-05,
+      "loss": 1.3994,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.21301775147929e-05,
+      "loss": 1.457,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.215976331360947e-05,
+      "loss": 1.4331,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2189349112426037e-05,
+      "loss": 1.2876,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2218934911242604e-05,
+      "loss": 1.6035,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2248520710059173e-05,
+      "loss": 1.4541,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.227810650887574e-05,
+      "loss": 1.4287,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.4189,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2337278106508877e-05,
+      "loss": 1.4961,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2366863905325445e-05,
+      "loss": 1.4219,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2396449704142012e-05,
+      "loss": 1.4023,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.242603550295858e-05,
+      "loss": 1.4204,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.245562130177515e-05,
+      "loss": 1.3105,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2485207100591716e-05,
+      "loss": 1.3613,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2514792899408285e-05,
+      "loss": 1.3428,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2544378698224854e-05,
+      "loss": 1.502,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2573964497041422e-05,
+      "loss": 1.4414,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2603550295857989e-05,
+      "loss": 1.3354,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2633136094674556e-05,
+      "loss": 1.3672,
+      "step": 427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2662721893491126e-05,
+      "loss": 1.3145,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 1.3906,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2721893491124262e-05,
+      "loss": 1.4434,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2751479289940829e-05,
+      "loss": 1.4336,
+      "step": 431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2781065088757399e-05,
+      "loss": 1.5566,
+      "step": 432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2810650887573966e-05,
+      "loss": 1.4268,
+      "step": 433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2840236686390533e-05,
+      "loss": 1.3994,
+      "step": 434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2869822485207101e-05,
+      "loss": 1.4248,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2899408284023668e-05,
+      "loss": 1.4087,
+      "step": 436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2928994082840239e-05,
+      "loss": 1.5156,
+      "step": 437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2958579881656806e-05,
+      "loss": 1.5459,
+      "step": 438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2988165680473374e-05,
+      "loss": 1.376,
+      "step": 439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3017751479289941e-05,
+      "loss": 1.4521,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.304733727810651e-05,
+      "loss": 1.5273,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 1.4131,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3106508875739645e-05,
+      "loss": 1.4395,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3136094674556214e-05,
+      "loss": 1.3115,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3165680473372782e-05,
+      "loss": 1.5068,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3195266272189351e-05,
+      "loss": 1.3159,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3224852071005918e-05,
+      "loss": 1.2773,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3254437869822488e-05,
+      "loss": 1.2207,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3284023668639055e-05,
+      "loss": 1.3794,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3313609467455622e-05,
+      "loss": 1.4883,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.334319526627219e-05,
+      "loss": 1.2852,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3372781065088758e-05,
+      "loss": 1.4521,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3402366863905328e-05,
+      "loss": 1.4058,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3431952662721895e-05,
+      "loss": 1.4209,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 1.4512,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.349112426035503e-05,
+      "loss": 1.4194,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3520710059171597e-05,
+      "loss": 1.5039,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3550295857988167e-05,
+      "loss": 1.5166,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3579881656804734e-05,
+      "loss": 1.2798,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3609467455621303e-05,
+      "loss": 1.3457,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.363905325443787e-05,
+      "loss": 1.4189,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.366863905325444e-05,
+      "loss": 1.4395,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3698224852071007e-05,
+      "loss": 1.3955,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3727810650887574e-05,
+      "loss": 1.4697,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3757396449704143e-05,
+      "loss": 1.5234,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3786982248520711e-05,
+      "loss": 1.5156,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.381656804733728e-05,
+      "loss": 1.4561,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.2905,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3875739644970417e-05,
+      "loss": 1.3662,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3905325443786984e-05,
+      "loss": 1.4209,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3934911242603551e-05,
+      "loss": 1.2754,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.396449704142012e-05,
+      "loss": 1.3574,
+      "step": 472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3994082840236686e-05,
+      "loss": 1.4336,
+      "step": 473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4023668639053257e-05,
+      "loss": 1.3208,
+      "step": 474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4053254437869824e-05,
+      "loss": 1.4346,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4082840236686392e-05,
+      "loss": 1.3906,
+      "step": 476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4112426035502959e-05,
+      "loss": 1.4473,
+      "step": 477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4142011834319526e-05,
+      "loss": 1.3867,
+      "step": 478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4171597633136096e-05,
+      "loss": 1.3364,
+      "step": 479
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4201183431952663e-05,
+      "loss": 1.5088,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 1.4189,
+      "step": 481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4260355029585799e-05,
+      "loss": 1.4463,
+      "step": 482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4289940828402369e-05,
+      "loss": 1.2886,
+      "step": 483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4319526627218936e-05,
+      "loss": 1.5146,
+      "step": 484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4349112426035503e-05,
+      "loss": 1.4336,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4378698224852072e-05,
+      "loss": 1.3857,
+      "step": 486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.440828402366864e-05,
+      "loss": 1.5273,
+      "step": 487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4437869822485209e-05,
+      "loss": 1.4111,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4467455621301776e-05,
+      "loss": 1.2632,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4497041420118346e-05,
+      "loss": 1.4702,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4526627218934913e-05,
+      "loss": 1.3687,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4556213017751481e-05,
+      "loss": 1.2671,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4585798816568048e-05,
+      "loss": 1.4316,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 1.3604,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4644970414201186e-05,
+      "loss": 1.4414,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4674556213017752e-05,
+      "loss": 1.4287,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4704142011834321e-05,
+      "loss": 1.2715,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4733727810650888e-05,
+      "loss": 1.3843,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4763313609467458e-05,
+      "loss": 1.293,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4792899408284025e-05,
+      "loss": 1.4443,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4822485207100592e-05,
+      "loss": 1.4336,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.485207100591716e-05,
+      "loss": 1.4033,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4881656804733728e-05,
+      "loss": 1.2974,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4911242603550298e-05,
+      "loss": 1.3438,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4940828402366865e-05,
+      "loss": 1.4189,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4970414201183433e-05,
+      "loss": 1.4082,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.2676,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5029585798816569e-05,
+      "loss": 1.3174,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5059171597633138e-05,
+      "loss": 1.4238,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5088757396449705e-05,
+      "loss": 1.3691,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5118343195266275e-05,
+      "loss": 1.4404,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5147928994082842e-05,
+      "loss": 1.2378,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.517751479289941e-05,
+      "loss": 1.2661,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5207100591715977e-05,
+      "loss": 1.2822,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5236686390532544e-05,
+      "loss": 1.4609,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5266272189349113e-05,
+      "loss": 1.271,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.529585798816568e-05,
+      "loss": 1.3457,
+      "step": 517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.532544378698225e-05,
+      "loss": 1.3838,
+      "step": 518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.535502958579882e-05,
+      "loss": 1.373,
+      "step": 519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.3975,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5414201183431952e-05,
+      "loss": 1.2412,
+      "step": 521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.544378698224852e-05,
+      "loss": 1.3833,
+      "step": 522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.547337278106509e-05,
+      "loss": 1.4707,
+      "step": 523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5502958579881658e-05,
+      "loss": 1.4229,
+      "step": 524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5532544378698227e-05,
+      "loss": 1.3711,
+      "step": 525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5562130177514792e-05,
+      "loss": 1.4814,
+      "step": 526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5591715976331364e-05,
+      "loss": 1.4561,
+      "step": 527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.562130177514793e-05,
+      "loss": 1.3838,
+      "step": 528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5650887573964498e-05,
+      "loss": 1.3359,
+      "step": 529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5680473372781066e-05,
+      "loss": 1.3398,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5710059171597635e-05,
+      "loss": 1.5146,
+      "step": 531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5739644970414204e-05,
+      "loss": 1.4619,
+      "step": 532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 1.4531,
+      "step": 533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.579881656804734e-05,
+      "loss": 1.417,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5828402366863906e-05,
+      "loss": 1.3848,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5857988165680475e-05,
+      "loss": 1.3418,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5887573964497043e-05,
+      "loss": 1.418,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.591715976331361e-05,
+      "loss": 1.4727,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.594674556213018e-05,
+      "loss": 1.2925,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5976331360946746e-05,
+      "loss": 1.5498,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6005917159763314e-05,
+      "loss": 1.4844,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6035502958579883e-05,
+      "loss": 1.4189,
+      "step": 542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.606508875739645e-05,
+      "loss": 1.4189,
+      "step": 543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.609467455621302e-05,
+      "loss": 1.3965,
+      "step": 544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6124260355029585e-05,
+      "loss": 1.3789,
+      "step": 545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 1.4082,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6183431952662723e-05,
+      "loss": 1.3311,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.621301775147929e-05,
+      "loss": 1.4033,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.624260355029586e-05,
+      "loss": 1.3887,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.627218934911243e-05,
+      "loss": 1.4004,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6301775147928997e-05,
+      "loss": 1.4023,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6331360946745562e-05,
+      "loss": 1.4131,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.636094674556213e-05,
+      "loss": 1.4951,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.63905325443787e-05,
+      "loss": 1.2422,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6420118343195268e-05,
+      "loss": 1.3242,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6449704142011837e-05,
+      "loss": 1.2725,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6479289940828405e-05,
+      "loss": 1.3477,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.650887573964497e-05,
+      "loss": 1.3408,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 1.459,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6568047337278108e-05,
+      "loss": 1.3994,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6597633136094676e-05,
+      "loss": 1.2524,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6627218934911245e-05,
+      "loss": 1.2979,
+      "step": 562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.665680473372781e-05,
+      "loss": 1.3491,
+      "step": 563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6686390532544382e-05,
+      "loss": 1.4331,
+      "step": 564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6715976331360947e-05,
+      "loss": 1.2573,
+      "step": 565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6745562130177516e-05,
+      "loss": 1.3228,
+      "step": 566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6775147928994085e-05,
+      "loss": 1.3223,
+      "step": 567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6804733727810653e-05,
+      "loss": 1.3096,
+      "step": 568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6834319526627222e-05,
+      "loss": 1.3311,
+      "step": 569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6863905325443787e-05,
+      "loss": 1.3936,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6893491124260356e-05,
+      "loss": 1.3984,
+      "step": 571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.2881,
+      "step": 572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6952662721893493e-05,
+      "loss": 1.251,
+      "step": 573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.698224852071006e-05,
+      "loss": 1.3789,
+      "step": 574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7011834319526627e-05,
+      "loss": 1.3618,
+      "step": 575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.70414201183432e-05,
+      "loss": 1.4482,
+      "step": 576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7071005917159764e-05,
+      "loss": 1.2568,
+      "step": 577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7100591715976332e-05,
+      "loss": 1.5127,
+      "step": 578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.71301775147929e-05,
+      "loss": 1.4199,
+      "step": 579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.715976331360947e-05,
+      "loss": 1.3091,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7189349112426038e-05,
+      "loss": 1.4141,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7218934911242603e-05,
+      "loss": 1.3906,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7248520710059172e-05,
+      "loss": 1.3867,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.727810650887574e-05,
+      "loss": 1.4717,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 1.3999,
+      "step": 585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7337278106508878e-05,
+      "loss": 1.377,
+      "step": 586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7366863905325447e-05,
+      "loss": 1.4131,
+      "step": 587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7396449704142012e-05,
+      "loss": 1.3506,
+      "step": 588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.742603550295858e-05,
+      "loss": 1.2788,
+      "step": 589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.745562130177515e-05,
+      "loss": 1.4209,
+      "step": 590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7485207100591718e-05,
+      "loss": 1.2993,
+      "step": 591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7514792899408286e-05,
+      "loss": 1.251,
+      "step": 592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7544378698224855e-05,
+      "loss": 1.4355,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7573964497041423e-05,
+      "loss": 1.4072,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.760355029585799e-05,
+      "loss": 1.2588,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7633136094674557e-05,
+      "loss": 1.4668,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7662721893491126e-05,
+      "loss": 1.2319,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.3867,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7721893491124263e-05,
+      "loss": 1.1978,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7751479289940828e-05,
+      "loss": 1.4209,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.77810650887574e-05,
+      "loss": 1.3032,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7810650887573965e-05,
+      "loss": 1.3799,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7840236686390534e-05,
+      "loss": 1.3867,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7869822485207103e-05,
+      "loss": 1.374,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7899408284023668e-05,
+      "loss": 1.3252,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.792899408284024e-05,
+      "loss": 1.3584,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7958579881656805e-05,
+      "loss": 1.4814,
+      "step": 607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7988165680473374e-05,
+      "loss": 1.3691,
+      "step": 608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8017751479289942e-05,
+      "loss": 1.2246,
+      "step": 609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.804733727810651e-05,
+      "loss": 1.4043,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 1.3188,
+      "step": 611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8106508875739645e-05,
+      "loss": 1.2236,
+      "step": 612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8136094674556213e-05,
+      "loss": 1.4727,
+      "step": 613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8165680473372782e-05,
+      "loss": 1.4053,
+      "step": 614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.819526627218935e-05,
+      "loss": 1.3984,
+      "step": 615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.822485207100592e-05,
+      "loss": 1.3008,
+      "step": 616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8254437869822488e-05,
+      "loss": 1.3281,
+      "step": 617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8284023668639056e-05,
+      "loss": 1.416,
+      "step": 618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.831360946745562e-05,
+      "loss": 1.3379,
+      "step": 619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.834319526627219e-05,
+      "loss": 1.3667,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.837278106508876e-05,
+      "loss": 1.376,
+      "step": 621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8402366863905327e-05,
+      "loss": 1.3594,
+      "step": 622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8431952662721896e-05,
+      "loss": 1.3345,
+      "step": 623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.4727,
+      "step": 624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.849112426035503e-05,
+      "loss": 1.4111,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.85207100591716e-05,
+      "loss": 1.4272,
+      "step": 626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8550295857988167e-05,
+      "loss": 1.3633,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8579881656804736e-05,
+      "loss": 1.3311,
+      "step": 628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8609467455621304e-05,
+      "loss": 1.335,
+      "step": 629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.863905325443787e-05,
+      "loss": 1.3086,
+      "step": 630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.866863905325444e-05,
+      "loss": 1.27,
+      "step": 631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8698224852071007e-05,
+      "loss": 1.3516,
+      "step": 632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8727810650887575e-05,
+      "loss": 1.4668,
+      "step": 633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8757396449704144e-05,
+      "loss": 1.4287,
+      "step": 634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8786982248520713e-05,
+      "loss": 1.4121,
+      "step": 635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.881656804733728e-05,
+      "loss": 1.4043,
+      "step": 636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 1.2358,
+      "step": 637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.887573964497042e-05,
+      "loss": 1.418,
+      "step": 638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8905325443786984e-05,
+      "loss": 1.3208,
+      "step": 639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8934911242603552e-05,
+      "loss": 1.4912,
+      "step": 640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.896449704142012e-05,
+      "loss": 1.3711,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8994082840236686e-05,
+      "loss": 1.3643,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9023668639053258e-05,
+      "loss": 1.4014,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9053254437869823e-05,
+      "loss": 1.458,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9082840236686392e-05,
+      "loss": 1.2842,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911242603550296e-05,
+      "loss": 1.3574,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9142011834319526e-05,
+      "loss": 1.2197,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9171597633136098e-05,
+      "loss": 1.3052,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9201183431952663e-05,
+      "loss": 1.3184,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.3809,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.92603550295858e-05,
+      "loss": 1.3721,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.928994082840237e-05,
+      "loss": 1.2642,
+      "step": 652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9319526627218937e-05,
+      "loss": 1.2925,
+      "step": 653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9349112426035502e-05,
+      "loss": 1.2168,
+      "step": 654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.937869822485207e-05,
+      "loss": 1.4648,
+      "step": 655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.940828402366864e-05,
+      "loss": 1.4053,
+      "step": 656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9437869822485208e-05,
+      "loss": 1.3662,
+      "step": 657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9467455621301777e-05,
+      "loss": 1.2412,
+      "step": 658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9497041420118345e-05,
+      "loss": 1.3887,
+      "step": 659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9526627218934914e-05,
+      "loss": 1.3662,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9556213017751483e-05,
+      "loss": 1.2754,
+      "step": 661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9585798816568048e-05,
+      "loss": 1.4629,
+      "step": 662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 1.5146,
+      "step": 663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9644970414201185e-05,
+      "loss": 1.3184,
+      "step": 664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9674556213017754e-05,
+      "loss": 1.4062,
+      "step": 665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9704142011834322e-05,
+      "loss": 1.2676,
+      "step": 666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9733727810650888e-05,
+      "loss": 1.3911,
+      "step": 667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.976331360946746e-05,
+      "loss": 1.4077,
+      "step": 668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9792899408284025e-05,
+      "loss": 1.3213,
+      "step": 669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9822485207100593e-05,
+      "loss": 1.5049,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9852071005917162e-05,
+      "loss": 1.3281,
+      "step": 671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9881656804733727e-05,
+      "loss": 1.231,
+      "step": 672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99112426035503e-05,
+      "loss": 1.375,
+      "step": 673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9940828402366864e-05,
+      "loss": 1.3604,
+      "step": 674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9970414201183433e-05,
+      "loss": 1.3193,
+      "step": 675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 1.4775,
+      "step": 676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999999896437613e-05,
+      "loss": 1.3467,
+      "step": 677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999958575044e-05,
+      "loss": 1.3564,
+      "step": 678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99999990679385e-05,
+      "loss": 1.2754,
+      "step": 679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999998343001794e-05,
+      "loss": 1.3545,
+      "step": 680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999997410940342e-05,
+      "loss": 1.3828,
+      "step": 681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999996271754164e-05,
+      "loss": 1.2803,
+      "step": 682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999492544328e-05,
+      "loss": 1.335,
+      "step": 683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999337200772e-05,
+      "loss": 1.417,
+      "step": 684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999991611447518e-05,
+      "loss": 1.4873,
+      "step": 685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999989643762704e-05,
+      "loss": 1.4219,
+      "step": 686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999987468953326e-05,
+      "loss": 1.4248,
+      "step": 687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999985087019426e-05,
+      "loss": 1.3779,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999982497961053e-05,
+      "loss": 1.3525,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999979701778263e-05,
+      "loss": 1.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999976698471113e-05,
+      "loss": 1.2832,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999997348803966e-05,
+      "loss": 1.4014,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999997007048398e-05,
+      "loss": 1.4014,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999966445804135e-05,
+      "loss": 1.3301,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999962614000205e-05,
+      "loss": 1.4141,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999958575072264e-05,
+      "loss": 1.3418,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999954329020406e-05,
+      "loss": 1.3369,
+      "step": 697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999949875844707e-05,
+      "loss": 1.208,
+      "step": 698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999945215545268e-05,
+      "loss": 1.3335,
+      "step": 699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999940348122184e-05,
+      "loss": 1.3521,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999935273575552e-05,
+      "loss": 1.4131,
+      "step": 701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999992999190548e-05,
+      "loss": 1.3784,
+      "step": 702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999924503112075e-05,
+      "loss": 1.2617,
+      "step": 703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999918807195457e-05,
+      "loss": 1.3398,
+      "step": 704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999912904155735e-05,
+      "loss": 1.4941,
+      "step": 705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999906793993036e-05,
+      "loss": 1.3467,
+      "step": 706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999900476707488e-05,
+      "loss": 1.1689,
+      "step": 707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999893952299216e-05,
+      "loss": 1.3164,
+      "step": 708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999887220768363e-05,
+      "loss": 1.3682,
+      "step": 709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999880282115063e-05,
+      "loss": 1.3545,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999873136339464e-05,
+      "loss": 1.4097,
+      "step": 711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999986578344171e-05,
+      "loss": 1.4023,
+      "step": 712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999858223421955e-05,
+      "loss": 1.3374,
+      "step": 713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999850456280353e-05,
+      "loss": 1.167,
+      "step": 714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999842482017068e-05,
+      "loss": 1.3477,
+      "step": 715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999834300632265e-05,
+      "loss": 1.5186,
+      "step": 716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999982591212611e-05,
+      "loss": 1.3164,
+      "step": 717
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999817316498785e-05,
+      "loss": 1.3008,
+      "step": 718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999980851375046e-05,
+      "loss": 1.4814,
+      "step": 719
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999799503881318e-05,
+      "loss": 1.4023,
+      "step": 720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999979028689155e-05,
+      "loss": 1.2461,
+      "step": 721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999780862781343e-05,
+      "loss": 1.3867,
+      "step": 722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999977123155089e-05,
+      "loss": 1.4443,
+      "step": 723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99997613932004e-05,
+      "loss": 1.4395,
+      "step": 724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999751347730068e-05,
+      "loss": 1.4653,
+      "step": 725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999741095140106e-05,
+      "loss": 1.3525,
+      "step": 726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999730635430727e-05,
+      "loss": 1.3184,
+      "step": 727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999971996860214e-05,
+      "loss": 1.4082,
+      "step": 728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999709094654576e-05,
+      "loss": 1.3242,
+      "step": 729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999698013588257e-05,
+      "loss": 1.3296,
+      "step": 730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999968672540341e-05,
+      "loss": 1.4019,
+      "step": 731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999675230100267e-05,
+      "loss": 1.2178,
+      "step": 732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999663527679076e-05,
+      "loss": 1.3564,
+      "step": 733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999651618140068e-05,
+      "loss": 1.374,
+      "step": 734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999639501483496e-05,
+      "loss": 1.3623,
+      "step": 735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999962717770961e-05,
+      "loss": 1.3135,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999614646818664e-05,
+      "loss": 1.375,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999960190881092e-05,
+      "loss": 1.4131,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999588963686638e-05,
+      "loss": 1.2979,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999575811446087e-05,
+      "loss": 1.3271,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999562452089548e-05,
+      "loss": 1.3623,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999548885617282e-05,
+      "loss": 1.3564,
+      "step": 742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999535112029583e-05,
+      "loss": 1.3525,
+      "step": 743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999952113132673e-05,
+      "loss": 1.2998,
+      "step": 744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999506943509014e-05,
+      "loss": 1.2749,
+      "step": 745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999949254857673e-05,
+      "loss": 1.4062,
+      "step": 746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999477946530175e-05,
+      "loss": 1.2847,
+      "step": 747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999946313736965e-05,
+      "loss": 1.3623,
+      "step": 748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999448121095467e-05,
+      "loss": 1.2632,
+      "step": 749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999432897707928e-05,
+      "loss": 1.4326,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999417467207358e-05,
+      "loss": 1.2773,
+      "step": 751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999401829594073e-05,
+      "loss": 1.4189,
+      "step": 752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999385984868395e-05,
+      "loss": 1.3506,
+      "step": 753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999369933030653e-05,
+      "loss": 1.2725,
+      "step": 754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999935367408118e-05,
+      "loss": 1.3203,
+      "step": 755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999337208020312e-05,
+      "loss": 1.2246,
+      "step": 756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999320534848392e-05,
+      "loss": 1.3013,
+      "step": 757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999303654565762e-05,
+      "loss": 1.3159,
+      "step": 758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999928656717278e-05,
+      "loss": 1.3047,
+      "step": 759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999269272669788e-05,
+      "loss": 1.4238,
+      "step": 760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999251771057147e-05,
+      "loss": 1.4097,
+      "step": 761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999234062335228e-05,
+      "loss": 1.3594,
+      "step": 762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999921614650439e-05,
+      "loss": 1.2856,
+      "step": 763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999198023565007e-05,
+      "loss": 1.3374,
+      "step": 764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999179693517454e-05,
+      "loss": 1.3789,
+      "step": 765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999916115636211e-05,
+      "loss": 1.25,
+      "step": 766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999914241209936e-05,
+      "loss": 1.3906,
+      "step": 767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999123460729588e-05,
+      "loss": 1.3096,
+      "step": 768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999104302253194e-05,
+      "loss": 1.208,
+      "step": 769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999084936670567e-05,
+      "loss": 1.3213,
+      "step": 770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999065363982116e-05,
+      "loss": 1.2734,
+      "step": 771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999045584188238e-05,
+      "loss": 1.1309,
+      "step": 772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999902559728935e-05,
+      "loss": 1.4297,
+      "step": 773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999005403285864e-05,
+      "loss": 1.2285,
+      "step": 774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998985002178194e-05,
+      "loss": 1.293,
+      "step": 775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998964393966768e-05,
+      "loss": 1.3848,
+      "step": 776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998943578652008e-05,
+      "loss": 1.2754,
+      "step": 777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999892255623435e-05,
+      "loss": 1.3672,
+      "step": 778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998901326714227e-05,
+      "loss": 1.2397,
+      "step": 779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999887989009208e-05,
+      "loss": 1.2974,
+      "step": 780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999885824636835e-05,
+      "loss": 1.5586,
+      "step": 781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998836395543486e-05,
+      "loss": 1.1904,
+      "step": 782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998814337617945e-05,
+      "loss": 1.2202,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998792072592178e-05,
+      "loss": 1.3472,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999876960046665e-05,
+      "loss": 1.2402,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998746921241827e-05,
+      "loss": 1.3535,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998724034918172e-05,
+      "loss": 1.3379,
+      "step": 787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999870094149617e-05,
+      "loss": 1.2773,
+      "step": 788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998677640976292e-05,
+      "loss": 1.3916,
+      "step": 789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999865413335902e-05,
+      "loss": 1.2969,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998630418644846e-05,
+      "loss": 1.3047,
+      "step": 791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998606496834255e-05,
+      "loss": 1.5215,
+      "step": 792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998582367927747e-05,
+      "loss": 1.4355,
+      "step": 793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998558031925817e-05,
+      "loss": 1.3335,
+      "step": 794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998533488828977e-05,
+      "loss": 1.2495,
+      "step": 795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999850873863773e-05,
+      "loss": 1.3086,
+      "step": 796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998483781352588e-05,
+      "loss": 1.3086,
+      "step": 797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999845861697407e-05,
+      "loss": 1.2529,
+      "step": 798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998433245502696e-05,
+      "loss": 1.3403,
+      "step": 799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999840766693899e-05,
+      "loss": 1.1069,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999838188128349e-05,
+      "loss": 1.3496,
+      "step": 801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999835588853672e-05,
+      "loss": 1.2773,
+      "step": 802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998329688699224e-05,
+      "loss": 1.2778,
+      "step": 803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998303281771537e-05,
+      "loss": 1.3354,
+      "step": 804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998276667754218e-05,
+      "loss": 1.2305,
+      "step": 805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998249846647813e-05,
+      "loss": 1.2817,
+      "step": 806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999822281845287e-05,
+      "loss": 1.3188,
+      "step": 807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998195583169964e-05,
+      "loss": 1.3149,
+      "step": 808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998168140799644e-05,
+      "loss": 1.3359,
+      "step": 809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998140491342488e-05,
+      "loss": 1.3545,
+      "step": 810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998112634799068e-05,
+      "loss": 1.2837,
+      "step": 811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998084571169953e-05,
+      "loss": 1.4209,
+      "step": 812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998056300455736e-05,
+      "loss": 1.2642,
+      "step": 813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999802782265699e-05,
+      "loss": 1.3809,
+      "step": 814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997999137774315e-05,
+      "loss": 1.4155,
+      "step": 815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99979702458083e-05,
+      "loss": 1.3008,
+      "step": 816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997941146759547e-05,
+      "loss": 1.3857,
+      "step": 817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997911840628657e-05,
+      "loss": 1.4414,
+      "step": 818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997882327416237e-05,
+      "loss": 1.3643,
+      "step": 819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997852607122895e-05,
+      "loss": 1.3506,
+      "step": 820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999782267974925e-05,
+      "loss": 1.2573,
+      "step": 821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997792545295923e-05,
+      "loss": 1.2031,
+      "step": 822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997762203763536e-05,
+      "loss": 1.3545,
+      "step": 823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997731655152715e-05,
+      "loss": 1.3081,
+      "step": 824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99977008994641e-05,
+      "loss": 1.3398,
+      "step": 825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997669936698322e-05,
+      "loss": 1.3672,
+      "step": 826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997638766856024e-05,
+      "loss": 1.2114,
+      "step": 827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999760738993785e-05,
+      "loss": 1.4707,
+      "step": 828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997575805944454e-05,
+      "loss": 1.2964,
+      "step": 829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999754401487649e-05,
+      "loss": 1.2012,
+      "step": 830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999751201673461e-05,
+      "loss": 1.3496,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997479811519484e-05,
+      "loss": 1.3271,
+      "step": 832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997447399231773e-05,
+      "loss": 1.2832,
+      "step": 833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997414779872153e-05,
+      "loss": 1.3877,
+      "step": 834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997381953441296e-05,
+      "loss": 1.2837,
+      "step": 835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999734891993989e-05,
+      "loss": 1.2197,
+      "step": 836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997315679368606e-05,
+      "loss": 1.2441,
+      "step": 837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997282231728143e-05,
+      "loss": 1.3594,
+      "step": 838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999724857701919e-05,
+      "loss": 1.4326,
+      "step": 839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997214715242444e-05,
+      "loss": 1.3662,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997180646398605e-05,
+      "loss": 1.3252,
+      "step": 841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997146370488384e-05,
+      "loss": 1.3135,
+      "step": 842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997111887512485e-05,
+      "loss": 1.3672,
+      "step": 843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999707719747163e-05,
+      "loss": 1.3071,
+      "step": 844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997042300366526e-05,
+      "loss": 1.3149,
+      "step": 845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997007196197904e-05,
+      "loss": 1.3506,
+      "step": 846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996971884966487e-05,
+      "loss": 1.3433,
+      "step": 847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999693636667301e-05,
+      "loss": 1.2554,
+      "step": 848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996900641318207e-05,
+      "loss": 1.1484,
+      "step": 849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999686470890282e-05,
+      "loss": 1.5352,
+      "step": 850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999682856942759e-05,
+      "loss": 1.249,
+      "step": 851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996792222893267e-05,
+      "loss": 1.3408,
+      "step": 852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99967556693006e-05,
+      "loss": 1.1621,
+      "step": 853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996718908650356e-05,
+      "loss": 1.4189,
+      "step": 854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996681940943286e-05,
+      "loss": 1.3633,
+      "step": 855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999664476618016e-05,
+      "loss": 1.3486,
+      "step": 856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999660738436175e-05,
+      "loss": 1.2725,
+      "step": 857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999656979548883e-05,
+      "loss": 1.1196,
+      "step": 858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996531999562174e-05,
+      "loss": 1.4473,
+      "step": 859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996493996582565e-05,
+      "loss": 1.2061,
+      "step": 860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996455786550798e-05,
+      "loss": 1.3857,
+      "step": 861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996417369467654e-05,
+      "loss": 1.2617,
+      "step": 862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996378745333937e-05,
+      "loss": 1.4102,
+      "step": 863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996339914150442e-05,
+      "loss": 1.3477,
+      "step": 864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996300875917977e-05,
+      "loss": 1.3486,
+      "step": 865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996261630637348e-05,
+      "loss": 1.4434,
+      "step": 866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999622217830937e-05,
+      "loss": 1.375,
+      "step": 867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996182518934858e-05,
+      "loss": 1.3203,
+      "step": 868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999614265251463e-05,
+      "loss": 1.2344,
+      "step": 869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996102579049523e-05,
+      "loss": 1.2222,
+      "step": 870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996062298540356e-05,
+      "loss": 1.4062,
+      "step": 871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996021810987965e-05,
+      "loss": 1.4131,
+      "step": 872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995981116393194e-05,
+      "loss": 1.2168,
+      "step": 873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999594021475688e-05,
+      "loss": 1.438,
+      "step": 874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995899106079875e-05,
+      "loss": 1.4053,
+      "step": 875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999585779036303e-05,
+      "loss": 1.1382,
+      "step": 876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995816267607198e-05,
+      "loss": 1.3105,
+      "step": 877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995774537813238e-05,
+      "loss": 1.2134,
+      "step": 878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999573260098202e-05,
+      "loss": 1.3105,
+      "step": 879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995690457114406e-05,
+      "loss": 1.4395,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999564810621127e-05,
+      "loss": 1.3037,
+      "step": 881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99956055482735e-05,
+      "loss": 1.3008,
+      "step": 882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999556278330196e-05,
+      "loss": 1.252,
+      "step": 883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999551981129755e-05,
+      "loss": 1.1836,
+      "step": 884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999547663226115e-05,
+      "loss": 1.4648,
+      "step": 885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999543324619366e-05,
+      "loss": 1.3789,
+      "step": 886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995389653095977e-05,
+      "loss": 1.312,
+      "step": 887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995345852969006e-05,
+      "loss": 1.3403,
+      "step": 888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995301845813654e-05,
+      "loss": 1.3428,
+      "step": 889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999525763163083e-05,
+      "loss": 1.126,
+      "step": 890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995213210421446e-05,
+      "loss": 1.2773,
+      "step": 891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995168582186433e-05,
+      "loss": 1.3716,
+      "step": 892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995123746926708e-05,
+      "loss": 1.3716,
+      "step": 893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99950787046432e-05,
+      "loss": 1.5361,
+      "step": 894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995033455336843e-05,
+      "loss": 1.1235,
+      "step": 895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994987999008573e-05,
+      "loss": 1.3096,
+      "step": 896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994942335659335e-05,
+      "loss": 1.3813,
+      "step": 897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999489646529007e-05,
+      "loss": 1.3237,
+      "step": 898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994850387901734e-05,
+      "loss": 1.3682,
+      "step": 899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994804103495274e-05,
+      "loss": 1.354,
+      "step": 900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994757612071655e-05,
+      "loss": 1.3223,
+      "step": 901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999471091363184e-05,
+      "loss": 1.2139,
+      "step": 902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994664008176787e-05,
+      "loss": 1.3091,
+      "step": 903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999461689570748e-05,
+      "loss": 1.3145,
+      "step": 904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994569576224888e-05,
+      "loss": 1.3921,
+      "step": 905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999452204972999e-05,
+      "loss": 1.3086,
+      "step": 906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994474316223774e-05,
+      "loss": 1.1631,
+      "step": 907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994426375707228e-05,
+      "loss": 1.3149,
+      "step": 908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999437822818134e-05,
+      "loss": 1.3213,
+      "step": 909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994329873647117e-05,
+      "loss": 1.3135,
+      "step": 910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994281312105553e-05,
+      "loss": 1.2222,
+      "step": 911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994232543557658e-05,
+      "loss": 1.3042,
+      "step": 912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994183568004437e-05,
+      "loss": 1.3545,
+      "step": 913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999413438544691e-05,
+      "loss": 1.3467,
+      "step": 914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999408499588609e-05,
+      "loss": 1.1934,
+      "step": 915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994035399323005e-05,
+      "loss": 1.2451,
+      "step": 916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999398559575868e-05,
+      "loss": 1.2646,
+      "step": 917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993935585194146e-05,
+      "loss": 1.4155,
+      "step": 918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999388536763044e-05,
+      "loss": 1.1865,
+      "step": 919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993834943068602e-05,
+      "loss": 1.231,
+      "step": 920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993784311509676e-05,
+      "loss": 1.2686,
+      "step": 921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993733472954712e-05,
+      "loss": 1.2715,
+      "step": 922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999368242740476e-05,
+      "loss": 1.1768,
+      "step": 923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993631174860885e-05,
+      "loss": 1.3672,
+      "step": 924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993579715324135e-05,
+      "loss": 1.1733,
+      "step": 925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993528048795587e-05,
+      "loss": 1.1309,
+      "step": 926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999347617527631e-05,
+      "loss": 1.2065,
+      "step": 927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993424094767374e-05,
+      "loss": 1.3672,
+      "step": 928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999337180726986e-05,
+      "loss": 1.1953,
+      "step": 929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993319312784855e-05,
+      "loss": 1.2959,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993266611313437e-05,
+      "loss": 1.3936,
+      "step": 931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993213702856704e-05,
+      "loss": 1.3164,
+      "step": 932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999316058741575e-05,
+      "loss": 1.2588,
+      "step": 933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999310726499168e-05,
+      "loss": 1.2988,
+      "step": 934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999305373558559e-05,
+      "loss": 1.2627,
+      "step": 935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999299999919859e-05,
+      "loss": 1.2505,
+      "step": 936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992946055831803e-05,
+      "loss": 1.3838,
+      "step": 937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992891905486337e-05,
+      "loss": 1.2744,
+      "step": 938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992837548163315e-05,
+      "loss": 1.251,
+      "step": 939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992782983863866e-05,
+      "loss": 1.3105,
+      "step": 940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992728212589118e-05,
+      "loss": 1.3271,
+      "step": 941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99926732343402e-05,
+      "loss": 1.2617,
+      "step": 942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992618049118263e-05,
+      "loss": 1.2197,
+      "step": 943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999256265692444e-05,
+      "loss": 1.1328,
+      "step": 944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999250705775988e-05,
+      "loss": 1.2617,
+      "step": 945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999245125162574e-05,
+      "loss": 1.2388,
+      "step": 946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999239523852317e-05,
+      "loss": 1.0615,
+      "step": 947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999233901845333e-05,
+      "loss": 1.3721,
+      "step": 948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999228259141739e-05,
+      "loss": 1.0869,
+      "step": 949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992225957416512e-05,
+      "loss": 1.2524,
+      "step": 950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992169116451875e-05,
+      "loss": 1.2363,
+      "step": 951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992112068524654e-05,
+      "loss": 1.2998,
+      "step": 952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999205481363603e-05,
+      "loss": 1.1646,
+      "step": 953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991997351787185e-05,
+      "loss": 1.3506,
+      "step": 954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999193968297932e-05,
+      "loss": 1.3086,
+      "step": 955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999188180721362e-05,
+      "loss": 1.2617,
+      "step": 956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991823724491285e-05,
+      "loss": 1.3174,
+      "step": 957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999176543481352e-05,
+      "loss": 1.29,
+      "step": 958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991706938181533e-05,
+      "loss": 1.3691,
+      "step": 959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991648234596534e-05,
+      "loss": 1.332,
+      "step": 960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999158932405974e-05,
+      "loss": 1.1807,
+      "step": 961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991530206572368e-05,
+      "loss": 1.3535,
+      "step": 962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991470882135647e-05,
+      "loss": 1.168,
+      "step": 963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991411350750804e-05,
+      "loss": 1.5117,
+      "step": 964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991351612419074e-05,
+      "loss": 1.3389,
+      "step": 965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991291667141687e-05,
+      "loss": 1.2427,
+      "step": 966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991231514919895e-05,
+      "loss": 1.3325,
+      "step": 967
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999117115575494e-05,
+      "loss": 1.1904,
+      "step": 968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991110589648068e-05,
+      "loss": 1.0874,
+      "step": 969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991049816600534e-05,
+      "loss": 1.3311,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99909888366136e-05,
+      "loss": 1.3049,
+      "step": 971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999092764968853e-05,
+      "loss": 1.3271,
+      "step": 972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999086625582659e-05,
+      "loss": 1.4326,
+      "step": 973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999080465502905e-05,
+      "loss": 1.4082,
+      "step": 974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990742847297187e-05,
+      "loss": 1.2119,
+      "step": 975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999068083263228e-05,
+      "loss": 1.3389,
+      "step": 976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990618611035616e-05,
+      "loss": 1.2393,
+      "step": 977
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990556182508483e-05,
+      "loss": 1.3281,
+      "step": 978
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990493547052175e-05,
+      "loss": 1.1987,
+      "step": 979
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990430704667986e-05,
+      "loss": 1.3325,
+      "step": 980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999036765535722e-05,
+      "loss": 1.3008,
+      "step": 981
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999030439912118e-05,
+      "loss": 1.251,
+      "step": 982
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999024093596118e-05,
+      "loss": 1.3428,
+      "step": 983
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990177265878534e-05,
+      "loss": 1.3291,
+      "step": 984
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999011338887456e-05,
+      "loss": 1.1641,
+      "step": 985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999004930495058e-05,
+      "loss": 1.2197,
+      "step": 986
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989985014107926e-05,
+      "loss": 1.2866,
+      "step": 987
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998992051634792e-05,
+      "loss": 1.1709,
+      "step": 988
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989855811671907e-05,
+      "loss": 1.2773,
+      "step": 989
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989790900081224e-05,
+      "loss": 1.3164,
+      "step": 990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989725781577214e-05,
+      "loss": 1.145,
+      "step": 991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998966045616123e-05,
+      "loss": 1.3428,
+      "step": 992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998959492383462e-05,
+      "loss": 1.3525,
+      "step": 993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998952918459874e-05,
+      "loss": 1.1348,
+      "step": 994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998946323845496e-05,
+      "loss": 1.2588,
+      "step": 995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989397085404642e-05,
+      "loss": 1.2881,
+      "step": 996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998933072544915e-05,
+      "loss": 1.2432,
+      "step": 997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998926415858987e-05,
+      "loss": 1.3789,
+      "step": 998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989197384828173e-05,
+      "loss": 1.2197,
+      "step": 999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998913040416544e-05,
+      "loss": 1.3555,
+      "step": 1000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989063216603066e-05,
+      "loss": 1.2344,
+      "step": 1001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988995822142434e-05,
+      "loss": 1.3369,
+      "step": 1002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998892822078495e-05,
+      "loss": 1.2422,
+      "step": 1003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988860412532006e-05,
+      "loss": 1.2832,
+      "step": 1004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988792397385008e-05,
+      "loss": 1.2334,
+      "step": 1005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988724175345367e-05,
+      "loss": 1.2529,
+      "step": 1006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988655746414497e-05,
+      "loss": 1.293,
+      "step": 1007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998858711059381e-05,
+      "loss": 1.4209,
+      "step": 1008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998851826788473e-05,
+      "loss": 1.3394,
+      "step": 1009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998844921828869e-05,
+      "loss": 1.3071,
+      "step": 1010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988379961807107e-05,
+      "loss": 1.2803,
+      "step": 1011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988310498441424e-05,
+      "loss": 1.2549,
+      "step": 1012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998824082819308e-05,
+      "loss": 1.3042,
+      "step": 1013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988170951063515e-05,
+      "loss": 1.2817,
+      "step": 1014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998810086705418e-05,
+      "loss": 1.4033,
+      "step": 1015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988030576166523e-05,
+      "loss": 1.3447,
+      "step": 1016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987960078402003e-05,
+      "loss": 1.3535,
+      "step": 1017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987889373762076e-05,
+      "loss": 1.3516,
+      "step": 1018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998781846224821e-05,
+      "loss": 1.3125,
+      "step": 1019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987747343861868e-05,
+      "loss": 1.2295,
+      "step": 1020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987676018604533e-05,
+      "loss": 1.2349,
+      "step": 1021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987604486477675e-05,
+      "loss": 1.2344,
+      "step": 1022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987532747482777e-05,
+      "loss": 1.375,
+      "step": 1023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987460801621324e-05,
+      "loss": 1.3076,
+      "step": 1024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998738864889481e-05,
+      "loss": 1.21,
+      "step": 1025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987316289304724e-05,
+      "loss": 1.2349,
+      "step": 1026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987243722852572e-05,
+      "loss": 1.3711,
+      "step": 1027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998717094953985e-05,
+      "loss": 1.3574,
+      "step": 1028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998709796936807e-05,
+      "loss": 1.332,
+      "step": 1029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987024782338742e-05,
+      "loss": 1.2461,
+      "step": 1030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986951388453378e-05,
+      "loss": 1.1475,
+      "step": 1031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986877787713506e-05,
+      "loss": 1.3271,
+      "step": 1032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986803980120642e-05,
+      "loss": 1.3223,
+      "step": 1033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986729965676323e-05,
+      "loss": 1.3154,
+      "step": 1034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986655744382078e-05,
+      "loss": 1.2676,
+      "step": 1035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986581316239443e-05,
+      "loss": 1.269,
+      "step": 1036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986506681249957e-05,
+      "loss": 1.2256,
+      "step": 1037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986431839415174e-05,
+      "loss": 1.2363,
+      "step": 1038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998635679073664e-05,
+      "loss": 1.3184,
+      "step": 1039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986281535215908e-05,
+      "loss": 1.2344,
+      "step": 1040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998620607285454e-05,
+      "loss": 1.3252,
+      "step": 1041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998613040365409e-05,
+      "loss": 1.2637,
+      "step": 1042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998605452761614e-05,
+      "loss": 1.3613,
+      "step": 1043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998597844474225e-05,
+      "loss": 1.3174,
+      "step": 1044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985902155034002e-05,
+      "loss": 1.2305,
+      "step": 1045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985825658492977e-05,
+      "loss": 1.2793,
+      "step": 1046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998574895512075e-05,
+      "loss": 1.2515,
+      "step": 1047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998567204491892e-05,
+      "loss": 1.3066,
+      "step": 1048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985594927889077e-05,
+      "loss": 1.2725,
+      "step": 1049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985517604032815e-05,
+      "loss": 1.209,
+      "step": 1050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985440073351738e-05,
+      "loss": 1.3008,
+      "step": 1051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985362335847455e-05,
+      "loss": 1.2607,
+      "step": 1052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998528439152157e-05,
+      "loss": 1.2314,
+      "step": 1053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99852062403757e-05,
+      "loss": 1.3203,
+      "step": 1054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985127882411467e-05,
+      "loss": 1.3955,
+      "step": 1055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985049317630487e-05,
+      "loss": 1.4834,
+      "step": 1056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9984970546034393e-05,
+      "loss": 1.27,
+      "step": 1057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984891567624816e-05,
+      "loss": 1.3848,
+      "step": 1058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998481238240339e-05,
+      "loss": 1.2871,
+      "step": 1059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984732990371755e-05,
+      "loss": 1.2559,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984653391531558e-05,
+      "loss": 1.2959,
+      "step": 1061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984573585884448e-05,
+      "loss": 1.2725,
+      "step": 1062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998449357343207e-05,
+      "loss": 1.4756,
+      "step": 1063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998441335417609e-05,
+      "loss": 1.2422,
+      "step": 1064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998433292811817e-05,
+      "loss": 1.2988,
+      "step": 1065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984252295259968e-05,
+      "loss": 1.3379,
+      "step": 1066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998417145560316e-05,
+      "loss": 1.458,
+      "step": 1067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998409040914942e-05,
+      "loss": 1.29,
+      "step": 1068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984009155900427e-05,
+      "loss": 1.2568,
+      "step": 1069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998392769585786e-05,
+      "loss": 1.2373,
+      "step": 1070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983846029023407e-05,
+      "loss": 1.3311,
+      "step": 1071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983764155398767e-05,
+      "loss": 1.3096,
+      "step": 1072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983682074985625e-05,
+      "loss": 1.3115,
+      "step": 1073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983599787785693e-05,
+      "loss": 1.3369,
+      "step": 1074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998351729380066e-05,
+      "loss": 1.3037,
+      "step": 1075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998343459303225e-05,
+      "loss": 1.2827,
+      "step": 1076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998335168548217e-05,
+      "loss": 1.27,
+      "step": 1077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983268571152132e-05,
+      "loss": 1.0811,
+      "step": 1078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983185250043863e-05,
+      "loss": 1.3633,
+      "step": 1079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998310172215909e-05,
+      "loss": 1.2637,
+      "step": 1080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998301798749954e-05,
+      "loss": 1.1846,
+      "step": 1081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982934046066945e-05,
+      "loss": 1.3447,
+      "step": 1082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998284989786305e-05,
+      "loss": 1.3418,
+      "step": 1083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982765542889595e-05,
+      "loss": 1.1714,
+      "step": 1084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998268098114833e-05,
+      "loss": 1.4209,
+      "step": 1085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982596212640997e-05,
+      "loss": 1.1919,
+      "step": 1086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982511237369363e-05,
+      "loss": 1.3477,
+      "step": 1087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998242605533518e-05,
+      "loss": 1.2446,
+      "step": 1088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982340666540217e-05,
+      "loss": 1.2275,
+      "step": 1089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998225507098624e-05,
+      "loss": 1.2036,
+      "step": 1090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982169268675024e-05,
+      "loss": 1.3042,
+      "step": 1091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982083259608348e-05,
+      "loss": 1.1812,
+      "step": 1092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981997043787986e-05,
+      "loss": 1.3008,
+      "step": 1093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981910621215732e-05,
+      "loss": 1.3516,
+      "step": 1094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998182399189337e-05,
+      "loss": 1.3467,
+      "step": 1095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981737155822702e-05,
+      "loss": 1.2617,
+      "step": 1096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981650113005514e-05,
+      "loss": 1.4277,
+      "step": 1097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981562863443622e-05,
+      "loss": 1.376,
+      "step": 1098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981475407138827e-05,
+      "loss": 1.3008,
+      "step": 1099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981387744092938e-05,
+      "loss": 1.1885,
+      "step": 1100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981299874307776e-05,
+      "loss": 1.3647,
+      "step": 1101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998121179778516e-05,
+      "loss": 1.3545,
+      "step": 1102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998112351452691e-05,
+      "loss": 1.3115,
+      "step": 1103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9981035024534854e-05,
+      "loss": 1.2656,
+      "step": 1104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980946327810834e-05,
+      "loss": 1.2344,
+      "step": 1105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980857424356682e-05,
+      "loss": 1.3042,
+      "step": 1106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980768314174237e-05,
+      "loss": 1.2661,
+      "step": 1107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980678997265347e-05,
+      "loss": 1.2808,
+      "step": 1108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998058947363186e-05,
+      "loss": 1.0918,
+      "step": 1109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980499743275633e-05,
+      "loss": 1.3193,
+      "step": 1110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980409806198523e-05,
+      "loss": 1.2354,
+      "step": 1111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980319662402392e-05,
+      "loss": 1.1074,
+      "step": 1112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980229311889107e-05,
+      "loss": 1.0088,
+      "step": 1113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980138754660544e-05,
+      "loss": 1.2471,
+      "step": 1114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980047990718575e-05,
+      "loss": 1.251,
+      "step": 1115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979957020065083e-05,
+      "loss": 1.3613,
+      "step": 1116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997986584270194e-05,
+      "loss": 1.3223,
+      "step": 1117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979774458631055e-05,
+      "loss": 1.3135,
+      "step": 1118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979682867854303e-05,
+      "loss": 1.3818,
+      "step": 1119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997959107037359e-05,
+      "loss": 1.1929,
+      "step": 1120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979499066190815e-05,
+      "loss": 1.2778,
+      "step": 1121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997940685530788e-05,
+      "loss": 1.3301,
+      "step": 1122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979314437726703e-05,
+      "loss": 1.332,
+      "step": 1123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979221813449196e-05,
+      "loss": 1.3145,
+      "step": 1124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979128982477273e-05,
+      "loss": 1.1089,
+      "step": 1125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979035944812857e-05,
+      "loss": 1.272,
+      "step": 1126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997894270045788e-05,
+      "loss": 1.4375,
+      "step": 1127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997884924941427e-05,
+      "loss": 1.3779,
+      "step": 1128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978755591683962e-05,
+      "loss": 1.377,
+      "step": 1129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978661727268898e-05,
+      "loss": 1.3027,
+      "step": 1130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997856765617102e-05,
+      "loss": 1.4033,
+      "step": 1131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997847337839228e-05,
+      "loss": 1.2271,
+      "step": 1132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978378893934627e-05,
+      "loss": 1.1772,
+      "step": 1133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978284202800017e-05,
+      "loss": 1.3008,
+      "step": 1134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978189304990416e-05,
+      "loss": 1.2764,
+      "step": 1135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997809420050779e-05,
+      "loss": 1.229,
+      "step": 1136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99779988893541e-05,
+      "loss": 1.2109,
+      "step": 1137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997790337153133e-05,
+      "loss": 1.2393,
+      "step": 1138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977807647041455e-05,
+      "loss": 1.2866,
+      "step": 1139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977711715886457e-05,
+      "loss": 1.2812,
+      "step": 1140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977615578068322e-05,
+      "loss": 1.3198,
+      "step": 1141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977519233589042e-05,
+      "loss": 1.1724,
+      "step": 1142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977422682450617e-05,
+      "loss": 1.2915,
+      "step": 1143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977325924655037e-05,
+      "loss": 1.2168,
+      "step": 1144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977228960204317e-05,
+      "loss": 1.1235,
+      "step": 1145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977131789100458e-05,
+      "loss": 1.147,
+      "step": 1146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977034411345476e-05,
+      "loss": 1.3184,
+      "step": 1147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976936826941387e-05,
+      "loss": 1.3579,
+      "step": 1148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997683903589021e-05,
+      "loss": 1.2871,
+      "step": 1149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976741038193973e-05,
+      "loss": 1.1162,
+      "step": 1150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976642833854706e-05,
+      "loss": 1.3042,
+      "step": 1151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976544422874444e-05,
+      "loss": 1.2451,
+      "step": 1152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976445805255223e-05,
+      "loss": 1.3394,
+      "step": 1153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976346980999085e-05,
+      "loss": 1.2515,
+      "step": 1154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997624795010808e-05,
+      "loss": 1.2944,
+      "step": 1155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997614871258426e-05,
+      "loss": 1.2573,
+      "step": 1156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976049268429675e-05,
+      "loss": 1.2793,
+      "step": 1157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975949617646386e-05,
+      "loss": 1.272,
+      "step": 1158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975849760236458e-05,
+      "loss": 1.3223,
+      "step": 1159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975749696201966e-05,
+      "loss": 1.249,
+      "step": 1160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975649425544968e-05,
+      "loss": 1.3018,
+      "step": 1161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975548948267557e-05,
+      "loss": 1.2417,
+      "step": 1162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975448264371803e-05,
+      "loss": 1.0874,
+      "step": 1163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975347373859798e-05,
+      "loss": 1.2246,
+      "step": 1164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975246276733627e-05,
+      "loss": 1.3262,
+      "step": 1165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975144972995384e-05,
+      "loss": 1.3052,
+      "step": 1166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997504346264717e-05,
+      "loss": 1.2178,
+      "step": 1167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997494174569109e-05,
+      "loss": 1.25,
+      "step": 1168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997483982212924e-05,
+      "loss": 1.2354,
+      "step": 1169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997473769196374e-05,
+      "loss": 1.2246,
+      "step": 1170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974635355196706e-05,
+      "loss": 1.3047,
+      "step": 1171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974532811830255e-05,
+      "loss": 1.3008,
+      "step": 1172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974430061866512e-05,
+      "loss": 1.0303,
+      "step": 1173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974327105307605e-05,
+      "loss": 1.4502,
+      "step": 1174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974223942155664e-05,
+      "loss": 1.2583,
+      "step": 1175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974120572412828e-05,
+      "loss": 1.3779,
+      "step": 1176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974016996081238e-05,
+      "loss": 1.333,
+      "step": 1177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997391321316304e-05,
+      "loss": 1.2329,
+      "step": 1178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997380922366038e-05,
+      "loss": 1.3311,
+      "step": 1179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973705027575417e-05,
+      "loss": 1.3486,
+      "step": 1180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973600624910308e-05,
+      "loss": 1.3516,
+      "step": 1181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997349601566721e-05,
+      "loss": 1.144,
+      "step": 1182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99733911998483e-05,
+      "loss": 1.2832,
+      "step": 1183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973286177455736e-05,
+      "loss": 1.2695,
+      "step": 1184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973180948491705e-05,
+      "loss": 1.2896,
+      "step": 1185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973075512958377e-05,
+      "loss": 1.3076,
+      "step": 1186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972969870857946e-05,
+      "loss": 1.2349,
+      "step": 1187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997286402219259e-05,
+      "loss": 1.3955,
+      "step": 1188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997275796696451e-05,
+      "loss": 1.2998,
+      "step": 1189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972651705175896e-05,
+      "loss": 1.1763,
+      "step": 1190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972545236828954e-05,
+      "loss": 1.2412,
+      "step": 1191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972438561925885e-05,
+      "loss": 1.2598,
+      "step": 1192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99723316804689e-05,
+      "loss": 1.2007,
+      "step": 1193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997222459246021e-05,
+      "loss": 1.2344,
+      "step": 1194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972117297902044e-05,
+      "loss": 1.1631,
+      "step": 1195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997200979679661e-05,
+      "loss": 1.292,
+      "step": 1196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971902089146145e-05,
+      "loss": 1.2891,
+      "step": 1197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971794174952874e-05,
+      "loss": 1.2715,
+      "step": 1198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997168605421904e-05,
+      "loss": 1.2031,
+      "step": 1199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971577726946868e-05,
+      "loss": 1.2783,
+      "step": 1200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971469193138615e-05,
+      "loss": 1.1582,
+      "step": 1201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971360452796523e-05,
+      "loss": 1.4404,
+      "step": 1202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971251505922848e-05,
+      "loss": 1.1211,
+      "step": 1203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997114235251984e-05,
+      "loss": 1.2451,
+      "step": 1204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971032992589765e-05,
+      "loss": 1.4492,
+      "step": 1205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970923426134888e-05,
+      "loss": 1.2988,
+      "step": 1206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970813653157478e-05,
+      "loss": 1.3496,
+      "step": 1207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970703673659807e-05,
+      "loss": 1.2432,
+      "step": 1208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970593487644158e-05,
+      "loss": 1.2266,
+      "step": 1209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970483095112806e-05,
+      "loss": 1.2812,
+      "step": 1210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970372496068045e-05,
+      "loss": 1.2959,
+      "step": 1211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997026169051216e-05,
+      "loss": 1.2363,
+      "step": 1212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970150678447445e-05,
+      "loss": 1.2754,
+      "step": 1213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970039459876203e-05,
+      "loss": 1.3184,
+      "step": 1214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969928034800735e-05,
+      "loss": 1.1123,
+      "step": 1215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969816403223353e-05,
+      "loss": 1.2988,
+      "step": 1216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969704565146366e-05,
+      "loss": 1.3545,
+      "step": 1217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969592520572095e-05,
+      "loss": 1.2612,
+      "step": 1218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996948026950285e-05,
+      "loss": 1.228,
+      "step": 1219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996936781194097e-05,
+      "loss": 1.2607,
+      "step": 1220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996925514788877e-05,
+      "loss": 1.0547,
+      "step": 1221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969142277348595e-05,
+      "loss": 1.2881,
+      "step": 1222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996902920032278e-05,
+      "loss": 1.1753,
+      "step": 1223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968915916813665e-05,
+      "loss": 1.2012,
+      "step": 1224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968802426823595e-05,
+      "loss": 1.3916,
+      "step": 1225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968688730354922e-05,
+      "loss": 1.4854,
+      "step": 1226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968574827410005e-05,
+      "loss": 1.2334,
+      "step": 1227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99684607179912e-05,
+      "loss": 1.4326,
+      "step": 1228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968346402100868e-05,
+      "loss": 1.2114,
+      "step": 1229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968231879741376e-05,
+      "loss": 1.2905,
+      "step": 1230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968117150915104e-05,
+      "loss": 1.2065,
+      "step": 1231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968002215624422e-05,
+      "loss": 1.5186,
+      "step": 1232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967887073871712e-05,
+      "loss": 1.2461,
+      "step": 1233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967771725659363e-05,
+      "loss": 1.3262,
+      "step": 1234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967656170989757e-05,
+      "loss": 1.2822,
+      "step": 1235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996754040986529e-05,
+      "loss": 1.3179,
+      "step": 1236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996742444228836e-05,
+      "loss": 1.3438,
+      "step": 1237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967308268261368e-05,
+      "loss": 1.2471,
+      "step": 1238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996719188778672e-05,
+      "loss": 1.3379,
+      "step": 1239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967075300866834e-05,
+      "loss": 1.3252,
+      "step": 1240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966958507504116e-05,
+      "loss": 1.2344,
+      "step": 1241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966841507700987e-05,
+      "loss": 1.1587,
+      "step": 1242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966724301459872e-05,
+      "loss": 1.335,
+      "step": 1243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966606888783197e-05,
+      "loss": 1.373,
+      "step": 1244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966489269673396e-05,
+      "loss": 1.2812,
+      "step": 1245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966371444132903e-05,
+      "loss": 1.3384,
+      "step": 1246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996625341216416e-05,
+      "loss": 1.2559,
+      "step": 1247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996613517376961e-05,
+      "loss": 1.2812,
+      "step": 1248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966016728951705e-05,
+      "loss": 1.1948,
+      "step": 1249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965898077712894e-05,
+      "loss": 1.1436,
+      "step": 1250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996577922005564e-05,
+      "loss": 1.1118,
+      "step": 1251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.99656601559824e-05,
+      "loss": 1.2197,
+      "step": 1252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965540885495643e-05,
+      "loss": 1.2285,
+      "step": 1253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996542140859784e-05,
+      "loss": 1.3145,
+      "step": 1254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996530172529146e-05,
+      "loss": 1.209,
+      "step": 1255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996518183557899e-05,
+      "loss": 1.1567,
+      "step": 1256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965061739462903e-05,
+      "loss": 1.2661,
+      "step": 1257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964941436945698e-05,
+      "loss": 1.3057,
+      "step": 1258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964820928029857e-05,
+      "loss": 1.2529,
+      "step": 1259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964700212717883e-05,
+      "loss": 1.2588,
+      "step": 1260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964579291012275e-05,
+      "loss": 1.2427,
+      "step": 1261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964458162915536e-05,
+      "loss": 1.3115,
+      "step": 1262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964336828430176e-05,
+      "loss": 1.2656,
+      "step": 1263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964215287558706e-05,
+      "loss": 1.2871,
+      "step": 1264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964093540303643e-05,
+      "loss": 1.2163,
+      "step": 1265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963971586667513e-05,
+      "loss": 1.2178,
+      "step": 1266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963849426652838e-05,
+      "loss": 1.1807,
+      "step": 1267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996372706026215e-05,
+      "loss": 1.3984,
+      "step": 1268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963604487497982e-05,
+      "loss": 1.1904,
+      "step": 1269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963481708362877e-05,
+      "loss": 1.2725,
+      "step": 1270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963358722859372e-05,
+      "loss": 1.1431,
+      "step": 1271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963235530990018e-05,
+      "loss": 1.2896,
+      "step": 1272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963112132757366e-05,
+      "loss": 1.1748,
+      "step": 1273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962988528163975e-05,
+      "loss": 1.2153,
+      "step": 1274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962864717212395e-05,
+      "loss": 1.2871,
+      "step": 1275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962740699905202e-05,
+      "loss": 1.2227,
+      "step": 1276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996261647624496e-05,
+      "loss": 1.2769,
+      "step": 1277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962492046234242e-05,
+      "loss": 1.061,
+      "step": 1278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962367409875628e-05,
+      "loss": 1.5,
+      "step": 1279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996224256717169e-05,
+      "loss": 1.2427,
+      "step": 1280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996211751812503e-05,
+      "loss": 1.1567,
+      "step": 1281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996199226273822e-05,
+      "loss": 1.1538,
+      "step": 1282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961866801013867e-05,
+      "loss": 1.4082,
+      "step": 1283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961741132954565e-05,
+      "loss": 1.1621,
+      "step": 1284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996161525856292e-05,
+      "loss": 1.2939,
+      "step": 1285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961489177841535e-05,
+      "loss": 1.2373,
+      "step": 1286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961362890793022e-05,
+      "loss": 1.2539,
+      "step": 1287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961236397419998e-05,
+      "loss": 1.2432,
+      "step": 1288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961109697725087e-05,
+      "loss": 1.2915,
+      "step": 1289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960982791710903e-05,
+      "loss": 1.2617,
+      "step": 1290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960855679380086e-05,
+      "loss": 1.1143,
+      "step": 1291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996072836073526e-05,
+      "loss": 1.1445,
+      "step": 1292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960600835779068e-05,
+      "loss": 1.3271,
+      "step": 1293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960473104514148e-05,
+      "loss": 1.1968,
+      "step": 1294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960345166943146e-05,
+      "loss": 1.2324,
+      "step": 1295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960217023068714e-05,
+      "loss": 1.2554,
+      "step": 1296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960088672893503e-05,
+      "loss": 1.0723,
+      "step": 1297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959960116420175e-05,
+      "loss": 1.1611,
+      "step": 1298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995983135365139e-05,
+      "loss": 1.3286,
+      "step": 1299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959702384589814e-05,
+      "loss": 1.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959573209238125e-05,
+      "loss": 1.1758,
+      "step": 1301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995944382759899e-05,
+      "loss": 1.2275,
+      "step": 1302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959314239675093e-05,
+      "loss": 1.3037,
+      "step": 1303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959184445469118e-05,
+      "loss": 1.2119,
+      "step": 1304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959054444983752e-05,
+      "loss": 1.1787,
+      "step": 1305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995892423822169e-05,
+      "loss": 1.1924,
+      "step": 1306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958793825185627e-05,
+      "loss": 1.1489,
+      "step": 1307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958663205878268e-05,
+      "loss": 1.3301,
+      "step": 1308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958532380302306e-05,
+      "loss": 1.1665,
+      "step": 1309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995840134846047e-05,
+      "loss": 1.166,
+      "step": 1310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958270110355457e-05,
+      "loss": 1.3833,
+      "step": 1311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958138665989994e-05,
+      "loss": 1.3057,
+      "step": 1312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958007015366804e-05,
+      "loss": 1.1587,
+      "step": 1313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957875158488608e-05,
+      "loss": 1.1904,
+      "step": 1314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957743095358144e-05,
+      "loss": 1.2161,
+      "step": 1315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995761082597814e-05,
+      "loss": 1.1372,
+      "step": 1316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957478350351343e-05,
+      "loss": 1.1475,
+      "step": 1317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957345668480492e-05,
+      "loss": 1.1821,
+      "step": 1318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957212780368336e-05,
+      "loss": 1.2354,
+      "step": 1319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995707968601763e-05,
+      "loss": 1.3262,
+      "step": 1320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956946385431126e-05,
+      "loss": 1.2354,
+      "step": 1321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995681287861159e-05,
+      "loss": 1.2607,
+      "step": 1322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956679165561782e-05,
+      "loss": 1.252,
+      "step": 1323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956545246284476e-05,
+      "loss": 1.3623,
+      "step": 1324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956411120782447e-05,
+      "loss": 1.2661,
+      "step": 1325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995627678905847e-05,
+      "loss": 1.2422,
+      "step": 1326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956142251115323e-05,
+      "loss": 1.2285,
+      "step": 1327
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99560075069558e-05,
+      "loss": 1.1362,
+      "step": 1328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955872556582693e-05,
+      "loss": 1.3076,
+      "step": 1329
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955737399998787e-05,
+      "loss": 1.2695,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955602037206893e-05,
+      "loss": 1.3223,
+      "step": 1331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955466468209804e-05,
+      "loss": 1.3682,
+      "step": 1332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995533069301034e-05,
+      "loss": 1.1929,
+      "step": 1333
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955194711611304e-05,
+      "loss": 1.3994,
+      "step": 1334
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955058524015513e-05,
+      "loss": 1.1768,
+      "step": 1335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954922130225794e-05,
+      "loss": 1.0947,
+      "step": 1336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954785530244966e-05,
+      "loss": 1.2168,
+      "step": 1337
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954648724075857e-05,
+      "loss": 1.1978,
+      "step": 1338
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954511711721308e-05,
+      "loss": 1.1108,
+      "step": 1339
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954374493184153e-05,
+      "loss": 1.1372,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954237068467232e-05,
+      "loss": 1.1436,
+      "step": 1341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954099437573395e-05,
+      "loss": 1.2847,
+      "step": 1342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953961600505488e-05,
+      "loss": 1.2461,
+      "step": 1343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953823557266372e-05,
+      "loss": 1.249,
+      "step": 1344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953685307858906e-05,
+      "loss": 1.2339,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953546852285948e-05,
+      "loss": 1.2412,
+      "step": 1346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953408190550364e-05,
+      "loss": 1.3013,
+      "step": 1347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953269322655035e-05,
+      "loss": 1.3105,
+      "step": 1348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953130248602834e-05,
+      "loss": 1.188,
+      "step": 1349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995299096839664e-05,
+      "loss": 1.1516,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952851482039338e-05,
+      "loss": 1.2324,
+      "step": 1351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952711789533813e-05,
+      "loss": 1.3281,
+      "step": 1352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995257189088297e-05,
+      "loss": 1.3071,
+      "step": 1353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952431786089693e-05,
+      "loss": 1.3008,
+      "step": 1354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952291475156892e-05,
+      "loss": 1.2056,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952150958087473e-05,
+      "loss": 1.2939,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952010234884342e-05,
+      "loss": 1.3848,
+      "step": 1357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995186930555042e-05,
+      "loss": 1.3545,
+      "step": 1358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995172817008862e-05,
+      "loss": 1.3545,
+      "step": 1359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995158682850187e-05,
+      "loss": 1.2861,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951445280793096e-05,
+      "loss": 1.2188,
+      "step": 1361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951303526965226e-05,
+      "loss": 1.2539,
+      "step": 1362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951161567021202e-05,
+      "loss": 1.1865,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951019400963962e-05,
+      "loss": 1.3086,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950877028796446e-05,
+      "loss": 1.2974,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995073445052161e-05,
+      "loss": 1.3408,
+      "step": 1366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950591666142408e-05,
+      "loss": 1.1909,
+      "step": 1367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995044867566179e-05,
+      "loss": 1.2881,
+      "step": 1368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995030547908272e-05,
+      "loss": 1.2827,
+      "step": 1369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995016207640817e-05,
+      "loss": 1.2417,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99500184676411e-05,
+      "loss": 1.189,
+      "step": 1371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9949874652784493e-05,
+      "loss": 1.2739,
+      "step": 1372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949730631841322e-05,
+      "loss": 1.1831,
+      "step": 1373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949586404814577e-05,
+      "loss": 1.3887,
+      "step": 1374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994944197170724e-05,
+      "loss": 1.2227,
+      "step": 1375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949297332522303e-05,
+      "loss": 1.2402,
+      "step": 1376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949152487262765e-05,
+      "loss": 1.2148,
+      "step": 1377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994900743593162e-05,
+      "loss": 1.3076,
+      "step": 1378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948862178531876e-05,
+      "loss": 1.2314,
+      "step": 1379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948716715066544e-05,
+      "loss": 1.2852,
+      "step": 1380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948571045538634e-05,
+      "loss": 1.2119,
+      "step": 1381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948425169951166e-05,
+      "loss": 1.334,
+      "step": 1382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948279088307155e-05,
+      "loss": 1.2412,
+      "step": 1383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948132800609633e-05,
+      "loss": 1.2993,
+      "step": 1384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947986306861627e-05,
+      "loss": 1.1904,
+      "step": 1385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994783960706617e-05,
+      "loss": 1.1855,
+      "step": 1386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994769270122631e-05,
+      "loss": 1.3398,
+      "step": 1387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947545589345075e-05,
+      "loss": 1.0889,
+      "step": 1388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994739827142552e-05,
+      "loss": 1.2476,
+      "step": 1389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.99472507474707e-05,
+      "loss": 1.2134,
+      "step": 1390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947103017483664e-05,
+      "loss": 1.2793,
+      "step": 1391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946955081467477e-05,
+      "loss": 1.376,
+      "step": 1392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946806939425197e-05,
+      "loss": 1.3418,
+      "step": 1393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946658591359896e-05,
+      "loss": 1.1685,
+      "step": 1394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946510037274647e-05,
+      "loss": 1.3027,
+      "step": 1395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946361277172527e-05,
+      "loss": 1.4053,
+      "step": 1396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994621231105662e-05,
+      "loss": 1.2925,
+      "step": 1397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946063138930003e-05,
+      "loss": 1.2568,
+      "step": 1398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994591376079577e-05,
+      "loss": 1.1616,
+      "step": 1399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994576417665702e-05,
+      "loss": 1.2207,
+      "step": 1400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945614386516844e-05,
+      "loss": 1.103,
+      "step": 1401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945464390378348e-05,
+      "loss": 1.2607,
+      "step": 1402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994531418824464e-05,
+      "loss": 1.2861,
+      "step": 1403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945163780118827e-05,
+      "loss": 1.2036,
+      "step": 1404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994501316600403e-05,
+      "loss": 1.0029,
+      "step": 1405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944862345903365e-05,
+      "loss": 1.1021,
+      "step": 1406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994471131981995e-05,
+      "loss": 1.1304,
+      "step": 1407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944560087756926e-05,
+      "loss": 1.0654,
+      "step": 1408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944408649717415e-05,
+      "loss": 1.0254,
+      "step": 1409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994425700570456e-05,
+      "loss": 1.4014,
+      "step": 1410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.99441051557215e-05,
+      "loss": 1.1948,
+      "step": 1411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943953099771373e-05,
+      "loss": 1.2744,
+      "step": 1412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994380083785734e-05,
+      "loss": 1.2822,
+      "step": 1413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994364836998255e-05,
+      "loss": 1.1807,
+      "step": 1414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943495696150158e-05,
+      "loss": 1.2104,
+      "step": 1415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994334281636333e-05,
+      "loss": 1.2754,
+      "step": 1416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994318973062523e-05,
+      "loss": 0.9673,
+      "step": 1417
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9943036438939033e-05,
+      "loss": 1.269,
+      "step": 1418
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994288294130791e-05,
+      "loss": 1.2417,
+      "step": 1419
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994272923773504e-05,
+      "loss": 1.1206,
+      "step": 1420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942575328223606e-05,
+      "loss": 1.3145,
+      "step": 1421
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942421212776802e-05,
+      "loss": 1.29,
+      "step": 1422
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942266891397817e-05,
+      "loss": 1.2744,
+      "step": 1423
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942112364089844e-05,
+      "loss": 1.3418,
+      "step": 1424
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941957630856087e-05,
+      "loss": 1.2402,
+      "step": 1425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994180269169975e-05,
+      "loss": 1.2422,
+      "step": 1426
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941647546624043e-05,
+      "loss": 1.2134,
+      "step": 1427
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941492195632176e-05,
+      "loss": 1.251,
+      "step": 1428
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941336638727373e-05,
+      "loss": 1.2021,
+      "step": 1429
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941180875912847e-05,
+      "loss": 1.1768,
+      "step": 1430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941024907191836e-05,
+      "loss": 1.2368,
+      "step": 1431
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940868732567557e-05,
+      "loss": 1.2861,
+      "step": 1432
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940712352043257e-05,
+      "loss": 1.2188,
+      "step": 1433
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940555765622168e-05,
+      "loss": 1.1162,
+      "step": 1434
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940398973307536e-05,
+      "loss": 1.2129,
+      "step": 1435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940241975102606e-05,
+      "loss": 1.1543,
+      "step": 1436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940084771010634e-05,
+      "loss": 1.1963,
+      "step": 1437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993992736103487e-05,
+      "loss": 1.3037,
+      "step": 1438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993976974517858e-05,
+      "loss": 1.1284,
+      "step": 1439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939611923445028e-05,
+      "loss": 1.1323,
+      "step": 1440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939453895837477e-05,
+      "loss": 1.1025,
+      "step": 1441
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939295662359206e-05,
+      "loss": 1.3086,
+      "step": 1442
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939137223013494e-05,
+      "loss": 1.2607,
+      "step": 1443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938978577803616e-05,
+      "loss": 1.3447,
+      "step": 1444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938819726732864e-05,
+      "loss": 1.4072,
+      "step": 1445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938660669804527e-05,
+      "loss": 1.2344,
+      "step": 1446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938501407021894e-05,
+      "loss": 1.2129,
+      "step": 1447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993834193838827e-05,
+      "loss": 1.3516,
+      "step": 1448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938182263906958e-05,
+      "loss": 1.166,
+      "step": 1449
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993802238358126e-05,
+      "loss": 1.2129,
+      "step": 1450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993786229741449e-05,
+      "loss": 1.3359,
+      "step": 1451
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937702005409968e-05,
+      "loss": 1.1934,
+      "step": 1452
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993754150757101e-05,
+      "loss": 1.3652,
+      "step": 1453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937380803900937e-05,
+      "loss": 1.4502,
+      "step": 1454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937219894403087e-05,
+      "loss": 1.1343,
+      "step": 1455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937058779080784e-05,
+      "loss": 1.0952,
+      "step": 1456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993689745793737e-05,
+      "loss": 1.2012,
+      "step": 1457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936735930976183e-05,
+      "loss": 1.1465,
+      "step": 1458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993657419820057e-05,
+      "loss": 1.2075,
+      "step": 1459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993641225961388e-05,
+      "loss": 1.251,
+      "step": 1460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936250115219473e-05,
+      "loss": 1.25,
+      "step": 1461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936087765020698e-05,
+      "loss": 1.2627,
+      "step": 1462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935925209020926e-05,
+      "loss": 1.1377,
+      "step": 1463
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993576244722352e-05,
+      "loss": 1.3413,
+      "step": 1464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993559947963185e-05,
+      "loss": 1.2969,
+      "step": 1465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935436306249293e-05,
+      "loss": 1.1143,
+      "step": 1466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935272927079233e-05,
+      "loss": 1.0854,
+      "step": 1467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935109342125047e-05,
+      "loss": 1.2529,
+      "step": 1468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934945551390125e-05,
+      "loss": 1.2529,
+      "step": 1469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993478155487786e-05,
+      "loss": 1.0488,
+      "step": 1470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934617352591647e-05,
+      "loss": 1.1763,
+      "step": 1471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934452944534892e-05,
+      "loss": 1.1289,
+      "step": 1472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934288330711e-05,
+      "loss": 1.3438,
+      "step": 1473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934123511123376e-05,
+      "loss": 1.1841,
+      "step": 1474
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933958485775436e-05,
+      "loss": 1.2773,
+      "step": 1475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933793254670596e-05,
+      "loss": 1.2891,
+      "step": 1476
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933627817812285e-05,
+      "loss": 1.2603,
+      "step": 1477
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933462175203924e-05,
+      "loss": 1.2251,
+      "step": 1478
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993329632684894e-05,
+      "loss": 1.1714,
+      "step": 1479
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993313027275078e-05,
+      "loss": 1.2388,
+      "step": 1480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932964012912874e-05,
+      "loss": 1.3086,
+      "step": 1481
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932797547338665e-05,
+      "loss": 1.2271,
+      "step": 1482
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932630876031606e-05,
+      "loss": 1.2827,
+      "step": 1483
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993246399899515e-05,
+      "loss": 1.1284,
+      "step": 1484
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993229691623275e-05,
+      "loss": 1.2598,
+      "step": 1485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993212962774786e-05,
+      "loss": 1.1826,
+      "step": 1486
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931962133543957e-05,
+      "loss": 1.2104,
+      "step": 1487
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993179443362451e-05,
+      "loss": 1.3711,
+      "step": 1488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931626527992982e-05,
+      "loss": 1.2207,
+      "step": 1489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993145841665286e-05,
+      "loss": 1.2666,
+      "step": 1490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993129009960762e-05,
+      "loss": 1.3721,
+      "step": 1491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931121576860754e-05,
+      "loss": 1.3359,
+      "step": 1492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930952848415747e-05,
+      "loss": 1.0493,
+      "step": 1493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.99307839142761e-05,
+      "loss": 1.1665,
+      "step": 1494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930614774445304e-05,
+      "loss": 1.2319,
+      "step": 1495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930445428926867e-05,
+      "loss": 1.1895,
+      "step": 1496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.99302758777243e-05,
+      "loss": 1.2607,
+      "step": 1497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993010612084111e-05,
+      "loss": 1.3027,
+      "step": 1498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992993615828081e-05,
+      "loss": 1.2832,
+      "step": 1499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929765990046926e-05,
+      "loss": 1.1445,
+      "step": 1500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929595616142983e-05,
+      "loss": 1.1553,
+      "step": 1501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992942503657251e-05,
+      "loss": 1.2773,
+      "step": 1502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929254251339033e-05,
+      "loss": 1.2266,
+      "step": 1503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.99290832604461e-05,
+      "loss": 1.1895,
+      "step": 1504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928912063897243e-05,
+      "loss": 1.1855,
+      "step": 1505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992874066169601e-05,
+      "loss": 1.2354,
+      "step": 1506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992856905384596e-05,
+      "loss": 1.2544,
+      "step": 1507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9928397240350636e-05,
+      "loss": 1.3379,
+      "step": 1508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.99282252212136e-05,
+      "loss": 1.3218,
+      "step": 1509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992805299643842e-05,
+      "loss": 1.2222,
+      "step": 1510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927880566028657e-05,
+      "loss": 1.0928,
+      "step": 1511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927707929987886e-05,
+      "loss": 1.3057,
+      "step": 1512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992753508831968e-05,
+      "loss": 1.2949,
+      "step": 1513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927362041027624e-05,
+      "loss": 1.0439,
+      "step": 1514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927188788115296e-05,
+      "loss": 1.2266,
+      "step": 1515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927015329586285e-05,
+      "loss": 1.3003,
+      "step": 1516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992684166544419e-05,
+      "loss": 1.1992,
+      "step": 1517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926667795692603e-05,
+      "loss": 1.3564,
+      "step": 1518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992649372033513e-05,
+      "loss": 1.0996,
+      "step": 1519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926319439375364e-05,
+      "loss": 1.1948,
+      "step": 1520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992614495281693e-05,
+      "loss": 1.1997,
+      "step": 1521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925970260663436e-05,
+      "loss": 1.1553,
+      "step": 1522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925795362918498e-05,
+      "loss": 1.0767,
+      "step": 1523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992562025958574e-05,
+      "loss": 1.1929,
+      "step": 1524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992544495066879e-05,
+      "loss": 1.1929,
+      "step": 1525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992526943617128e-05,
+      "loss": 1.2544,
+      "step": 1526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992509371609684e-05,
+      "loss": 1.2012,
+      "step": 1527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992491779044912e-05,
+      "loss": 1.3662,
+      "step": 1528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924741659231755e-05,
+      "loss": 1.3452,
+      "step": 1529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924565322448394e-05,
+      "loss": 1.4082,
+      "step": 1530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992438878010269e-05,
+      "loss": 1.0972,
+      "step": 1531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.99242120321983e-05,
+      "loss": 1.3066,
+      "step": 1532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992403507873889e-05,
+      "loss": 1.2754,
+      "step": 1533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923857919728117e-05,
+      "loss": 1.2192,
+      "step": 1534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992368055516965e-05,
+      "loss": 1.3633,
+      "step": 1535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992350298506717e-05,
+      "loss": 1.2285,
+      "step": 1536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923325209424354e-05,
+      "loss": 1.2412,
+      "step": 1537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992314722824488e-05,
+      "loss": 1.2544,
+      "step": 1538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922969041532435e-05,
+      "loss": 1.3394,
+      "step": 1539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992279064929071e-05,
+      "loss": 1.0947,
+      "step": 1540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922612051523402e-05,
+      "loss": 1.231,
+      "step": 1541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992243324823421e-05,
+      "loss": 1.1494,
+      "step": 1542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922254239426834e-05,
+      "loss": 1.353,
+      "step": 1543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922075025104984e-05,
+      "loss": 1.2153,
+      "step": 1544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921895605272376e-05,
+      "loss": 1.2188,
+      "step": 1545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992171597993272e-05,
+      "loss": 1.1899,
+      "step": 1546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921536149089736e-05,
+      "loss": 1.2246,
+      "step": 1547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921356112747155e-05,
+      "loss": 1.1792,
+      "step": 1548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.99211758709087e-05,
+      "loss": 1.1392,
+      "step": 1549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920995423578108e-05,
+      "loss": 1.1104,
+      "step": 1550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920814770759117e-05,
+      "loss": 1.1357,
+      "step": 1551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920633912455462e-05,
+      "loss": 1.1626,
+      "step": 1552
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920452848670898e-05,
+      "loss": 1.2139,
+      "step": 1553
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.992027157940917e-05,
+      "loss": 1.2095,
+      "step": 1554
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920090104674036e-05,
+      "loss": 1.229,
+      "step": 1555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991990842446925e-05,
+      "loss": 1.145,
+      "step": 1556
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991972653879858e-05,
+      "loss": 1.189,
+      "step": 1557
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919544447665788e-05,
+      "loss": 1.3726,
+      "step": 1558
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991936215107465e-05,
+      "loss": 1.3135,
+      "step": 1559
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991917964902894e-05,
+      "loss": 1.1426,
+      "step": 1560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991899694153244e-05,
+      "loss": 1.0073,
+      "step": 1561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918814028588932e-05,
+      "loss": 1.2549,
+      "step": 1562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918630910202203e-05,
+      "loss": 1.2275,
+      "step": 1563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991844758637605e-05,
+      "loss": 1.2832,
+      "step": 1564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991826405711427e-05,
+      "loss": 1.2764,
+      "step": 1565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918080322420662e-05,
+      "loss": 1.2651,
+      "step": 1566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991789638229903e-05,
+      "loss": 1.1782,
+      "step": 1567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991771223675319e-05,
+      "loss": 1.2144,
+      "step": 1568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991752788578695e-05,
+      "loss": 1.1953,
+      "step": 1569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917343329404128e-05,
+      "loss": 1.1821,
+      "step": 1570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991715856760855e-05,
+      "loss": 1.2271,
+      "step": 1571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916973600404043e-05,
+      "loss": 1.1445,
+      "step": 1572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916788427794437e-05,
+      "loss": 1.3066,
+      "step": 1573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991660304978357e-05,
+      "loss": 1.1128,
+      "step": 1574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916417466375275e-05,
+      "loss": 1.29,
+      "step": 1575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916231677573404e-05,
+      "loss": 1.1494,
+      "step": 1576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916045683381798e-05,
+      "loss": 1.2046,
+      "step": 1577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915859483804313e-05,
+      "loss": 1.3018,
+      "step": 1578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915673078844808e-05,
+      "loss": 1.3389,
+      "step": 1579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915486468507138e-05,
+      "loss": 1.2949,
+      "step": 1580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915299652795173e-05,
+      "loss": 1.2432,
+      "step": 1581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915112631712782e-05,
+      "loss": 1.2188,
+      "step": 1582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914925405263835e-05,
+      "loss": 1.3379,
+      "step": 1583
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914737973452214e-05,
+      "loss": 1.2319,
+      "step": 1584
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914550336281798e-05,
+      "loss": 1.2568,
+      "step": 1585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914362493756477e-05,
+      "loss": 1.2275,
+      "step": 1586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914174445880138e-05,
+      "loss": 1.2178,
+      "step": 1587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991398619265668e-05,
+      "loss": 1.2217,
+      "step": 1588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991379773409e-05,
+      "loss": 1.3052,
+      "step": 1589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913609070183998e-05,
+      "loss": 1.2227,
+      "step": 1590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913420200942584e-05,
+      "loss": 1.0669,
+      "step": 1591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913231126369675e-05,
+      "loss": 1.1548,
+      "step": 1592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991304184646918e-05,
+      "loss": 1.3643,
+      "step": 1593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912852361245023e-05,
+      "loss": 1.2007,
+      "step": 1594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912662670701128e-05,
+      "loss": 1.1235,
+      "step": 1595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912472774841426e-05,
+      "loss": 1.2197,
+      "step": 1596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912282673669846e-05,
+      "loss": 1.23,
+      "step": 1597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991209236719033e-05,
+      "loss": 1.2637,
+      "step": 1598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911901855406818e-05,
+      "loss": 1.2539,
+      "step": 1599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911711138323256e-05,
+      "loss": 1.2148,
+      "step": 1600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911520215943588e-05,
+      "loss": 1.1387,
+      "step": 1601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991132908827178e-05,
+      "loss": 1.3174,
+      "step": 1602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911137755311786e-05,
+      "loss": 1.2207,
+      "step": 1603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910946217067563e-05,
+      "loss": 1.3545,
+      "step": 1604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910754473543087e-05,
+      "loss": 1.29,
+      "step": 1605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910562524742327e-05,
+      "loss": 1.2646,
+      "step": 1606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991037037066925e-05,
+      "loss": 1.1025,
+      "step": 1607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991017801132785e-05,
+      "loss": 1.2036,
+      "step": 1608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909985446722103e-05,
+      "loss": 1.3848,
+      "step": 1609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909792676856e-05,
+      "loss": 1.3066,
+      "step": 1610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990959970173353e-05,
+      "loss": 1.2559,
+      "step": 1611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909406521358695e-05,
+      "loss": 1.2852,
+      "step": 1612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909213135735496e-05,
+      "loss": 1.207,
+      "step": 1613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909019544867934e-05,
+      "loss": 1.0957,
+      "step": 1614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908825748760022e-05,
+      "loss": 1.165,
+      "step": 1615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908631747415773e-05,
+      "loss": 1.2427,
+      "step": 1616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908437540839206e-05,
+      "loss": 1.1011,
+      "step": 1617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908243129034344e-05,
+      "loss": 1.1616,
+      "step": 1618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990804851200521e-05,
+      "loss": 1.2852,
+      "step": 1619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907853689755844e-05,
+      "loss": 1.1855,
+      "step": 1620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990765866229027e-05,
+      "loss": 1.2085,
+      "step": 1621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907463429612533e-05,
+      "loss": 1.2627,
+      "step": 1622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907267991726678e-05,
+      "loss": 1.1006,
+      "step": 1623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990707234863675e-05,
+      "loss": 1.2822,
+      "step": 1624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906876500346802e-05,
+      "loss": 1.1138,
+      "step": 1625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990668044686089e-05,
+      "loss": 1.2715,
+      "step": 1626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990648418818308e-05,
+      "loss": 1.2163,
+      "step": 1627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990628772431743e-05,
+      "loss": 1.2393,
+      "step": 1628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906091055268013e-05,
+      "loss": 1.2104,
+      "step": 1629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905894181038902e-05,
+      "loss": 1.269,
+      "step": 1630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905697101634172e-05,
+      "loss": 1.2656,
+      "step": 1631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905499817057913e-05,
+      "loss": 1.1519,
+      "step": 1632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.99053023273142e-05,
+      "loss": 1.2568,
+      "step": 1633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990510463240713e-05,
+      "loss": 1.1182,
+      "step": 1634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.99049067323408e-05,
+      "loss": 1.2075,
+      "step": 1635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904708627119302e-05,
+      "loss": 1.271,
+      "step": 1636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904510316746747e-05,
+      "loss": 1.1914,
+      "step": 1637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904311801227237e-05,
+      "loss": 1.249,
+      "step": 1638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904113080564885e-05,
+      "loss": 1.1758,
+      "step": 1639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903914154763808e-05,
+      "loss": 1.1865,
+      "step": 1640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903715023828126e-05,
+      "loss": 1.1323,
+      "step": 1641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903515687761962e-05,
+      "loss": 1.1904,
+      "step": 1642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9903316146569446e-05,
+      "loss": 1.1948,
+      "step": 1643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9903116400254712e-05,
+      "loss": 1.2402,
+      "step": 1644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902916448821894e-05,
+      "loss": 1.3354,
+      "step": 1645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902716292275138e-05,
+      "loss": 1.23,
+      "step": 1646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902515930618588e-05,
+      "loss": 1.2441,
+      "step": 1647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902315363856388e-05,
+      "loss": 1.2627,
+      "step": 1648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.99021145919927e-05,
+      "loss": 1.228,
+      "step": 1649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901913615031682e-05,
+      "loss": 1.229,
+      "step": 1650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901712432977492e-05,
+      "loss": 1.2632,
+      "step": 1651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901511045834304e-05,
+      "loss": 1.2812,
+      "step": 1652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901309453606282e-05,
+      "loss": 1.0679,
+      "step": 1653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901107656297608e-05,
+      "loss": 1.1284,
+      "step": 1654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900905653912456e-05,
+      "loss": 1.2139,
+      "step": 1655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990070344645501e-05,
+      "loss": 1.1963,
+      "step": 1656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990050103392946e-05,
+      "loss": 1.1812,
+      "step": 1657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990029841634e-05,
+      "loss": 1.1816,
+      "step": 1658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990009559369083e-05,
+      "loss": 1.1465,
+      "step": 1659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899892565986142e-05,
+      "loss": 1.3052,
+      "step": 1660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899689333230146e-05,
+      "loss": 1.1021,
+      "step": 1661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989948589542705e-05,
+      "loss": 1.0771,
+      "step": 1662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899282252581073e-05,
+      "loss": 1.2979,
+      "step": 1663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899078404696425e-05,
+      "loss": 1.1655,
+      "step": 1664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898874351777336e-05,
+      "loss": 1.106,
+      "step": 1665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898670093828026e-05,
+      "loss": 1.1221,
+      "step": 1666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898465630852724e-05,
+      "loss": 1.3042,
+      "step": 1667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898260962855676e-05,
+      "loss": 1.2119,
+      "step": 1668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989805608984111e-05,
+      "loss": 1.1279,
+      "step": 1669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897851011813274e-05,
+      "loss": 1.2637,
+      "step": 1670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897645728776418e-05,
+      "loss": 1.0625,
+      "step": 1671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989744024073479e-05,
+      "loss": 1.1211,
+      "step": 1672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989723454769265e-05,
+      "loss": 1.1338,
+      "step": 1673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989702864965425e-05,
+      "loss": 1.2949,
+      "step": 1674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896822546623862e-05,
+      "loss": 1.2061,
+      "step": 1675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896616238605758e-05,
+      "loss": 1.1587,
+      "step": 1676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.98964097256042e-05,
+      "loss": 1.2881,
+      "step": 1677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896203007623475e-05,
+      "loss": 1.2339,
+      "step": 1678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895996084667865e-05,
+      "loss": 1.1865,
+      "step": 1679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895788956741648e-05,
+      "loss": 1.0796,
+      "step": 1680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989558162384912e-05,
+      "loss": 1.1616,
+      "step": 1681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895374085994574e-05,
+      "loss": 1.2305,
+      "step": 1682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989516634318231e-05,
+      "loss": 1.1812,
+      "step": 1683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894958395416627e-05,
+      "loss": 1.2495,
+      "step": 1684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989475024270184e-05,
+      "loss": 1.1846,
+      "step": 1685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989454188504225e-05,
+      "loss": 1.1401,
+      "step": 1686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894333322442177e-05,
+      "loss": 1.1973,
+      "step": 1687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9894124554905946e-05,
+      "loss": 1.2192,
+      "step": 1688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989391558243787e-05,
+      "loss": 1.3047,
+      "step": 1689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893706405042287e-05,
+      "loss": 1.2627,
+      "step": 1690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893497022723526e-05,
+      "loss": 1.2437,
+      "step": 1691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893287435485926e-05,
+      "loss": 1.1895,
+      "step": 1692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893077643333827e-05,
+      "loss": 1.1836,
+      "step": 1693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892867646271573e-05,
+      "loss": 1.5225,
+      "step": 1694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892657444303512e-05,
+      "loss": 1.2427,
+      "step": 1695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892447037434e-05,
+      "loss": 1.23,
+      "step": 1696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892236425667396e-05,
+      "loss": 1.2246,
+      "step": 1697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892025609008063e-05,
+      "loss": 1.0562,
+      "step": 1698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891814587460364e-05,
+      "loss": 1.1514,
+      "step": 1699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891603361028674e-05,
+      "loss": 1.0967,
+      "step": 1700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891391929717362e-05,
+      "loss": 1.186,
+      "step": 1701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891180293530813e-05,
+      "loss": 1.2251,
+      "step": 1702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989096845247341e-05,
+      "loss": 1.2441,
+      "step": 1703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890756406549535e-05,
+      "loss": 1.3428,
+      "step": 1704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890544155763588e-05,
+      "loss": 1.189,
+      "step": 1705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989033170011996e-05,
+      "loss": 1.3672,
+      "step": 1706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890119039623053e-05,
+      "loss": 1.1025,
+      "step": 1707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988990617427727e-05,
+      "loss": 1.2588,
+      "step": 1708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889693104087028e-05,
+      "loss": 1.1899,
+      "step": 1709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988947982905673e-05,
+      "loss": 1.3271,
+      "step": 1710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889266349190796e-05,
+      "loss": 1.1768,
+      "step": 1711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988905266449365e-05,
+      "loss": 1.1377,
+      "step": 1712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888838774969718e-05,
+      "loss": 1.0986,
+      "step": 1713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888624680623425e-05,
+      "loss": 1.1841,
+      "step": 1714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888410381459212e-05,
+      "loss": 1.2021,
+      "step": 1715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888195877481515e-05,
+      "loss": 1.1968,
+      "step": 1716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887981168694778e-05,
+      "loss": 1.2144,
+      "step": 1717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887766255103447e-05,
+      "loss": 1.249,
+      "step": 1718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887551136711973e-05,
+      "loss": 1.2017,
+      "step": 1719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887335813524813e-05,
+      "loss": 1.2036,
+      "step": 1720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887120285546427e-05,
+      "loss": 1.2461,
+      "step": 1721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886904552781276e-05,
+      "loss": 1.1899,
+      "step": 1722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988668861523383e-05,
+      "loss": 1.3213,
+      "step": 1723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886472472908568e-05,
+      "loss": 1.1909,
+      "step": 1724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988625612580996e-05,
+      "loss": 1.2949,
+      "step": 1725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988603957394248e-05,
+      "loss": 1.0908,
+      "step": 1726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988582281731063e-05,
+      "loss": 1.1465,
+      "step": 1727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885605855918887e-05,
+      "loss": 1.2988,
+      "step": 1728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885388689771748e-05,
+      "loss": 1.2529,
+      "step": 1729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885171318873716e-05,
+      "loss": 1.248,
+      "step": 1730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884953743229288e-05,
+      "loss": 1.2871,
+      "step": 1731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988473596284297e-05,
+      "loss": 1.2754,
+      "step": 1732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884517977719274e-05,
+      "loss": 1.1982,
+      "step": 1733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884299787862718e-05,
+      "loss": 1.1328,
+      "step": 1734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884081393277817e-05,
+      "loss": 1.3442,
+      "step": 1735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98838627939691e-05,
+      "loss": 1.0288,
+      "step": 1736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883643989941088e-05,
+      "loss": 1.0967,
+      "step": 1737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883424981198317e-05,
+      "loss": 1.2505,
+      "step": 1738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988320576774532e-05,
+      "loss": 1.229,
+      "step": 1739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988298634958664e-05,
+      "loss": 1.0786,
+      "step": 1740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988276672672682e-05,
+      "loss": 1.1196,
+      "step": 1741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882546899170414e-05,
+      "loss": 1.1562,
+      "step": 1742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988232686692197e-05,
+      "loss": 1.2891,
+      "step": 1743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882106629986047e-05,
+      "loss": 1.2344,
+      "step": 1744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881886188367207e-05,
+      "loss": 1.1812,
+      "step": 1745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881665542070014e-05,
+      "loss": 1.334,
+      "step": 1746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988144469109904e-05,
+      "loss": 1.2612,
+      "step": 1747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988122363545886e-05,
+      "loss": 1.0698,
+      "step": 1748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988100237515405e-05,
+      "loss": 1.2329,
+      "step": 1749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880780910189197e-05,
+      "loss": 1.1484,
+      "step": 1750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880559240568885e-05,
+      "loss": 1.2119,
+      "step": 1751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880337366297705e-05,
+      "loss": 1.3145,
+      "step": 1752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988011528738025e-05,
+      "loss": 1.1865,
+      "step": 1753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879893003821128e-05,
+      "loss": 1.1948,
+      "step": 1754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879670515624937e-05,
+      "loss": 1.0977,
+      "step": 1755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879447822796285e-05,
+      "loss": 1.2666,
+      "step": 1756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879224925339785e-05,
+      "loss": 1.1133,
+      "step": 1757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987900182326006e-05,
+      "loss": 1.2705,
+      "step": 1758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987877851656172e-05,
+      "loss": 1.1748,
+      "step": 1759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98785550052494e-05,
+      "loss": 1.2305,
+      "step": 1760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878331289327722e-05,
+      "loss": 1.1865,
+      "step": 1761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878107368801324e-05,
+      "loss": 1.0298,
+      "step": 1762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877883243674847e-05,
+      "loss": 1.146,
+      "step": 1763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877658913952924e-05,
+      "loss": 1.335,
+      "step": 1764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987743437964021e-05,
+      "loss": 1.2041,
+      "step": 1765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877209640741354e-05,
+      "loss": 1.2979,
+      "step": 1766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876984697261004e-05,
+      "loss": 1.147,
+      "step": 1767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876759549203827e-05,
+      "loss": 1.2217,
+      "step": 1768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876534196574485e-05,
+      "loss": 1.2871,
+      "step": 1769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876308639377643e-05,
+      "loss": 1.2441,
+      "step": 1770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876082877617974e-05,
+      "loss": 1.1479,
+      "step": 1771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875856911300156e-05,
+      "loss": 1.2212,
+      "step": 1772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987563074042887e-05,
+      "loss": 1.167,
+      "step": 1773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987540436500879e-05,
+      "loss": 1.2051,
+      "step": 1774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875177785044623e-05,
+      "loss": 1.1909,
+      "step": 1775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874951000541045e-05,
+      "loss": 1.2114,
+      "step": 1776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874724011502762e-05,
+      "loss": 1.21,
+      "step": 1777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874496817934475e-05,
+      "loss": 1.103,
+      "step": 1778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987426941984089e-05,
+      "loss": 1.2095,
+      "step": 1779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874041817226714e-05,
+      "loss": 1.1533,
+      "step": 1780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987381401009666e-05,
+      "loss": 1.2559,
+      "step": 1781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873585998455453e-05,
+      "loss": 1.3672,
+      "step": 1782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873357782307813e-05,
+      "loss": 1.2197,
+      "step": 1783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987312936165846e-05,
+      "loss": 1.1855,
+      "step": 1784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872900736512136e-05,
+      "loss": 1.1348,
+      "step": 1785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987267190687357e-05,
+      "loss": 1.2344,
+      "step": 1786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98724428727475e-05,
+      "loss": 1.2158,
+      "step": 1787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872213634138678e-05,
+      "loss": 1.228,
+      "step": 1788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871984191051844e-05,
+      "loss": 1.2759,
+      "step": 1789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871754543491754e-05,
+      "loss": 1.1172,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871524691463163e-05,
+      "loss": 1.1104,
+      "step": 1791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871294634970833e-05,
+      "loss": 1.1782,
+      "step": 1792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871064374019533e-05,
+      "loss": 1.2402,
+      "step": 1793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987083390861402e-05,
+      "loss": 1.1758,
+      "step": 1794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870603238759083e-05,
+      "loss": 1.2725,
+      "step": 1795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987037236445949e-05,
+      "loss": 1.1846,
+      "step": 1796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870141285720022e-05,
+      "loss": 1.373,
+      "step": 1797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869910002545475e-05,
+      "loss": 1.2515,
+      "step": 1798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869678514940626e-05,
+      "loss": 1.2114,
+      "step": 1799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986944682291028e-05,
+      "loss": 1.166,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986921492645923e-05,
+      "loss": 1.1895,
+      "step": 1801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868982825592287e-05,
+      "loss": 1.114,
+      "step": 1802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986875052031425e-05,
+      "loss": 1.2705,
+      "step": 1803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868518010629932e-05,
+      "loss": 1.2354,
+      "step": 1804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986828529654415e-05,
+      "loss": 1.2676,
+      "step": 1805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868052378061727e-05,
+      "loss": 1.2744,
+      "step": 1806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867819255187485e-05,
+      "loss": 1.0864,
+      "step": 1807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867585927926252e-05,
+      "loss": 1.1909,
+      "step": 1808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986735239628286e-05,
+      "loss": 1.0701,
+      "step": 1809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986711866026215e-05,
+      "loss": 1.1641,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986688471986896e-05,
+      "loss": 1.2227,
+      "step": 1811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986665057510813e-05,
+      "loss": 1.2319,
+      "step": 1812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986641622598452e-05,
+      "loss": 1.1855,
+      "step": 1813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866181672502982e-05,
+      "loss": 1.2422,
+      "step": 1814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986594691466837e-05,
+      "loss": 1.2241,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986571195248555e-05,
+      "loss": 1.1924,
+      "step": 1816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865476785959384e-05,
+      "loss": 1.2754,
+      "step": 1817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865241415094747e-05,
+      "loss": 1.1475,
+      "step": 1818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865005839896512e-05,
+      "loss": 1.2769,
+      "step": 1819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864770060369564e-05,
+      "loss": 1.29,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864534076518777e-05,
+      "loss": 1.2422,
+      "step": 1821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864297888349043e-05,
+      "loss": 1.2559,
+      "step": 1822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864061495865256e-05,
+      "loss": 1.2412,
+      "step": 1823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863824899072314e-05,
+      "loss": 1.2671,
+      "step": 1824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986358809797511e-05,
+      "loss": 1.1328,
+      "step": 1825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863351092578556e-05,
+      "loss": 1.2773,
+      "step": 1826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863113882887556e-05,
+      "loss": 1.2012,
+      "step": 1827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862876468907028e-05,
+      "loss": 1.2383,
+      "step": 1828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862638850641883e-05,
+      "loss": 1.2588,
+      "step": 1829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862401028097053e-05,
+      "loss": 1.1333,
+      "step": 1830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986216300127745e-05,
+      "loss": 1.2285,
+      "step": 1831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861924770188017e-05,
+      "loss": 1.1118,
+      "step": 1832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986168633483368e-05,
+      "loss": 1.1699,
+      "step": 1833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861447695219382e-05,
+      "loss": 1.1973,
+      "step": 1834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861208851350065e-05,
+      "loss": 1.0479,
+      "step": 1835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986096980323067e-05,
+      "loss": 1.2695,
+      "step": 1836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986073055086616e-05,
+      "loss": 1.0908,
+      "step": 1837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986049109426148e-05,
+      "loss": 1.1724,
+      "step": 1838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860251433421595e-05,
+      "loss": 1.0938,
+      "step": 1839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986001156835147e-05,
+      "loss": 1.2773,
+      "step": 1840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859771499056068e-05,
+      "loss": 1.1787,
+      "step": 1841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859531225540364e-05,
+      "loss": 1.2393,
+      "step": 1842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859290747809337e-05,
+      "loss": 1.1021,
+      "step": 1843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859050065867966e-05,
+      "loss": 1.1538,
+      "step": 1844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858809179721237e-05,
+      "loss": 1.2026,
+      "step": 1845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985856808937414e-05,
+      "loss": 1.1777,
+      "step": 1846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858326794831662e-05,
+      "loss": 1.0396,
+      "step": 1847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858085296098806e-05,
+      "loss": 1.1602,
+      "step": 1848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857843593180577e-05,
+      "loss": 1.3345,
+      "step": 1849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857601686081977e-05,
+      "loss": 1.1191,
+      "step": 1850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985735957480802e-05,
+      "loss": 1.2314,
+      "step": 1851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857117259363714e-05,
+      "loss": 1.3149,
+      "step": 1852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856874739754085e-05,
+      "loss": 1.0645,
+      "step": 1853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856632015984152e-05,
+      "loss": 1.2817,
+      "step": 1854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856389088058946e-05,
+      "loss": 1.0361,
+      "step": 1855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856145955983496e-05,
+      "loss": 1.2588,
+      "step": 1856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855902619762836e-05,
+      "loss": 1.0337,
+      "step": 1857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985565907940201e-05,
+      "loss": 1.1387,
+      "step": 1858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985541533490606e-05,
+      "loss": 1.189,
+      "step": 1859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855171386280038e-05,
+      "loss": 1.1689,
+      "step": 1860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854927233528992e-05,
+      "loss": 1.1279,
+      "step": 1861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985468287665798e-05,
+      "loss": 1.1304,
+      "step": 1862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854438315672067e-05,
+      "loss": 1.2275,
+      "step": 1863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854193550576316e-05,
+      "loss": 1.1025,
+      "step": 1864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853948581375796e-05,
+      "loss": 1.1904,
+      "step": 1865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853703408075582e-05,
+      "loss": 1.1582,
+      "step": 1866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985345803068075e-05,
+      "loss": 1.3354,
+      "step": 1867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9853212449196386e-05,
+      "loss": 1.1489,
+      "step": 1868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852966663627575e-05,
+      "loss": 1.2471,
+      "step": 1869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852720673979405e-05,
+      "loss": 1.1553,
+      "step": 1870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852474480256974e-05,
+      "loss": 1.2656,
+      "step": 1871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852228082465382e-05,
+      "loss": 1.1665,
+      "step": 1872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985198148060973e-05,
+      "loss": 1.1626,
+      "step": 1873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851734674695127e-05,
+      "loss": 1.1987,
+      "step": 1874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985148766472669e-05,
+      "loss": 1.2246,
+      "step": 1875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985124045070952e-05,
+      "loss": 1.1753,
+      "step": 1876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850993032648754e-05,
+      "loss": 1.2041,
+      "step": 1877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985074541054951e-05,
+      "loss": 1.2246,
+      "step": 1878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850497584416915e-05,
+      "loss": 1.2871,
+      "step": 1879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985024955425611e-05,
+      "loss": 1.2612,
+      "step": 1880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985000132007222e-05,
+      "loss": 1.3018,
+      "step": 1881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849752881870396e-05,
+      "loss": 1.2085,
+      "step": 1882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984950423965578e-05,
+      "loss": 1.2847,
+      "step": 1883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849255393433523e-05,
+      "loss": 1.1992,
+      "step": 1884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849006343208777e-05,
+      "loss": 1.1772,
+      "step": 1885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848757088986708e-05,
+      "loss": 1.2861,
+      "step": 1886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984850763077247e-05,
+      "loss": 1.1572,
+      "step": 1887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984825796857123e-05,
+      "loss": 1.2842,
+      "step": 1888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848008102388166e-05,
+      "loss": 1.3818,
+      "step": 1889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847758032228447e-05,
+      "loss": 1.3721,
+      "step": 1890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984750775809726e-05,
+      "loss": 1.063,
+      "step": 1891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984725727999978e-05,
+      "loss": 1.2744,
+      "step": 1892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98470065979412e-05,
+      "loss": 1.2324,
+      "step": 1893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846755711926708e-05,
+      "loss": 1.1934,
+      "step": 1894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846504621961505e-05,
+      "loss": 1.1772,
+      "step": 1895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846253328050797e-05,
+      "loss": 1.1514,
+      "step": 1896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846001830199776e-05,
+      "loss": 1.1162,
+      "step": 1897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845750128413656e-05,
+      "loss": 1.3188,
+      "step": 1898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845498222697653e-05,
+      "loss": 1.124,
+      "step": 1899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845246113056988e-05,
+      "loss": 1.0601,
+      "step": 1900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844993799496874e-05,
+      "loss": 1.1289,
+      "step": 1901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984474128202254e-05,
+      "loss": 1.2539,
+      "step": 1902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844488560639222e-05,
+      "loss": 1.2373,
+      "step": 1903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844235635352144e-05,
+      "loss": 1.042,
+      "step": 1904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843982506166556e-05,
+      "loss": 1.1318,
+      "step": 1905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984372917308769e-05,
+      "loss": 1.0176,
+      "step": 1906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843475636120804e-05,
+      "loss": 1.2217,
+      "step": 1907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843221895271142e-05,
+      "loss": 1.1704,
+      "step": 1908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842967950543964e-05,
+      "loss": 1.189,
+      "step": 1909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842713801944527e-05,
+      "loss": 1.1885,
+      "step": 1910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842459449478094e-05,
+      "loss": 1.0015,
+      "step": 1911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842204893149936e-05,
+      "loss": 1.1763,
+      "step": 1912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841950132965322e-05,
+      "loss": 1.3477,
+      "step": 1913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984169516892953e-05,
+      "loss": 1.2012,
+      "step": 1914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841440001047846e-05,
+      "loss": 1.2271,
+      "step": 1915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841184629325553e-05,
+      "loss": 1.0742,
+      "step": 1916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840929053767935e-05,
+      "loss": 1.2349,
+      "step": 1917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840673274380288e-05,
+      "loss": 1.0889,
+      "step": 1918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840417291167915e-05,
+      "loss": 1.2515,
+      "step": 1919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840161104136108e-05,
+      "loss": 1.2188,
+      "step": 1920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839904713290186e-05,
+      "loss": 1.1313,
+      "step": 1921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983964811863545e-05,
+      "loss": 1.2349,
+      "step": 1922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839391320177218e-05,
+      "loss": 1.2295,
+      "step": 1923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839134317920807e-05,
+      "loss": 1.1431,
+      "step": 1924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838877111871542e-05,
+      "loss": 1.2148,
+      "step": 1925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838619702034753e-05,
+      "loss": 1.1138,
+      "step": 1926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838362088415765e-05,
+      "loss": 1.1699,
+      "step": 1927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838104271019918e-05,
+      "loss": 1.2451,
+      "step": 1928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837846249852552e-05,
+      "loss": 1.147,
+      "step": 1929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983758802491901e-05,
+      "loss": 1.1953,
+      "step": 1930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983732959622464e-05,
+      "loss": 1.2578,
+      "step": 1931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837070963774797e-05,
+      "loss": 1.21,
+      "step": 1932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983681212757484e-05,
+      "loss": 1.1987,
+      "step": 1933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836553087630123e-05,
+      "loss": 1.1924,
+      "step": 1934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836293843946016e-05,
+      "loss": 1.1807,
+      "step": 1935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983603439652789e-05,
+      "loss": 1.125,
+      "step": 1936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835774745381112e-05,
+      "loss": 1.248,
+      "step": 1937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983551489051107e-05,
+      "loss": 1.2173,
+      "step": 1938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835254831923137e-05,
+      "loss": 1.2026,
+      "step": 1939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834994569622704e-05,
+      "loss": 1.1777,
+      "step": 1940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834734103615167e-05,
+      "loss": 1.2715,
+      "step": 1941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834473433905906e-05,
+      "loss": 1.1323,
+      "step": 1942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834212560500336e-05,
+      "loss": 1.2373,
+      "step": 1943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833951483403847e-05,
+      "loss": 1.1357,
+      "step": 1944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833690202621857e-05,
+      "loss": 1.2959,
+      "step": 1945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833428718159773e-05,
+      "loss": 1.2056,
+      "step": 1946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833167030023013e-05,
+      "loss": 1.3413,
+      "step": 1947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832905138216995e-05,
+      "loss": 1.0708,
+      "step": 1948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832643042747142e-05,
+      "loss": 1.1963,
+      "step": 1949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983238074361889e-05,
+      "loss": 1.271,
+      "step": 1950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832118240837663e-05,
+      "loss": 1.1816,
+      "step": 1951
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831855534408905e-05,
+      "loss": 1.2515,
+      "step": 1952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983159262433805e-05,
+      "loss": 1.0894,
+      "step": 1953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983132951063055e-05,
+      "loss": 1.3174,
+      "step": 1954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831066193291852e-05,
+      "loss": 1.2695,
+      "step": 1955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830802672327413e-05,
+      "loss": 1.1748,
+      "step": 1956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830538947742684e-05,
+      "loss": 1.1387,
+      "step": 1957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9830275019543134e-05,
+      "loss": 1.1455,
+      "step": 1958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9830010887734228e-05,
+      "loss": 1.1377,
+      "step": 1959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829746552321437e-05,
+      "loss": 1.3105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829482013310236e-05,
+      "loss": 1.0415,
+      "step": 1961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829217270706102e-05,
+      "loss": 1.3008,
+      "step": 1962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828952324514522e-05,
+      "loss": 1.1816,
+      "step": 1963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982868717474098e-05,
+      "loss": 1.0776,
+      "step": 1964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828421821390972e-05,
+      "loss": 1.1577,
+      "step": 1965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982815626446999e-05,
+      "loss": 1.0952,
+      "step": 1966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827890503983536e-05,
+      "loss": 1.1841,
+      "step": 1967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827624539937116e-05,
+      "loss": 1.127,
+      "step": 1968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827358372336237e-05,
+      "loss": 1.1099,
+      "step": 1969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827092001186412e-05,
+      "loss": 1.293,
+      "step": 1970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982682542649316e-05,
+      "loss": 1.1436,
+      "step": 1971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826558648262e-05,
+      "loss": 1.1572,
+      "step": 1972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982629166649846e-05,
+      "loss": 1.1782,
+      "step": 1973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826024481208073e-05,
+      "loss": 1.2539,
+      "step": 1974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982575709239636e-05,
+      "loss": 1.2109,
+      "step": 1975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825489500068874e-05,
+      "loss": 1.2148,
+      "step": 1976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825221704231153e-05,
+      "loss": 1.2114,
+      "step": 1977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982495370488874e-05,
+      "loss": 1.1631,
+      "step": 1978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824685502047188e-05,
+      "loss": 1.2295,
+      "step": 1979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824417095712052e-05,
+      "loss": 1.0913,
+      "step": 1980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982414848588889e-05,
+      "loss": 1.082,
+      "step": 1981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982387967258327e-05,
+      "loss": 1.2437,
+      "step": 1982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823610655800757e-05,
+      "loss": 1.1938,
+      "step": 1983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823341435546922e-05,
+      "loss": 1.0913,
+      "step": 1984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982307201182734e-05,
+      "loss": 1.125,
+      "step": 1985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98228023846476e-05,
+      "loss": 1.272,
+      "step": 1986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822532554013276e-05,
+      "loss": 1.2686,
+      "step": 1987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822262519929963e-05,
+      "loss": 1.1167,
+      "step": 1988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982199228240325e-05,
+      "loss": 1.0522,
+      "step": 1989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982172184143874e-05,
+      "loss": 1.1177,
+      "step": 1990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821451197042028e-05,
+      "loss": 1.0747,
+      "step": 1991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821180349218726e-05,
+      "loss": 1.2969,
+      "step": 1992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820909297974437e-05,
+      "loss": 1.1699,
+      "step": 1993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982063804331478e-05,
+      "loss": 1.251,
+      "step": 1994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820366585245378e-05,
+      "loss": 1.166,
+      "step": 1995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982009492377184e-05,
+      "loss": 1.0781,
+      "step": 1996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819823058899804e-05,
+      "loss": 1.2812,
+      "step": 1997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819550990634897e-05,
+      "loss": 1.1812,
+      "step": 1998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819278718982757e-05,
+      "loss": 1.0879,
+      "step": 1999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819006243949017e-05,
+      "loss": 1.2783,
+      "step": 2000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981873356553933e-05,
+      "loss": 1.1738,
+      "step": 2001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818460683759334e-05,
+      "loss": 1.188,
+      "step": 2002
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9818187598614688e-05,
+      "loss": 1.1724,
+      "step": 2003
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817914310111044e-05,
+      "loss": 1.1345,
+      "step": 2004
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817640818254065e-05,
+      "loss": 1.1851,
+      "step": 2005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981736712304942e-05,
+      "loss": 1.0068,
+      "step": 2006
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981709322450277e-05,
+      "loss": 1.2705,
+      "step": 2007
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981681912261979e-05,
+      "loss": 1.2881,
+      "step": 2008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9816544817406163e-05,
+      "loss": 1.2378,
+      "step": 2009
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981627030886756e-05,
+      "loss": 1.1396,
+      "step": 2010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815995597009676e-05,
+      "loss": 1.1914,
+      "step": 2011
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98157206818382e-05,
+      "loss": 1.0618,
+      "step": 2012
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815445563358818e-05,
+      "loss": 1.1523,
+      "step": 2013
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815170241577243e-05,
+      "loss": 1.1943,
+      "step": 2014
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981489471649916e-05,
+      "loss": 1.1689,
+      "step": 2015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981461898813029e-05,
+      "loss": 1.1592,
+      "step": 2016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814343056476337e-05,
+      "loss": 1.1938,
+      "step": 2017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814066921543018e-05,
+      "loss": 1.1948,
+      "step": 2018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813790583336055e-05,
+      "loss": 1.1675,
+      "step": 2019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981351404186117e-05,
+      "loss": 1.2153,
+      "step": 2020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813237297124087e-05,
+      "loss": 1.1226,
+      "step": 2021
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981296034913054e-05,
+      "loss": 1.0464,
+      "step": 2022
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812683197886267e-05,
+      "loss": 1.0664,
+      "step": 2023
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812405843397008e-05,
+      "loss": 1.1484,
+      "step": 2024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812128285668512e-05,
+      "loss": 1.1055,
+      "step": 2025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981185052470652e-05,
+      "loss": 1.2139,
+      "step": 2026
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811572560516784e-05,
+      "loss": 1.2886,
+      "step": 2027
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811294393105072e-05,
+      "loss": 1.1201,
+      "step": 2028
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981101602247714e-05,
+      "loss": 1.2207,
+      "step": 2029
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810737448638747e-05,
+      "loss": 1.2422,
+      "step": 2030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810458671595674e-05,
+      "loss": 1.2324,
+      "step": 2031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981017969135369e-05,
+      "loss": 1.0376,
+      "step": 2032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809900507918575e-05,
+      "loss": 1.1465,
+      "step": 2033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809621121296106e-05,
+      "loss": 1.1987,
+      "step": 2034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809341531492076e-05,
+      "loss": 1.2061,
+      "step": 2035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809061738512275e-05,
+      "loss": 1.3472,
+      "step": 2036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808781742362495e-05,
+      "loss": 1.2788,
+      "step": 2037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808501543048538e-05,
+      "loss": 1.1318,
+      "step": 2038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808221140576212e-05,
+      "loss": 1.1777,
+      "step": 2039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807940534951315e-05,
+      "loss": 1.1978,
+      "step": 2040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807659726179667e-05,
+      "loss": 1.1328,
+      "step": 2041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807378714267078e-05,
+      "loss": 1.2739,
+      "step": 2042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807097499219373e-05,
+      "loss": 1.1426,
+      "step": 2043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980681608104238e-05,
+      "loss": 1.2021,
+      "step": 2044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980653445974192e-05,
+      "loss": 1.1719,
+      "step": 2045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806252635323827e-05,
+      "loss": 1.1235,
+      "step": 2046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9805970607793944e-05,
+      "loss": 1.2256,
+      "step": 2047
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805688377158106e-05,
+      "loss": 1.1646,
+      "step": 2048
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805405943422165e-05,
+      "loss": 1.1772,
+      "step": 2049
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980512330659197e-05,
+      "loss": 1.1514,
+      "step": 2050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804840466673368e-05,
+      "loss": 1.062,
+      "step": 2051
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804557423672225e-05,
+      "loss": 1.1133,
+      "step": 2052
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804274177594398e-05,
+      "loss": 1.252,
+      "step": 2053
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803990728445758e-05,
+      "loss": 1.1484,
+      "step": 2054
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803707076232176e-05,
+      "loss": 1.1079,
+      "step": 2055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803423220959522e-05,
+      "loss": 1.2139,
+      "step": 2056
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803139162633684e-05,
+      "loss": 1.105,
+      "step": 2057
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802854901260536e-05,
+      "loss": 1.2158,
+      "step": 2058
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802570436845974e-05,
+      "loss": 1.1797,
+      "step": 2059
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802285769395883e-05,
+      "loss": 1.2207,
+      "step": 2060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802000898916166e-05,
+      "loss": 1.083,
+      "step": 2061
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801715825412714e-05,
+      "loss": 1.1753,
+      "step": 2062
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980143054889144e-05,
+      "loss": 1.1675,
+      "step": 2063
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980114506935825e-05,
+      "loss": 1.2095,
+      "step": 2064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980085938681906e-05,
+      "loss": 1.147,
+      "step": 2065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800573501279784e-05,
+      "loss": 1.1675,
+      "step": 2066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980028741274634e-05,
+      "loss": 1.2339,
+      "step": 2067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800001121224658e-05,
+      "loss": 1.248,
+      "step": 2068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799714626720668e-05,
+      "loss": 1.1196,
+      "step": 2069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799427929240306e-05,
+      "loss": 1.1362,
+      "step": 2070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799141028789505e-05,
+      "loss": 1.1245,
+      "step": 2071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798853925374207e-05,
+      "loss": 1.1963,
+      "step": 2072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798566619000362e-05,
+      "loss": 1.1787,
+      "step": 2073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798279109673923e-05,
+      "loss": 0.9883,
+      "step": 2074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979799139740084e-05,
+      "loss": 1.3291,
+      "step": 2075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797703482187075e-05,
+      "loss": 1.1621,
+      "step": 2076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979741536403859e-05,
+      "loss": 1.1382,
+      "step": 2077
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797127042961356e-05,
+      "loss": 1.1968,
+      "step": 2078
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796838518961343e-05,
+      "loss": 1.2061,
+      "step": 2079
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796549792044524e-05,
+      "loss": 1.1841,
+      "step": 2080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796260862216883e-05,
+      "loss": 1.3027,
+      "step": 2081
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795971729484404e-05,
+      "loss": 1.1763,
+      "step": 2082
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979568239385307e-05,
+      "loss": 1.3647,
+      "step": 2083
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979539285532888e-05,
+      "loss": 1.0811,
+      "step": 2084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795103113917835e-05,
+      "loss": 1.2666,
+      "step": 2085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794813169625925e-05,
+      "loss": 1.1353,
+      "step": 2086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794523022459168e-05,
+      "loss": 1.208,
+      "step": 2087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979423267242356e-05,
+      "loss": 1.1924,
+      "step": 2088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793942119525126e-05,
+      "loss": 1.1455,
+      "step": 2089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793651363769876e-05,
+      "loss": 1.0176,
+      "step": 2090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979336040516384e-05,
+      "loss": 1.1875,
+      "step": 2091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979306924371304e-05,
+      "loss": 1.2451,
+      "step": 2092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979277787942351e-05,
+      "loss": 1.0391,
+      "step": 2093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9792486312301277e-05,
+      "loss": 1.1875,
+      "step": 2094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979219454235239e-05,
+      "loss": 1.2041,
+      "step": 2095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791902569582885e-05,
+      "loss": 1.2173,
+      "step": 2096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979161039399881e-05,
+      "loss": 1.2354,
+      "step": 2097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791318015606224e-05,
+      "loss": 1.1611,
+      "step": 2098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791025434411177e-05,
+      "loss": 1.2974,
+      "step": 2099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790732650419725e-05,
+      "loss": 1.1333,
+      "step": 2100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790439663637944e-05,
+      "loss": 1.1582,
+      "step": 2101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979014647407189e-05,
+      "loss": 1.2256,
+      "step": 2102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789853081727644e-05,
+      "loss": 1.0764,
+      "step": 2103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978955948661128e-05,
+      "loss": 1.2715,
+      "step": 2104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789265688728877e-05,
+      "loss": 1.1631,
+      "step": 2105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788971688086524e-05,
+      "loss": 1.1646,
+      "step": 2106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788677484690307e-05,
+      "loss": 1.1909,
+      "step": 2107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788383078546326e-05,
+      "loss": 1.1436,
+      "step": 2108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788088469660672e-05,
+      "loss": 1.1343,
+      "step": 2109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978779365803945e-05,
+      "loss": 1.0693,
+      "step": 2110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787498643688764e-05,
+      "loss": 1.209,
+      "step": 2111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787203426614728e-05,
+      "loss": 1.2871,
+      "step": 2112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786908006823457e-05,
+      "loss": 1.1895,
+      "step": 2113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786612384321065e-05,
+      "loss": 1.1943,
+      "step": 2114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978631655911368e-05,
+      "loss": 1.0742,
+      "step": 2115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786020531207427e-05,
+      "loss": 1.1094,
+      "step": 2116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785724300608437e-05,
+      "loss": 1.2241,
+      "step": 2117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785427867322845e-05,
+      "loss": 1.3047,
+      "step": 2118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785131231356793e-05,
+      "loss": 1.1846,
+      "step": 2119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784834392716425e-05,
+      "loss": 1.0996,
+      "step": 2120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978453735140789e-05,
+      "loss": 1.0942,
+      "step": 2121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784240107437336e-05,
+      "loss": 1.1831,
+      "step": 2122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783942660810928e-05,
+      "loss": 1.2212,
+      "step": 2123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783645011534815e-05,
+      "loss": 1.1196,
+      "step": 2124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783347159615173e-05,
+      "loss": 1.3506,
+      "step": 2125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783049105058165e-05,
+      "loss": 1.1836,
+      "step": 2126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782750847869968e-05,
+      "loss": 1.1104,
+      "step": 2127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782452388056755e-05,
+      "loss": 1.0972,
+      "step": 2128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782153725624713e-05,
+      "loss": 1.0977,
+      "step": 2129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781854860580028e-05,
+      "loss": 1.2148,
+      "step": 2130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781555792928884e-05,
+      "loss": 1.1465,
+      "step": 2131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781256522677482e-05,
+      "loss": 1.0474,
+      "step": 2132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780957049832017e-05,
+      "loss": 1.1157,
+      "step": 2133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780657374398694e-05,
+      "loss": 1.1582,
+      "step": 2134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780357496383716e-05,
+      "loss": 1.1685,
+      "step": 2135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97800574157933e-05,
+      "loss": 1.145,
+      "step": 2136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9779757132633663e-05,
+      "loss": 1.2324,
+      "step": 2137
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9779456646911014e-05,
+      "loss": 1.2202,
+      "step": 2138
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9779155958631583e-05,
+      "loss": 1.3545,
+      "step": 2139
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97788550678016e-05,
+      "loss": 1.1836,
+      "step": 2140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778553974427294e-05,
+      "loss": 1.2012,
+      "step": 2141
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778252678514908e-05,
+      "loss": 1.3179,
+      "step": 2142
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977795118007067e-05,
+      "loss": 1.2285,
+      "step": 2143
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9777649479100835e-05,
+      "loss": 1.1279,
+      "step": 2144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977734757561165e-05,
+      "loss": 1.2236,
+      "step": 2145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9777045469609364e-05,
+      "loss": 1.1475,
+      "step": 2146
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977674316110024e-05,
+      "loss": 1.1201,
+      "step": 2147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776440650090535e-05,
+      "loss": 1.0486,
+      "step": 2148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776137936586522e-05,
+      "loss": 1.2109,
+      "step": 2149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977583502059446e-05,
+      "loss": 1.1592,
+      "step": 2150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977553190212063e-05,
+      "loss": 1.1509,
+      "step": 2151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9775228581171314e-05,
+      "loss": 1.2412,
+      "step": 2152
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774925057752784e-05,
+      "loss": 1.1724,
+      "step": 2153
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774621331871337e-05,
+      "loss": 1.104,
+      "step": 2154
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774317403533257e-05,
+      "loss": 1.1299,
+      "step": 2155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774013272744842e-05,
+      "loss": 1.2705,
+      "step": 2156
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977370893951239e-05,
+      "loss": 1.0757,
+      "step": 2157
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773404403842203e-05,
+      "loss": 1.2266,
+      "step": 2158
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773099665740594e-05,
+      "loss": 1.249,
+      "step": 2159
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772794725213873e-05,
+      "loss": 1.2607,
+      "step": 2160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977248958226835e-05,
+      "loss": 0.9897,
+      "step": 2161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772184236910358e-05,
+      "loss": 1.1714,
+      "step": 2162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771878689146205e-05,
+      "loss": 1.1355,
+      "step": 2163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977157293898224e-05,
+      "loss": 1.3799,
+      "step": 2164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771266986424774e-05,
+      "loss": 1.2764,
+      "step": 2165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770960831480157e-05,
+      "loss": 1.1079,
+      "step": 2166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770654474154727e-05,
+      "loss": 1.1382,
+      "step": 2167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977034791445483e-05,
+      "loss": 1.2095,
+      "step": 2168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770041152386815e-05,
+      "loss": 1.1948,
+      "step": 2169
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976973418795704e-05,
+      "loss": 1.1328,
+      "step": 2170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9769427021171858e-05,
+      "loss": 1.1812,
+      "step": 2171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9769119652037627e-05,
+      "loss": 1.2075,
+      "step": 2172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768812080560723e-05,
+      "loss": 1.3711,
+      "step": 2173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768504306747513e-05,
+      "loss": 1.2451,
+      "step": 2174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976819633060437e-05,
+      "loss": 1.0864,
+      "step": 2175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767888152137676e-05,
+      "loss": 1.3115,
+      "step": 2176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767579771353812e-05,
+      "loss": 1.1494,
+      "step": 2177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976727118825916e-05,
+      "loss": 1.1753,
+      "step": 2178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766962402860127e-05,
+      "loss": 1.1191,
+      "step": 2179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766653415163092e-05,
+      "loss": 1.2021,
+      "step": 2180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766344225174464e-05,
+      "loss": 1.1094,
+      "step": 2181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976603483290064e-05,
+      "loss": 1.1328,
+      "step": 2182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9765725238348035e-05,
+      "loss": 1.0596,
+      "step": 2183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9765415441523066e-05,
+      "loss": 1.1265,
+      "step": 2184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9765105442432137e-05,
+      "loss": 1.1265,
+      "step": 2185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764795241081678e-05,
+      "loss": 1.1636,
+      "step": 2186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764484837478107e-05,
+      "loss": 1.2432,
+      "step": 2187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764174231627863e-05,
+      "loss": 1.2002,
+      "step": 2188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976386342353737e-05,
+      "loss": 1.1211,
+      "step": 2189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976355241321307e-05,
+      "loss": 1.2056,
+      "step": 2190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763241200661403e-05,
+      "loss": 1.2495,
+      "step": 2191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976292978588882e-05,
+      "loss": 1.1426,
+      "step": 2192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762618168901767e-05,
+      "loss": 1.0474,
+      "step": 2193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762306349706695e-05,
+      "loss": 1.0854,
+      "step": 2194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761994328310067e-05,
+      "loss": 1.1753,
+      "step": 2195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761682104718346e-05,
+      "loss": 1.2046,
+      "step": 2196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761369678937997e-05,
+      "loss": 1.2402,
+      "step": 2197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761057050975495e-05,
+      "loss": 1.2026,
+      "step": 2198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760744220837308e-05,
+      "loss": 1.0723,
+      "step": 2199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760431188529923e-05,
+      "loss": 1.1133,
+      "step": 2200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976011795405982e-05,
+      "loss": 1.2539,
+      "step": 2201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759804517433485e-05,
+      "loss": 1.0391,
+      "step": 2202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759490878657412e-05,
+      "loss": 1.2939,
+      "step": 2203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97591770377381e-05,
+      "loss": 1.2295,
+      "step": 2204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758862994682046e-05,
+      "loss": 1.2524,
+      "step": 2205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758548749495757e-05,
+      "loss": 1.2178,
+      "step": 2206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975823430218574e-05,
+      "loss": 1.2734,
+      "step": 2207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757919652758505e-05,
+      "loss": 1.1743,
+      "step": 2208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757604801220577e-05,
+      "loss": 1.1035,
+      "step": 2209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757289747578473e-05,
+      "loss": 1.3467,
+      "step": 2210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756974491838713e-05,
+      "loss": 1.1768,
+      "step": 2211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975665903400784e-05,
+      "loss": 0.9512,
+      "step": 2212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756343374092376e-05,
+      "loss": 1.0972,
+      "step": 2213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756027512098863e-05,
+      "loss": 1.0815,
+      "step": 2214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755711448033844e-05,
+      "loss": 1.2471,
+      "step": 2215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755395181903863e-05,
+      "loss": 1.1484,
+      "step": 2216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755078713715474e-05,
+      "loss": 1.0244,
+      "step": 2217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754762043475232e-05,
+      "loss": 1.1865,
+      "step": 2218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754445171189694e-05,
+      "loss": 1.1377,
+      "step": 2219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754128096865425e-05,
+      "loss": 1.1484,
+      "step": 2220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753810820508986e-05,
+      "loss": 1.0347,
+      "step": 2221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753493342126955e-05,
+      "loss": 1.2148,
+      "step": 2222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975317566172591e-05,
+      "loss": 1.1938,
+      "step": 2223
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752857779312427e-05,
+      "loss": 1.2622,
+      "step": 2224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752539694893088e-05,
+      "loss": 1.0322,
+      "step": 2225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752221408474484e-05,
+      "loss": 1.251,
+      "step": 2226
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975190292006321e-05,
+      "loss": 1.3076,
+      "step": 2227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975158422966586e-05,
+      "loss": 1.2041,
+      "step": 2228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975126533728903e-05,
+      "loss": 1.0864,
+      "step": 2229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750946242939336e-05,
+      "loss": 1.105,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750626946623376e-05,
+      "loss": 1.21,
+      "step": 2231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750307448347772e-05,
+      "loss": 1.1445,
+      "step": 2232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749987748119137e-05,
+      "loss": 1.377,
+      "step": 2233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749667845944095e-05,
+      "loss": 1.0854,
+      "step": 2234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974934774182927e-05,
+      "loss": 1.2314,
+      "step": 2235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974902743578129e-05,
+      "loss": 1.1558,
+      "step": 2236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748706927806798e-05,
+      "loss": 1.2686,
+      "step": 2237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748386217912426e-05,
+      "loss": 1.272,
+      "step": 2238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748065306104814e-05,
+      "loss": 1.1279,
+      "step": 2239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9747744192390616e-05,
+      "loss": 1.2231,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974742287677648e-05,
+      "loss": 1.2124,
+      "step": 2241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9747101359269057e-05,
+      "loss": 1.1382,
+      "step": 2242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746779639875012e-05,
+      "loss": 1.2437,
+      "step": 2243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746457718601008e-05,
+      "loss": 1.1206,
+      "step": 2244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746135595453713e-05,
+      "loss": 1.1255,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745813270439794e-05,
+      "loss": 1.1318,
+      "step": 2246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974549074356593e-05,
+      "loss": 1.1709,
+      "step": 2247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745168014838805e-05,
+      "loss": 1.1963,
+      "step": 2248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744845084265102e-05,
+      "loss": 1.1978,
+      "step": 2249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744521951851504e-05,
+      "loss": 1.2881,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744198617604713e-05,
+      "loss": 1.2324,
+      "step": 2251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743875081531415e-05,
+      "loss": 1.168,
+      "step": 2252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743551343638324e-05,
+      "loss": 1.1792,
+      "step": 2253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743227403932135e-05,
+      "loss": 1.0435,
+      "step": 2254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974290326241956e-05,
+      "loss": 1.0229,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974257891910732e-05,
+      "loss": 1.1323,
+      "step": 2256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742254374002122e-05,
+      "loss": 1.0835,
+      "step": 2257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741929627110696e-05,
+      "loss": 1.1953,
+      "step": 2258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741604678439765e-05,
+      "loss": 0.9995,
+      "step": 2259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974127952799606e-05,
+      "loss": 1.2666,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740954175786317e-05,
+      "loss": 1.1426,
+      "step": 2261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974062862181727e-05,
+      "loss": 1.168,
+      "step": 2262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740302866095668e-05,
+      "loss": 1.0952,
+      "step": 2263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739976908628257e-05,
+      "loss": 1.0898,
+      "step": 2264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739650749421785e-05,
+      "loss": 1.2612,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739324388483013e-05,
+      "loss": 1.2119,
+      "step": 2266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738997825818696e-05,
+      "loss": 1.1694,
+      "step": 2267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738671061435596e-05,
+      "loss": 1.2168,
+      "step": 2268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738344095340488e-05,
+      "loss": 1.2183,
+      "step": 2269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738016927540143e-05,
+      "loss": 1.106,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973768955804133e-05,
+      "loss": 1.2437,
+      "step": 2271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737361986850838e-05,
+      "loss": 1.0688,
+      "step": 2272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973703421397545e-05,
+      "loss": 1.1157,
+      "step": 2273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736706239421953e-05,
+      "loss": 1.2256,
+      "step": 2274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736378063197138e-05,
+      "loss": 1.167,
+      "step": 2275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973604968530781e-05,
+      "loss": 1.0762,
+      "step": 2276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735721105760763e-05,
+      "loss": 1.2451,
+      "step": 2277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735392324562807e-05,
+      "loss": 1.0957,
+      "step": 2278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735063341720747e-05,
+      "loss": 1.2065,
+      "step": 2279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734734157241403e-05,
+      "loss": 1.0269,
+      "step": 2280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973440477113159e-05,
+      "loss": 1.0898,
+      "step": 2281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734075183398133e-05,
+      "loss": 1.2529,
+      "step": 2282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733745394047855e-05,
+      "loss": 1.4141,
+      "step": 2283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733415403087588e-05,
+      "loss": 1.2334,
+      "step": 2284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973308521052417e-05,
+      "loss": 1.1611,
+      "step": 2285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732754816364435e-05,
+      "loss": 1.2295,
+      "step": 2286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732424220615227e-05,
+      "loss": 1.3662,
+      "step": 2287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.97320934232834e-05,
+      "loss": 1.1216,
+      "step": 2288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731762424375794e-05,
+      "loss": 1.2461,
+      "step": 2289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731431223899275e-05,
+      "loss": 1.0439,
+      "step": 2290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731099821860702e-05,
+      "loss": 1.1895,
+      "step": 2291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730768218266933e-05,
+      "loss": 1.1343,
+      "step": 2292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973043641312484e-05,
+      "loss": 1.1519,
+      "step": 2293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730104406441303e-05,
+      "loss": 1.0293,
+      "step": 2294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9729772198223184e-05,
+      "loss": 1.1929,
+      "step": 2295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972943978847737e-05,
+      "loss": 1.1401,
+      "step": 2296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9729107177210754e-05,
+      "loss": 1.1274,
+      "step": 2297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728774364430212e-05,
+      "loss": 1.3584,
+      "step": 2298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728441350142646e-05,
+      "loss": 1.1909,
+      "step": 2299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972810813435495e-05,
+      "loss": 1.2539,
+      "step": 2300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727774717074027e-05,
+      "loss": 1.1187,
+      "step": 2301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727441098306785e-05,
+      "loss": 1.1719,
+      "step": 2302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727107278060132e-05,
+      "loss": 1.2148,
+      "step": 2303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972677325634098e-05,
+      "loss": 1.1553,
+      "step": 2304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726439033156252e-05,
+      "loss": 1.0972,
+      "step": 2305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972610460851287e-05,
+      "loss": 1.2178,
+      "step": 2306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725769982417756e-05,
+      "loss": 1.207,
+      "step": 2307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972543515487784e-05,
+      "loss": 1.1948,
+      "step": 2308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972510012590007e-05,
+      "loss": 1.2173,
+      "step": 2309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972476489549137e-05,
+      "loss": 1.0698,
+      "step": 2310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972442946365869e-05,
+      "loss": 1.0132,
+      "step": 2311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9724093830408978e-05,
+      "loss": 1.0918,
+      "step": 2312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723757995749186e-05,
+      "loss": 1.1172,
+      "step": 2313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972342195968627e-05,
+      "loss": 1.1597,
+      "step": 2314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723085722227186e-05,
+      "loss": 1.0398,
+      "step": 2315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722749283378905e-05,
+      "loss": 1.1143,
+      "step": 2316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722412643148392e-05,
+      "loss": 1.2427,
+      "step": 2317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722075801542616e-05,
+      "loss": 1.1387,
+      "step": 2318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721738758568564e-05,
+      "loss": 1.2256,
+      "step": 2319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721401514233204e-05,
+      "loss": 1.1304,
+      "step": 2320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972106406854353e-05,
+      "loss": 1.0806,
+      "step": 2321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9720726421506532e-05,
+      "loss": 1.147,
+      "step": 2322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97203885731292e-05,
+      "loss": 1.0962,
+      "step": 2323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972005052341853e-05,
+      "loss": 1.2456,
+      "step": 2324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719712272381523e-05,
+      "loss": 1.1494,
+      "step": 2325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719373820025192e-05,
+      "loss": 1.0923,
+      "step": 2326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719035166356544e-05,
+      "loss": 1.335,
+      "step": 2327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971869631138259e-05,
+      "loss": 1.2422,
+      "step": 2328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718357255110355e-05,
+      "loss": 1.146,
+      "step": 2329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718017997546853e-05,
+      "loss": 1.144,
+      "step": 2330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971767853869912e-05,
+      "loss": 1.2417,
+      "step": 2331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971733887857418e-05,
+      "loss": 1.1748,
+      "step": 2332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716999017179073e-05,
+      "loss": 1.1577,
+      "step": 2333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716658954520834e-05,
+      "loss": 1.3232,
+      "step": 2334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971631869060651e-05,
+      "loss": 1.2598,
+      "step": 2335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715978225443147e-05,
+      "loss": 1.2407,
+      "step": 2336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97156375590378e-05,
+      "loss": 1.1978,
+      "step": 2337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971529669139752e-05,
+      "loss": 1.2676,
+      "step": 2338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714955622529372e-05,
+      "loss": 1.2178,
+      "step": 2339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714614352440415e-05,
+      "loss": 1.0918,
+      "step": 2340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714272881137725e-05,
+      "loss": 1.1187,
+      "step": 2341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9713931208628368e-05,
+      "loss": 0.9971,
+      "step": 2342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971358933491942e-05,
+      "loss": 1.2188,
+      "step": 2343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9713247260017968e-05,
+      "loss": 1.207,
+      "step": 2344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97129049839311e-05,
+      "loss": 1.252,
+      "step": 2345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971256250666589e-05,
+      "loss": 1.1108,
+      "step": 2346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971221982822945e-05,
+      "loss": 1.165,
+      "step": 2347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711876948628862e-05,
+      "loss": 0.9749,
+      "step": 2348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971153386787124e-05,
+      "loss": 1.187,
+      "step": 2349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711190585963682e-05,
+      "loss": 1.1997,
+      "step": 2350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710847102913303e-05,
+      "loss": 1.165,
+      "step": 2351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971050341872722e-05,
+      "loss": 1.0889,
+      "step": 2352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710159533412542e-05,
+      "loss": 1.2261,
+      "step": 2353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9709815446976396e-05,
+      "loss": 1.1509,
+      "step": 2354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970947115942591e-05,
+      "loss": 1.106,
+      "step": 2355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970912667076822e-05,
+      "loss": 1.0364,
+      "step": 2356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970878198101045e-05,
+      "loss": 1.2163,
+      "step": 2357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970843709015975e-05,
+      "loss": 1.1265,
+      "step": 2358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708091998223257e-05,
+      "loss": 1.1953,
+      "step": 2359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707746705208122e-05,
+      "loss": 1.1069,
+      "step": 2360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707401211121495e-05,
+      "loss": 1.1143,
+      "step": 2361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707055515970533e-05,
+      "loss": 1.1475,
+      "step": 2362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9706709619762393e-05,
+      "loss": 1.1514,
+      "step": 2363
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9706363522504248e-05,
+      "loss": 1.0645,
+      "step": 2364
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970601722420326e-05,
+      "loss": 1.0771,
+      "step": 2365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.97056707248666e-05,
+      "loss": 1.1562,
+      "step": 2366
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705324024501446e-05,
+      "loss": 1.1694,
+      "step": 2367
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970497712311498e-05,
+      "loss": 1.1587,
+      "step": 2368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970463002071439e-05,
+      "loss": 1.2041,
+      "step": 2369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970428271730686e-05,
+      "loss": 1.1724,
+      "step": 2370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703935212899592e-05,
+      "loss": 1.1792,
+      "step": 2371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703587507499777e-05,
+      "loss": 1.1963,
+      "step": 2372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970323960111461e-05,
+      "loss": 1.1157,
+      "step": 2373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702891493751313e-05,
+      "loss": 1.1172,
+      "step": 2374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970254318541709e-05,
+      "loss": 1.2007,
+      "step": 2375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702194676119147e-05,
+      "loss": 1.1013,
+      "step": 2376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701845965864712e-05,
+      "loss": 1.2334,
+      "step": 2377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701497054661004e-05,
+      "loss": 1.1294,
+      "step": 2378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970114794251525e-05,
+      "loss": 1.2114,
+      "step": 2379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970079862943468e-05,
+      "loss": 1.1646,
+      "step": 2380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970044911542653e-05,
+      "loss": 1.2046,
+      "step": 2381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700099400498044e-05,
+      "loss": 1.1392,
+      "step": 2382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969974948465646e-05,
+      "loss": 1.0796,
+      "step": 2383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9699399367909025e-05,
+      "loss": 1.2383,
+      "step": 2384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9699049050262996e-05,
+      "loss": 1.29,
+      "step": 2385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698698531725623e-05,
+      "loss": 1.167,
+      "step": 2386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969834781230417e-05,
+      "loss": 1.1426,
+      "step": 2387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.96979968920059e-05,
+      "loss": 1.1191,
+      "step": 2388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969764577083808e-05,
+      "loss": 1.0669,
+      "step": 2389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697294448807985e-05,
+      "loss": 1.1826,
+      "step": 2390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969694292592289e-05,
+      "loss": 1.1069,
+      "step": 2391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696591202190077e-05,
+      "loss": 1.2637,
+      "step": 2392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969623927761683e-05,
+      "loss": 1.2407,
+      "step": 2393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969588715221044e-05,
+      "loss": 1.2202,
+      "step": 2394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695534825978203e-05,
+      "loss": 1.0605,
+      "step": 2395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969518229892741e-05,
+      "loss": 1.209,
+      "step": 2396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694829571065364e-05,
+      "loss": 1.2471,
+      "step": 2397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694476642399372e-05,
+      "loss": 1.0259,
+      "step": 2398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694123512936747e-05,
+      "loss": 1.2173,
+      "step": 2399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.96937701826848e-05,
+      "loss": 1.2236,
+      "step": 2400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969341665165085e-05,
+      "loss": 1.1372,
+      "step": 2401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9693062919842218e-05,
+      "loss": 1.3203,
+      "step": 2402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692708987266235e-05,
+      "loss": 1.1489,
+      "step": 2403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692354853930227e-05,
+      "loss": 1.2397,
+      "step": 2404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692000519841535e-05,
+      "loss": 1.0107,
+      "step": 2405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969164598500749e-05,
+      "loss": 1.0005,
+      "step": 2406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969129124943544e-05,
+      "loss": 1.2163,
+      "step": 2407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9690936313132732e-05,
+      "loss": 1.2017,
+      "step": 2408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9690581176106717e-05,
+      "loss": 1.1616,
+      "step": 2409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9690225838364753e-05,
+      "loss": 1.1895,
+      "step": 2410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689870299914198e-05,
+      "loss": 1.1367,
+      "step": 2411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689514560762415e-05,
+      "loss": 1.2646,
+      "step": 2412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689158620916776e-05,
+      "loss": 1.1147,
+      "step": 2413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968880248038465e-05,
+      "loss": 1.147,
+      "step": 2414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968844613917341e-05,
+      "loss": 0.9299,
+      "step": 2415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968808959729045e-05,
+      "loss": 1.1929,
+      "step": 2416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687732854743138e-05,
+      "loss": 1.1489,
+      "step": 2417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687375911538872e-05,
+      "loss": 1.1426,
+      "step": 2418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687018767685048e-05,
+      "loss": 1.1611,
+      "step": 2419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968666142318906e-05,
+      "loss": 1.1592,
+      "step": 2420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686303878058307e-05,
+      "loss": 1.165,
+      "step": 2421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685946132300197e-05,
+      "loss": 1.127,
+      "step": 2422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968558818592214e-05,
+      "loss": 1.2397,
+      "step": 2423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685230038931546e-05,
+      "loss": 1.1567,
+      "step": 2424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968487169133584e-05,
+      "loss": 1.2197,
+      "step": 2425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9684513143142442e-05,
+      "loss": 1.1313,
+      "step": 2426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9684154394358775e-05,
+      "loss": 1.2031,
+      "step": 2427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968379544499227e-05,
+      "loss": 1.1655,
+      "step": 2428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968343629505037e-05,
+      "loss": 1.0693,
+      "step": 2429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9683076944540503e-05,
+      "loss": 1.2192,
+      "step": 2430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682717393470118e-05,
+      "loss": 1.3311,
+      "step": 2431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968235764184666e-05,
+      "loss": 1.1055,
+      "step": 2432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681997689677577e-05,
+      "loss": 1.1567,
+      "step": 2433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681637536970336e-05,
+      "loss": 1.2705,
+      "step": 2434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681277183732384e-05,
+      "loss": 1.1294,
+      "step": 2435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9680916629971194e-05,
+      "loss": 1.0166,
+      "step": 2436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9680555875694228e-05,
+      "loss": 1.2334,
+      "step": 2437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968019492090896e-05,
+      "loss": 1.1221,
+      "step": 2438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9679833765622867e-05,
+      "loss": 1.1484,
+      "step": 2439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9679472409843428e-05,
+      "loss": 1.1641,
+      "step": 2440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9679110853578127e-05,
+      "loss": 1.0566,
+      "step": 2441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967874909683446e-05,
+      "loss": 1.2422,
+      "step": 2442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678387139619906e-05,
+      "loss": 1.3076,
+      "step": 2443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678024981941975e-05,
+      "loss": 1.21,
+      "step": 2444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967766262380816e-05,
+      "loss": 0.9849,
+      "step": 2445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9677300065225974e-05,
+      "loss": 1.1138,
+      "step": 2446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676937306202918e-05,
+      "loss": 1.1797,
+      "step": 2447
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967657434674651e-05,
+      "loss": 1.209,
+      "step": 2448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967621118686427e-05,
+      "loss": 1.0879,
+      "step": 2449
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675847826563716e-05,
+      "loss": 1.1187,
+      "step": 2450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675484265852374e-05,
+      "loss": 1.1284,
+      "step": 2451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967512050473778e-05,
+      "loss": 1.1943,
+      "step": 2452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967475654322746e-05,
+      "loss": 1.1279,
+      "step": 2453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9674392381328958e-05,
+      "loss": 1.1875,
+      "step": 2454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9674028019049817e-05,
+      "loss": 1.1831,
+      "step": 2455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673663456397578e-05,
+      "loss": 1.0649,
+      "step": 2456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96732986933798e-05,
+      "loss": 1.1602,
+      "step": 2457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9672933730004032e-05,
+      "loss": 1.3208,
+      "step": 2458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967256856627784e-05,
+      "loss": 1.1924,
+      "step": 2459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9672203202208777e-05,
+      "loss": 1.0635,
+      "step": 2460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967183763780442e-05,
+      "loss": 1.2427,
+      "step": 2461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9671471873072336e-05,
+      "loss": 1.022,
+      "step": 2462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9671105908020105e-05,
+      "loss": 1.0801,
+      "step": 2463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96707397426553e-05,
+      "loss": 1.2578,
+      "step": 2464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670373376985515e-05,
+      "loss": 1.2451,
+      "step": 2465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967000681101833e-05,
+      "loss": 1.0847,
+      "step": 2466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966964004476134e-05,
+      "loss": 1.1445,
+      "step": 2467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9669273078222142e-05,
+      "loss": 1.2139,
+      "step": 2468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9668905911408337e-05,
+      "loss": 1.2061,
+      "step": 2469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966853854432753e-05,
+      "loss": 1.1724,
+      "step": 2470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966817097698733e-05,
+      "loss": 1.2446,
+      "step": 2471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966780320939535e-05,
+      "loss": 1.2017,
+      "step": 2472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667435241559207e-05,
+      "loss": 1.145,
+      "step": 2473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667067073486522e-05,
+      "loss": 1.1372,
+      "step": 2474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9666698705184924e-05,
+      "loss": 1.0806,
+      "step": 2475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966633013666204e-05,
+      "loss": 1.1079,
+      "step": 2476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665961367925502e-05,
+      "loss": 1.2617,
+      "step": 2477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665592398982952e-05,
+      "loss": 1.1841,
+      "step": 2478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665223229842033e-05,
+      "loss": 1.062,
+      "step": 2479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966485386051039e-05,
+      "loss": 1.1182,
+      "step": 2480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966448429099567e-05,
+      "loss": 1.21,
+      "step": 2481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664114521305528e-05,
+      "loss": 1.1426,
+      "step": 2482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966374455144763e-05,
+      "loss": 1.2173,
+      "step": 2483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663374381429634e-05,
+      "loss": 1.1895,
+      "step": 2484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96630040112592e-05,
+      "loss": 1.1304,
+      "step": 2485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9662633440944018e-05,
+      "loss": 1.1699,
+      "step": 2486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9662262670491744e-05,
+      "loss": 1.2368,
+      "step": 2487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661891699910068e-05,
+      "loss": 1.2063,
+      "step": 2488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661520529206673e-05,
+      "loss": 1.1362,
+      "step": 2489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661149158389243e-05,
+      "loss": 1.2129,
+      "step": 2490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660777587465474e-05,
+      "loss": 1.2227,
+      "step": 2491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966040581644306e-05,
+      "loss": 1.166,
+      "step": 2492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96600338453297e-05,
+      "loss": 1.2612,
+      "step": 2493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9659661674133103e-05,
+      "loss": 1.1836,
+      "step": 2494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9659289302860977e-05,
+      "loss": 1.2095,
+      "step": 2495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658916731521028e-05,
+      "loss": 1.1436,
+      "step": 2496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658543960120976e-05,
+      "loss": 1.0078,
+      "step": 2497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658170988668547e-05,
+      "loss": 1.2021,
+      "step": 2498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657797817171463e-05,
+      "loss": 1.209,
+      "step": 2499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657424445637452e-05,
+      "loss": 1.1465,
+      "step": 2500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657050874074246e-05,
+      "loss": 1.2119,
+      "step": 2501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965667710248959e-05,
+      "loss": 1.1797,
+      "step": 2502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9656303130891214e-05,
+      "loss": 1.1826,
+      "step": 2503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655928959286875e-05,
+      "loss": 0.9858,
+      "step": 2504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655554587684316e-05,
+      "loss": 1.1138,
+      "step": 2505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655180016091298e-05,
+      "loss": 1.2456,
+      "step": 2506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9654805244515573e-05,
+      "loss": 1.0752,
+      "step": 2507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9654430272964903e-05,
+      "loss": 1.2158,
+      "step": 2508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965405510144706e-05,
+      "loss": 1.1646,
+      "step": 2509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965367972996981e-05,
+      "loss": 1.0718,
+      "step": 2510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9653304158540933e-05,
+      "loss": 1.1284,
+      "step": 2511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652928387168202e-05,
+      "loss": 1.0972,
+      "step": 2512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652552415859402e-05,
+      "loss": 1.1353,
+      "step": 2513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965217624462232e-05,
+      "loss": 1.1143,
+      "step": 2514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651799873464754e-05,
+      "loss": 1.0645,
+      "step": 2515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965142330239449e-05,
+      "loss": 1.1113,
+      "step": 2516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651046531419335e-05,
+      "loss": 1.1519,
+      "step": 2517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650669560547085e-05,
+      "loss": 1.2324,
+      "step": 2518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650292389785557e-05,
+      "loss": 1.1099,
+      "step": 2519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964991501914256e-05,
+      "loss": 1.2637,
+      "step": 2520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9649537448625905e-05,
+      "loss": 1.02,
+      "step": 2521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964915967824342e-05,
+      "loss": 1.1274,
+      "step": 2522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9648781708002923e-05,
+      "loss": 1.2788,
+      "step": 2523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964840353791225e-05,
+      "loss": 1.2124,
+      "step": 2524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9648025167979224e-05,
+      "loss": 1.106,
+      "step": 2525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964764659821169e-05,
+      "loss": 1.0283,
+      "step": 2526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647267828617493e-05,
+      "loss": 0.9561,
+      "step": 2527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646888859204462e-05,
+      "loss": 1.1128,
+      "step": 2528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646509689980462e-05,
+      "loss": 1.0503,
+      "step": 2529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964613032095334e-05,
+      "loss": 1.0869,
+      "step": 2530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645750752130953e-05,
+      "loss": 1.0476,
+      "step": 2531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645370983521168e-05,
+      "loss": 1.126,
+      "step": 2532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964499101513185e-05,
+      "loss": 1.1394,
+      "step": 2533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964461084697086e-05,
+      "loss": 1.1709,
+      "step": 2534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964423047904608e-05,
+      "loss": 1.1211,
+      "step": 2535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643849911365388e-05,
+      "loss": 1.085,
+      "step": 2536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643469143936666e-05,
+      "loss": 1.1406,
+      "step": 2537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96430881767678e-05,
+      "loss": 1.2104,
+      "step": 2538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964270700986668e-05,
+      "loss": 1.293,
+      "step": 2539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642325643241206e-05,
+      "loss": 1.1855,
+      "step": 2540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641944076899268e-05,
+      "loss": 1.0869,
+      "step": 2541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641562310848777e-05,
+      "loss": 1.1768,
+      "step": 2542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641180345097634e-05,
+      "loss": 1.0605,
+      "step": 2543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9640798179653758e-05,
+      "loss": 1.146,
+      "step": 2544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9640415814525056e-05,
+      "loss": 1.1777,
+      "step": 2545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9640033249719457e-05,
+      "loss": 1.1367,
+      "step": 2546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639650485244878e-05,
+      "loss": 1.1973,
+      "step": 2547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963926752110925e-05,
+      "loss": 1.1001,
+      "step": 2548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638884357320505e-05,
+      "loss": 1.1187,
+      "step": 2549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638500993886575e-05,
+      "loss": 1.0957,
+      "step": 2550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638117430815408e-05,
+      "loss": 1.0244,
+      "step": 2551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637733668114944e-05,
+      "loss": 1.1357,
+      "step": 2552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637349705793127e-05,
+      "loss": 1.1426,
+      "step": 2553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9636965543857923e-05,
+      "loss": 1.2002,
+      "step": 2554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9636581182317275e-05,
+      "loss": 1.0547,
+      "step": 2555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963619662117915e-05,
+      "loss": 1.1611,
+      "step": 2556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635811860451514e-05,
+      "loss": 1.2007,
+      "step": 2557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963542690014234e-05,
+      "loss": 1.1328,
+      "step": 2558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635041740259592e-05,
+      "loss": 1.1934,
+      "step": 2559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963465638081125e-05,
+      "loss": 1.2241,
+      "step": 2560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634270821805305e-05,
+      "loss": 1.1465,
+      "step": 2561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633885063249733e-05,
+      "loss": 1.1221,
+      "step": 2562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633499105152532e-05,
+      "loss": 1.1306,
+      "step": 2563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633112947521686e-05,
+      "loss": 1.2505,
+      "step": 2564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.96327265903652e-05,
+      "loss": 1.1333,
+      "step": 2565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9632340033691075e-05,
+      "loss": 0.9614,
+      "step": 2566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963195327750732e-05,
+      "loss": 1.291,
+      "step": 2567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963156632182194e-05,
+      "loss": 1.0796,
+      "step": 2568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631179166642957e-05,
+      "loss": 1.1797,
+      "step": 2569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963079181197838e-05,
+      "loss": 1.042,
+      "step": 2570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9630404257836245e-05,
+      "loss": 1.2568,
+      "step": 2571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963001650422457e-05,
+      "loss": 1.2559,
+      "step": 2572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629628551151385e-05,
+      "loss": 1.0234,
+      "step": 2573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629240398624734e-05,
+      "loss": 1.0625,
+      "step": 2574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962885204665265e-05,
+      "loss": 1.127,
+      "step": 2575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962846349524318e-05,
+      "loss": 1.2812,
+      "step": 2576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9628074744404367e-05,
+      "loss": 1.1313,
+      "step": 2577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962768579414427e-05,
+      "loss": 1.251,
+      "step": 2578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962729664447094e-05,
+      "loss": 1.2466,
+      "step": 2579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626907295392436e-05,
+      "loss": 1.104,
+      "step": 2580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962651774691683e-05,
+      "loss": 1.1074,
+      "step": 2581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626127999052182e-05,
+      "loss": 1.2061,
+      "step": 2582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625738051806573e-05,
+      "loss": 1.1748,
+      "step": 2583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625347905188073e-05,
+      "loss": 1.1973,
+      "step": 2584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624957559204763e-05,
+      "loss": 1.0781,
+      "step": 2585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962456701386473e-05,
+      "loss": 1.1602,
+      "step": 2586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624176269176065e-05,
+      "loss": 1.2671,
+      "step": 2587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9623785325146857e-05,
+      "loss": 1.1631,
+      "step": 2588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962339418178521e-05,
+      "loss": 1.1499,
+      "step": 2589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962300283909922e-05,
+      "loss": 1.1616,
+      "step": 2590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962261129709699e-05,
+      "loss": 1.2119,
+      "step": 2591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962221955578664e-05,
+      "loss": 1.1704,
+      "step": 2592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621827615176275e-05,
+      "loss": 0.9976,
+      "step": 2593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621435475274016e-05,
+      "loss": 1.1387,
+      "step": 2594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621043136087984e-05,
+      "loss": 1.1328,
+      "step": 2595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962065059762631e-05,
+      "loss": 1.1157,
+      "step": 2596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620257859897116e-05,
+      "loss": 1.2275,
+      "step": 2597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619864922908546e-05,
+      "loss": 1.0967,
+      "step": 2598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961947178666873e-05,
+      "loss": 1.3032,
+      "step": 2599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619078451185814e-05,
+      "loss": 1.2251,
+      "step": 2600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961868491646795e-05,
+      "loss": 1.103,
+      "step": 2601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618291182523285e-05,
+      "loss": 1.0513,
+      "step": 2602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617897249359973e-05,
+      "loss": 1.1392,
+      "step": 2603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617503116986175e-05,
+      "loss": 1.1621,
+      "step": 2604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617108785410054e-05,
+      "loss": 1.2656,
+      "step": 2605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961671425463978e-05,
+      "loss": 1.1152,
+      "step": 2606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616319524683516e-05,
+      "loss": 1.1333,
+      "step": 2607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961592459554945e-05,
+      "loss": 1.0422,
+      "step": 2608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9615529467245754e-05,
+      "loss": 1.1104,
+      "step": 2609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9615134139780614e-05,
+      "loss": 1.2979,
+      "step": 2610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961473861316222e-05,
+      "loss": 1.2148,
+      "step": 2611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9614342887398758e-05,
+      "loss": 1.2466,
+      "step": 2612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613946962498433e-05,
+      "loss": 1.1694,
+      "step": 2613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613550838469442e-05,
+      "loss": 1.2764,
+      "step": 2614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961315451531999e-05,
+      "loss": 1.2061,
+      "step": 2615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9612757993058282e-05,
+      "loss": 1.1416,
+      "step": 2616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9612361271692535e-05,
+      "loss": 1.1792,
+      "step": 2617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611964351230966e-05,
+      "loss": 1.1626,
+      "step": 2618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611567231681797e-05,
+      "loss": 1.0957,
+      "step": 2619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961116991305325e-05,
+      "loss": 1.1318,
+      "step": 2620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610772395353562e-05,
+      "loss": 1.1816,
+      "step": 2621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610374678590952e-05,
+      "loss": 1.2139,
+      "step": 2622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609976762773672e-05,
+      "loss": 1.3359,
+      "step": 2623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609578647909958e-05,
+      "loss": 1.2695,
+      "step": 2624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609180334008056e-05,
+      "loss": 1.1357,
+      "step": 2625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608781821076218e-05,
+      "loss": 1.1919,
+      "step": 2626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608383109122692e-05,
+      "loss": 1.2471,
+      "step": 2627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9607984198155744e-05,
+      "loss": 1.1812,
+      "step": 2628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9607585088183635e-05,
+      "loss": 1.144,
+      "step": 2629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9607185779214627e-05,
+      "loss": 1.2783,
+      "step": 2630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9606786271256997e-05,
+      "loss": 1.1885,
+      "step": 2631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9606386564319012e-05,
+      "loss": 1.1436,
+      "step": 2632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960598665840896e-05,
+      "loss": 1.1782,
+      "step": 2633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9605586553535115e-05,
+      "loss": 1.3301,
+      "step": 2634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9605186249705772e-05,
+      "loss": 1.1199,
+      "step": 2635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604785746929216e-05,
+      "loss": 1.1416,
+      "step": 2636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960438504521375e-05,
+      "loss": 1.2427,
+      "step": 2637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603984144567664e-05,
+      "loss": 1.166,
+      "step": 2638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603583044999266e-05,
+      "loss": 1.1514,
+      "step": 2639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603181746516866e-05,
+      "loss": 1.1982,
+      "step": 2640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602780249128775e-05,
+      "loss": 1.0933,
+      "step": 2641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602378552843308e-05,
+      "loss": 1.1558,
+      "step": 2642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9601976657668785e-05,
+      "loss": 1.2256,
+      "step": 2643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960157456361353e-05,
+      "loss": 1.0864,
+      "step": 2644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960117227068587e-05,
+      "loss": 1.252,
+      "step": 2645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9600769778894143e-05,
+      "loss": 1.1748,
+      "step": 2646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960036708824668e-05,
+      "loss": 1.3252,
+      "step": 2647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599964198751823e-05,
+      "loss": 1.0864,
+      "step": 2648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959956111041792e-05,
+      "loss": 1.1294,
+      "step": 2649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599157823253313e-05,
+      "loss": 1.144,
+      "step": 2650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9598754337266365e-05,
+      "loss": 1.0938,
+      "step": 2651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9598350652465422e-05,
+      "loss": 0.9277,
+      "step": 2652
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597946768858854e-05,
+      "loss": 1.1138,
+      "step": 2653
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959754268645502e-05,
+      "loss": 1.0703,
+      "step": 2654
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.95971384052623e-05,
+      "loss": 1.1631,
+      "step": 2655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9596733925289056e-05,
+      "loss": 1.1626,
+      "step": 2656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959632924654367e-05,
+      "loss": 1.1123,
+      "step": 2657
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595924369034524e-05,
+      "loss": 1.085,
+      "step": 2658
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595519292770007e-05,
+      "loss": 1.1475,
+      "step": 2659
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595114017758506e-05,
+      "loss": 1.082,
+      "step": 2660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9594708544008414e-05,
+      "loss": 1.1411,
+      "step": 2661
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9594302871528136e-05,
+      "loss": 1.2168,
+      "step": 2662
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593897000326064e-05,
+      "loss": 1.126,
+      "step": 2663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593490930410612e-05,
+      "loss": 1.126,
+      "step": 2664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593084661790186e-05,
+      "loss": 1.1211,
+      "step": 2665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9592678194473208e-05,
+      "loss": 1.1738,
+      "step": 2666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959227152846809e-05,
+      "loss": 1.1895,
+      "step": 2667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959186466378326e-05,
+      "loss": 1.208,
+      "step": 2668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959145760042714e-05,
+      "loss": 1.1455,
+      "step": 2669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591050338408168e-05,
+      "loss": 1.1758,
+      "step": 2670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590642877734775e-05,
+      "loss": 1.1001,
+      "step": 2671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590235218415397e-05,
+      "loss": 1.2173,
+      "step": 2672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589827360458486e-05,
+      "loss": 1.1016,
+      "step": 2673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589419303872485e-05,
+      "loss": 1.1226,
+      "step": 2674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589011048665846e-05,
+      "loss": 1.084,
+      "step": 2675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9588602594847023e-05,
+      "loss": 1.0161,
+      "step": 2676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9588193942424484e-05,
+      "loss": 1.106,
+      "step": 2677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958778509140668e-05,
+      "loss": 1.0747,
+      "step": 2678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958737604180209e-05,
+      "loss": 1.2861,
+      "step": 2679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9586966793619186e-05,
+      "loss": 1.1802,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958655734686644e-05,
+      "loss": 1.0889,
+      "step": 2681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9586147701552335e-05,
+      "loss": 1.0928,
+      "step": 2682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9585737857685355e-05,
+      "loss": 1.0498,
+      "step": 2683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958532781527399e-05,
+      "loss": 1.0396,
+      "step": 2684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9584917574326732e-05,
+      "loss": 1.1255,
+      "step": 2685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9584507134852075e-05,
+      "loss": 0.9902,
+      "step": 2686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958409649685853e-05,
+      "loss": 1.2124,
+      "step": 2687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583685660354593e-05,
+      "loss": 1.1338,
+      "step": 2688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583274625348773e-05,
+      "loss": 1.2783,
+      "step": 2689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9582863391849592e-05,
+      "loss": 1.207,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958245195986556e-05,
+      "loss": 1.2378,
+      "step": 2691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95820403294052e-05,
+      "loss": 1.2842,
+      "step": 2692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581628500477042e-05,
+      "loss": 1.2197,
+      "step": 2693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958121647308961e-05,
+      "loss": 1.2085,
+      "step": 2694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9580804247251443e-05,
+      "loss": 1.2402,
+      "step": 2695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9580391822971077e-05,
+      "loss": 1.106,
+      "step": 2696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957997920025706e-05,
+      "loss": 1.0596,
+      "step": 2697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9579566379117925e-05,
+      "loss": 1.1714,
+      "step": 2698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9579153359562235e-05,
+      "loss": 1.1611,
+      "step": 2699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957874014159854e-05,
+      "loss": 1.27,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95783267252354e-05,
+      "loss": 1.0093,
+      "step": 2701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577913110481377e-05,
+      "loss": 1.3057,
+      "step": 2702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957749929734504e-05,
+      "loss": 1.1265,
+      "step": 2703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577085285834953e-05,
+      "loss": 1.1411,
+      "step": 2704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576671075959698e-05,
+      "loss": 1.1279,
+      "step": 2705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576256667727857e-05,
+      "loss": 1.2168,
+      "step": 2706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575842061148008e-05,
+      "loss": 1.2319,
+      "step": 2707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575427256228737e-05,
+      "loss": 1.1855,
+      "step": 2708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957501225297864e-05,
+      "loss": 1.1372,
+      "step": 2709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957459705140631e-05,
+      "loss": 1.0933,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957418165152035e-05,
+      "loss": 1.1519,
+      "step": 2711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957376605332936e-05,
+      "loss": 1.2432,
+      "step": 2712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957335025684195e-05,
+      "loss": 1.1069,
+      "step": 2713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572934262066732e-05,
+      "loss": 1.1807,
+      "step": 2714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572518069012323e-05,
+      "loss": 1.1797,
+      "step": 2715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572101677687342e-05,
+      "loss": 1.2095,
+      "step": 2716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571685088100415e-05,
+      "loss": 1.1831,
+      "step": 2717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571268300260172e-05,
+      "loss": 1.1108,
+      "step": 2718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570851314175242e-05,
+      "loss": 1.1841,
+      "step": 2719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570434129854262e-05,
+      "loss": 1.1245,
+      "step": 2720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570016747305874e-05,
+      "loss": 1.0659,
+      "step": 2721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.956959916653873e-05,
+      "loss": 1.1299,
+      "step": 2722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569181387561467e-05,
+      "loss": 1.0791,
+      "step": 2723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9568763410382743e-05,
+      "loss": 1.1572,
+      "step": 2724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9568345235011216e-05,
+      "loss": 1.1772,
+      "step": 2725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567926861455547e-05,
+      "loss": 1.1211,
+      "step": 2726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567508289724405e-05,
+      "loss": 1.0483,
+      "step": 2727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567089519826455e-05,
+      "loss": 1.145,
+      "step": 2728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9566670551770375e-05,
+      "loss": 1.2393,
+      "step": 2729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956625138556484e-05,
+      "loss": 1.1514,
+      "step": 2730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956583202121853e-05,
+      "loss": 1.2358,
+      "step": 2731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9565412458740136e-05,
+      "loss": 1.0518,
+      "step": 2732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9564992698138346e-05,
+      "loss": 1.2256,
+      "step": 2733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956457273942185e-05,
+      "loss": 1.0054,
+      "step": 2734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956415258259935e-05,
+      "loss": 1.1704,
+      "step": 2735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563732227679554e-05,
+      "loss": 1.2769,
+      "step": 2736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563311674671158e-05,
+      "loss": 1.0654,
+      "step": 2737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562890923582884e-05,
+      "loss": 1.1494,
+      "step": 2738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562469974423436e-05,
+      "loss": 1.144,
+      "step": 2739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562048827201542e-05,
+      "loss": 1.1597,
+      "step": 2740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561627481925917e-05,
+      "loss": 1.2622,
+      "step": 2741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956120593860529e-05,
+      "loss": 1.1514,
+      "step": 2742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95607841972484e-05,
+      "loss": 1.2188,
+      "step": 2743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560362257863974e-05,
+      "loss": 1.166,
+      "step": 2744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955994012046075e-05,
+      "loss": 1.1797,
+      "step": 2745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955951778504748e-05,
+      "loss": 1.1094,
+      "step": 2746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9559095251632906e-05,
+      "loss": 1.1606,
+      "step": 2747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558672520225782e-05,
+      "loss": 1.1943,
+      "step": 2748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955824959083486e-05,
+      "loss": 1.1724,
+      "step": 2749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557826463468902e-05,
+      "loss": 1.1509,
+      "step": 2750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557403138136672e-05,
+      "loss": 1.2632,
+      "step": 2751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955697961484694e-05,
+      "loss": 1.2568,
+      "step": 2752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556555893608476e-05,
+      "loss": 1.2637,
+      "step": 2753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556131974430056e-05,
+      "loss": 1.125,
+      "step": 2754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9555707857320466e-05,
+      "loss": 1.1299,
+      "step": 2755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9555283542288483e-05,
+      "loss": 1.1279,
+      "step": 2756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554859029342898e-05,
+      "loss": 1.1582,
+      "step": 2757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554434318492506e-05,
+      "loss": 1.0913,
+      "step": 2758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95540094097461e-05,
+      "loss": 1.1929,
+      "step": 2759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9553584303112487e-05,
+      "loss": 1.0586,
+      "step": 2760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9553158998600463e-05,
+      "loss": 1.208,
+      "step": 2761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552733496218843e-05,
+      "loss": 1.1479,
+      "step": 2762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552307795976442e-05,
+      "loss": 1.0815,
+      "step": 2763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551881897882073e-05,
+      "loss": 1.1992,
+      "step": 2764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551455801944558e-05,
+      "loss": 1.3057,
+      "step": 2765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551029508172727e-05,
+      "loss": 1.1821,
+      "step": 2766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550603016575402e-05,
+      "loss": 1.2969,
+      "step": 2767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550176327161423e-05,
+      "loss": 1.1865,
+      "step": 2768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9549749439939623e-05,
+      "loss": 1.0933,
+      "step": 2769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9549322354918853e-05,
+      "loss": 1.1846,
+      "step": 2770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548895072107947e-05,
+      "loss": 0.9863,
+      "step": 2771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548467591515763e-05,
+      "loss": 1.1226,
+      "step": 2772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954803991315115e-05,
+      "loss": 1.0503,
+      "step": 2773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9547612037022975e-05,
+      "loss": 1.0874,
+      "step": 2774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954718396314009e-05,
+      "loss": 1.1245,
+      "step": 2775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9546755691511367e-05,
+      "loss": 1.3086,
+      "step": 2776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9546327222145673e-05,
+      "loss": 1.1758,
+      "step": 2777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545898555051888e-05,
+      "loss": 1.1621,
+      "step": 2778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545469690238885e-05,
+      "loss": 1.1504,
+      "step": 2779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545040627715554e-05,
+      "loss": 1.2266,
+      "step": 2780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9544611367490776e-05,
+      "loss": 1.2759,
+      "step": 2781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954418190957344e-05,
+      "loss": 1.248,
+      "step": 2782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954375225397245e-05,
+      "loss": 1.126,
+      "step": 2783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95433224006967e-05,
+      "loss": 1.2866,
+      "step": 2784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954289234975509e-05,
+      "loss": 1.2881,
+      "step": 2785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9542462101156534e-05,
+      "loss": 1.021,
+      "step": 2786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9542031654909942e-05,
+      "loss": 1.2363,
+      "step": 2787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541601011024224e-05,
+      "loss": 1.1587,
+      "step": 2788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541170169508304e-05,
+      "loss": 1.1748,
+      "step": 2789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540739130371107e-05,
+      "loss": 1.1079,
+      "step": 2790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540307893621558e-05,
+      "loss": 1.0527,
+      "step": 2791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539876459268593e-05,
+      "loss": 1.1577,
+      "step": 2792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539444827321144e-05,
+      "loss": 1.0757,
+      "step": 2793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539012997788152e-05,
+      "loss": 1.0923,
+      "step": 2794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953858097067856e-05,
+      "loss": 1.0566,
+      "step": 2795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953814874600132e-05,
+      "loss": 1.1318,
+      "step": 2796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537716323765386e-05,
+      "loss": 1.0713,
+      "step": 2797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537283703979708e-05,
+      "loss": 1.041,
+      "step": 2798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953685088665325e-05,
+      "loss": 1.1943,
+      "step": 2799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9536417871794978e-05,
+      "loss": 1.0752,
+      "step": 2800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535984659413855e-05,
+      "loss": 1.3223,
+      "step": 2801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535551249518866e-05,
+      "loss": 1.2168,
+      "step": 2802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535117642118972e-05,
+      "loss": 1.2842,
+      "step": 2803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953468383722317e-05,
+      "loss": 1.1309,
+      "step": 2804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953424983484043e-05,
+      "loss": 1.2617,
+      "step": 2805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533815634979754e-05,
+      "loss": 1.1348,
+      "step": 2806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533381237650126e-05,
+      "loss": 0.958,
+      "step": 2807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532946642860554e-05,
+      "loss": 1.1382,
+      "step": 2808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532511850620027e-05,
+      "loss": 1.2197,
+      "step": 2809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953207686093756e-05,
+      "loss": 1.1353,
+      "step": 2810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531641673822157e-05,
+      "loss": 1.2148,
+      "step": 2811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531206289282836e-05,
+      "loss": 1.1489,
+      "step": 2812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9530770707328616e-05,
+      "loss": 1.1455,
+      "step": 2813
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9530334927968512e-05,
+      "loss": 1.1206,
+      "step": 2814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952989895121155e-05,
+      "loss": 1.1763,
+      "step": 2815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952946277706677e-05,
+      "loss": 1.1606,
+      "step": 2816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9529026405543204e-05,
+      "loss": 1.2241,
+      "step": 2817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528589836649883e-05,
+      "loss": 1.2109,
+      "step": 2818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528153070395853e-05,
+      "loss": 1.1289,
+      "step": 2819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9527716106790163e-05,
+      "loss": 1.0991,
+      "step": 2820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952727894584186e-05,
+      "loss": 1.1064,
+      "step": 2821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9526841587559996e-05,
+      "loss": 1.1924,
+      "step": 2822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952640403195364e-05,
+      "loss": 1.2808,
+      "step": 2823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525966279031846e-05,
+      "loss": 1.1865,
+      "step": 2824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525528328803687e-05,
+      "loss": 1.2017,
+      "step": 2825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525090181278227e-05,
+      "loss": 1.1191,
+      "step": 2826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524651836464547e-05,
+      "loss": 1.2612,
+      "step": 2827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524213294371722e-05,
+      "loss": 1.1694,
+      "step": 2828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952377455500884e-05,
+      "loss": 1.0449,
+      "step": 2829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9523335618384984e-05,
+      "loss": 1.2046,
+      "step": 2830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522896484509247e-05,
+      "loss": 1.1592,
+      "step": 2831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522457153390725e-05,
+      "loss": 1.1362,
+      "step": 2832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952201762503852e-05,
+      "loss": 1.2158,
+      "step": 2833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9521577899461732e-05,
+      "loss": 1.0264,
+      "step": 2834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952113797666947e-05,
+      "loss": 1.1338,
+      "step": 2835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520697856670843e-05,
+      "loss": 1.127,
+      "step": 2836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520257539474975e-05,
+      "loss": 0.9722,
+      "step": 2837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9519817025090978e-05,
+      "loss": 1.1226,
+      "step": 2838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9519376313527977e-05,
+      "loss": 1.168,
+      "step": 2839
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9518935404795104e-05,
+      "loss": 1.0867,
+      "step": 2840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951849429890149e-05,
+      "loss": 1.1748,
+      "step": 2841
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951805299585627e-05,
+      "loss": 1.0708,
+      "step": 2842
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517611495668588e-05,
+      "loss": 1.1763,
+      "step": 2843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517169798347583e-05,
+      "loss": 1.1436,
+      "step": 2844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516727903902407e-05,
+      "loss": 1.1821,
+      "step": 2845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516285812342215e-05,
+      "loss": 1.1313,
+      "step": 2846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9515843523676163e-05,
+      "loss": 1.0444,
+      "step": 2847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9515401037913404e-05,
+      "loss": 1.2588,
+      "step": 2848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951495835506311e-05,
+      "loss": 1.1675,
+      "step": 2849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9514515475134453e-05,
+      "loss": 1.2295,
+      "step": 2850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.95140723981366e-05,
+      "loss": 1.1562,
+      "step": 2851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951362912407873e-05,
+      "loss": 1.2412,
+      "step": 2852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9513185652970025e-05,
+      "loss": 1.1606,
+      "step": 2853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951274198481967e-05,
+      "loss": 1.1602,
+      "step": 2854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9512298119636855e-05,
+      "loss": 1.2451,
+      "step": 2855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511854057430778e-05,
+      "loss": 1.165,
+      "step": 2856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511409798210624e-05,
+      "loss": 1.1003,
+      "step": 2857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951096534198561e-05,
+      "loss": 1.1392,
+      "step": 2858
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510520688764926e-05,
+      "loss": 1.1636,
+      "step": 2859
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510075838557796e-05,
+      "loss": 1.0742,
+      "step": 2860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509630791373425e-05,
+      "loss": 1.0283,
+      "step": 2861
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509185547221036e-05,
+      "loss": 1.1909,
+      "step": 2862
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9508740106109848e-05,
+      "loss": 1.1509,
+      "step": 2863
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9508294468049093e-05,
+      "loss": 1.166,
+      "step": 2864
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950784863304799e-05,
+      "loss": 1.2461,
+      "step": 2865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9507402601115787e-05,
+      "loss": 1.1099,
+      "step": 2866
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950695637226171e-05,
+      "loss": 1.1606,
+      "step": 2867
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950650994649501e-05,
+      "loss": 0.9937,
+      "step": 2868
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950606332382493e-05,
+      "loss": 0.9673,
+      "step": 2869
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950561650426072e-05,
+      "loss": 1.0913,
+      "step": 2870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950516948781164e-05,
+      "loss": 1.2168,
+      "step": 2871
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950472227448694e-05,
+      "loss": 1.1743,
+      "step": 2872
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950427486429589e-05,
+      "loss": 1.1084,
+      "step": 2873
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9503827257247753e-05,
+      "loss": 1.1152,
+      "step": 2874
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9503379453351803e-05,
+      "loss": 1.1074,
+      "step": 2875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502931452617314e-05,
+      "loss": 1.0312,
+      "step": 2876
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502483255053565e-05,
+      "loss": 1.0171,
+      "step": 2877
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950203486066984e-05,
+      "loss": 1.0757,
+      "step": 2878
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501586269475426e-05,
+      "loss": 1.1484,
+      "step": 2879
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501137481479612e-05,
+      "loss": 1.1685,
+      "step": 2880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.95006884966917e-05,
+      "loss": 1.1348,
+      "step": 2881
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950023931512098e-05,
+      "loss": 1.1353,
+      "step": 2882
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9499789936776765e-05,
+      "loss": 1.2646,
+      "step": 2883
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9499340361668355e-05,
+      "loss": 1.2295,
+      "step": 2884
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9498890589805067e-05,
+      "loss": 1.1333,
+      "step": 2885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949844062119622e-05,
+      "loss": 1.1582,
+      "step": 2886
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949799045585112e-05,
+      "loss": 1.249,
+      "step": 2887
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9497540093779107e-05,
+      "loss": 1.1182,
+      "step": 2888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9497089534989503e-05,
+      "loss": 1.2139,
+      "step": 2889
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9496638779491634e-05,
+      "loss": 1.2061,
+      "step": 2890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949618782729485e-05,
+      "loss": 1.0024,
+      "step": 2891
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495736678408476e-05,
+      "loss": 1.1851,
+      "step": 2892
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495285332841866e-05,
+      "loss": 1.0649,
+      "step": 2893
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949483379060437e-05,
+      "loss": 1.1919,
+      "step": 2894
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494382051705332e-05,
+      "loss": 1.2197,
+      "step": 2895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493930116154114e-05,
+      "loss": 1.2085,
+      "step": 2896
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493477983960075e-05,
+      "loss": 1.2407,
+      "step": 2897
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493025655132584e-05,
+      "loss": 1.1841,
+      "step": 2898
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949257312968101e-05,
+      "loss": 1.2129,
+      "step": 2899
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9492120407614715e-05,
+      "loss": 1.2227,
+      "step": 2900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9491667488943086e-05,
+      "loss": 1.0786,
+      "step": 2901
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9491214373675502e-05,
+      "loss": 1.3511,
+      "step": 2902
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949076106182135e-05,
+      "loss": 1.0806,
+      "step": 2903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9490307553390014e-05,
+      "loss": 1.2271,
+      "step": 2904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489853848390893e-05,
+      "loss": 1.1738,
+      "step": 2905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489399946833382e-05,
+      "loss": 1.3457,
+      "step": 2906
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948894584872688e-05,
+      "loss": 1.1323,
+      "step": 2907
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9488491554080795e-05,
+      "loss": 1.2529,
+      "step": 2908
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9488037062904538e-05,
+      "loss": 1.2236,
+      "step": 2909
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9487582375207515e-05,
+      "loss": 1.0742,
+      "step": 2910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9487127490999157e-05,
+      "loss": 1.1733,
+      "step": 2911
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486672410288873e-05,
+      "loss": 1.167,
+      "step": 2912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486217133086098e-05,
+      "loss": 1.2139,
+      "step": 2913
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9485761659400254e-05,
+      "loss": 1.0581,
+      "step": 2914
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9485305989240785e-05,
+      "loss": 1.1621,
+      "step": 2915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484850122617117e-05,
+      "loss": 0.855,
+      "step": 2916
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94843940595387e-05,
+      "loss": 1.1406,
+      "step": 2917
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948393780001498e-05,
+      "loss": 1.1621,
+      "step": 2918
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9483481344055404e-05,
+      "loss": 1.1196,
+      "step": 2919
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948302469166943e-05,
+      "loss": 1.2261,
+      "step": 2920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482567842866513e-05,
+      "loss": 1.126,
+      "step": 2921
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948211079765612e-05,
+      "loss": 1.1108,
+      "step": 2922
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948165355604771e-05,
+      "loss": 1.1372,
+      "step": 2923
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481196118050756e-05,
+      "loss": 1.0986,
+      "step": 2924
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948073848367474e-05,
+      "loss": 1.1099,
+      "step": 2925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9480280652929136e-05,
+      "loss": 1.0967,
+      "step": 2926
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9479822625823425e-05,
+      "loss": 1.1851,
+      "step": 2927
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9479364402367094e-05,
+      "loss": 1.0103,
+      "step": 2928
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478905982569635e-05,
+      "loss": 1.1865,
+      "step": 2929
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478447366440542e-05,
+      "loss": 1.0703,
+      "step": 2930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947798855398932e-05,
+      "loss": 1.1997,
+      "step": 2931
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9477529545225463e-05,
+      "loss": 1.1562,
+      "step": 2932
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9477070340158482e-05,
+      "loss": 1.0356,
+      "step": 2933
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947661093879789e-05,
+      "loss": 1.1689,
+      "step": 2934
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9476151341153203e-05,
+      "loss": 1.1299,
+      "step": 2935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947569154723394e-05,
+      "loss": 1.2002,
+      "step": 2936
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947523155704962e-05,
+      "loss": 1.02,
+      "step": 2937
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9474771370609772e-05,
+      "loss": 1.1382,
+      "step": 2938
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947431098792393e-05,
+      "loss": 1.1104,
+      "step": 2939
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9473850409001632e-05,
+      "loss": 1.1836,
+      "step": 2940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9473389633852412e-05,
+      "loss": 1.1689,
+      "step": 2941
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9472928662485818e-05,
+      "loss": 1.1699,
+      "step": 2942
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9472467494911393e-05,
+      "loss": 1.1108,
+      "step": 2943
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9472006131138697e-05,
+      "loss": 1.1968,
+      "step": 2944
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9471544571177276e-05,
+      "loss": 1.1011,
+      "step": 2945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94710828150367e-05,
+      "loss": 1.1519,
+      "step": 2946
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9470620862726527e-05,
+      "loss": 1.1157,
+      "step": 2947
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9470158714256322e-05,
+      "loss": 1.0635,
+      "step": 2948
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9469696369635668e-05,
+      "loss": 1.228,
+      "step": 2949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946923382887413e-05,
+      "loss": 1.1582,
+      "step": 2950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468771091981296e-05,
+      "loss": 1.168,
+      "step": 2951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468308158966746e-05,
+      "loss": 1.2607,
+      "step": 2952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9467845029840073e-05,
+      "loss": 1.2402,
+      "step": 2953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9467381704610866e-05,
+      "loss": 1.1655,
+      "step": 2954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946691818328872e-05,
+      "loss": 1.0354,
+      "step": 2955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946645446588324e-05,
+      "loss": 1.2051,
+      "step": 2956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946599055240403e-05,
+      "loss": 0.9565,
+      "step": 2957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9465526442860697e-05,
+      "loss": 1.0752,
+      "step": 2958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9465062137262855e-05,
+      "loss": 1.0024,
+      "step": 2959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946459763562012e-05,
+      "loss": 1.1899,
+      "step": 2960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9464132937942113e-05,
+      "loss": 1.1621,
+      "step": 2961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946366804423846e-05,
+      "loss": 1.2168,
+      "step": 2962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9463202954518788e-05,
+      "loss": 1.1333,
+      "step": 2963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9462737668792733e-05,
+      "loss": 1.1016,
+      "step": 2964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9462272187069932e-05,
+      "loss": 1.1436,
+      "step": 2965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9461806509360024e-05,
+      "loss": 1.1226,
+      "step": 2966
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9461340635672656e-05,
+      "loss": 0.9448,
+      "step": 2967
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460874566017476e-05,
+      "loss": 1.1846,
+      "step": 2968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460408300404137e-05,
+      "loss": 1.1167,
+      "step": 2969
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9459941838842303e-05,
+      "loss": 1.1367,
+      "step": 2970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9459475181341626e-05,
+      "loss": 1.2656,
+      "step": 2971
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945900832791178e-05,
+      "loss": 1.0864,
+      "step": 2972
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945854127856243e-05,
+      "loss": 1.1421,
+      "step": 2973
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458074033303247e-05,
+      "loss": 1.0977,
+      "step": 2974
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9457606592143916e-05,
+      "loss": 1.0674,
+      "step": 2975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9457138955094112e-05,
+      "loss": 1.125,
+      "step": 2976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9456671122163527e-05,
+      "loss": 1.1353,
+      "step": 2977
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945620309336185e-05,
+      "loss": 1.1406,
+      "step": 2978
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945573486869877e-05,
+      "loss": 1.1138,
+      "step": 2979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455266448183988e-05,
+      "loss": 1.1406,
+      "step": 2980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945479783182721e-05,
+      "loss": 1.2129,
+      "step": 2981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9454329019638134e-05,
+      "loss": 1.105,
+      "step": 2982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9453860011626477e-05,
+      "loss": 1.1772,
+      "step": 2983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945339080780195e-05,
+      "loss": 1.2373,
+      "step": 2984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452921408174273e-05,
+      "loss": 0.937,
+      "step": 2985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945245181275317e-05,
+      "loss": 1.1494,
+      "step": 2986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451982021548366e-05,
+      "loss": 1.1104,
+      "step": 2987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451512034569588e-05,
+      "loss": 1.3311,
+      "step": 2988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451041851826574e-05,
+      "loss": 1.1479,
+      "step": 2989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450571473329065e-05,
+      "loss": 1.1372,
+      "step": 2990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94501008990868e-05,
+      "loss": 1.29,
+      "step": 2991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449630129109525e-05,
+      "loss": 0.9888,
+      "step": 2992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449159163406995e-05,
+      "loss": 1.1157,
+      "step": 2993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944868800198896e-05,
+      "loss": 1.25,
+      "step": 2994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9448216644865182e-05,
+      "loss": 1.2095,
+      "step": 2995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9447745092045424e-05,
+      "loss": 1.0889,
+      "step": 2996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9447273343539455e-05,
+      "loss": 1.1313,
+      "step": 2997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9446801399357042e-05,
+      "loss": 1.1699,
+      "step": 2998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944632925950796e-05,
+      "loss": 1.0728,
+      "step": 2999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944585692400199e-05,
+      "loss": 1.0806,
+      "step": 3000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445384392848915e-05,
+      "loss": 1.0908,
+      "step": 3001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944491166605852e-05,
+      "loss": 1.1406,
+      "step": 3002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9444438743640603e-05,
+      "loss": 1.1978,
+      "step": 3003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9443965625604955e-05,
+      "loss": 1.2432,
+      "step": 3004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9443492311961373e-05,
+      "loss": 1.0894,
+      "step": 3005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9443018802719666e-05,
+      "loss": 1.0054,
+      "step": 3006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442545097889635e-05,
+      "loss": 1.1758,
+      "step": 3007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442071197481095e-05,
+      "loss": 1.0586,
+      "step": 3008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9441597101503863e-05,
+      "loss": 1.1777,
+      "step": 3009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9441122809967757e-05,
+      "loss": 1.2178,
+      "step": 3010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440648322882603e-05,
+      "loss": 1.1069,
+      "step": 3011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440173640258226e-05,
+      "loss": 1.1445,
+      "step": 3012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943969876210446e-05,
+      "loss": 1.1865,
+      "step": 3013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943922368843114e-05,
+      "loss": 1.0784,
+      "step": 3014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9438748419248104e-05,
+      "loss": 1.1313,
+      "step": 3015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94382729545652e-05,
+      "loss": 1.0918,
+      "step": 3016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943779729439227e-05,
+      "loss": 1.1309,
+      "step": 3017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437321438739172e-05,
+      "loss": 1.1157,
+      "step": 3018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943684538761576e-05,
+      "loss": 1.1514,
+      "step": 3019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9436369141031895e-05,
+      "loss": 1.1177,
+      "step": 3020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943589269899744e-05,
+      "loss": 1.2026,
+      "step": 3021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435416061522267e-05,
+      "loss": 1.1104,
+      "step": 3022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434939228616238e-05,
+      "loss": 1.0327,
+      "step": 3023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434462200289242e-05,
+      "loss": 1.2124,
+      "step": 3024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433984976551153e-05,
+      "loss": 1.2012,
+      "step": 3025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433507557411856e-05,
+      "loss": 1.1577,
+      "step": 3026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433029942881242e-05,
+      "loss": 1.2314,
+      "step": 3027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9432552132969197e-05,
+      "loss": 1.1719,
+      "step": 3028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9432074127685625e-05,
+      "loss": 1.2891,
+      "step": 3029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431595927040425e-05,
+      "loss": 1.1357,
+      "step": 3030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94311175310435e-05,
+      "loss": 1.0957,
+      "step": 3031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9430638939704756e-05,
+      "loss": 1.1514,
+      "step": 3032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9430160153034113e-05,
+      "loss": 1.2227,
+      "step": 3033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9429681171041484e-05,
+      "loss": 1.209,
+      "step": 3034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9429201993736787e-05,
+      "loss": 1.0493,
+      "step": 3035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428722621129954e-05,
+      "loss": 1.1523,
+      "step": 3036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428243053230907e-05,
+      "loss": 1.2959,
+      "step": 3037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9427763290049583e-05,
+      "loss": 1.104,
+      "step": 3038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9427283331595916e-05,
+      "loss": 1.1558,
+      "step": 3039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426803177879853e-05,
+      "loss": 1.1582,
+      "step": 3040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942632282891133e-05,
+      "loss": 1.168,
+      "step": 3041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9425842284700307e-05,
+      "loss": 1.208,
+      "step": 3042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942536154525673e-05,
+      "loss": 1.1147,
+      "step": 3043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9424880610590557e-05,
+      "loss": 1.0376,
+      "step": 3044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942439948071175e-05,
+      "loss": 1.1033,
+      "step": 3045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423918155630278e-05,
+      "loss": 1.1646,
+      "step": 3046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423436635356103e-05,
+      "loss": 1.0938,
+      "step": 3047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9422954919899205e-05,
+      "loss": 1.0483,
+      "step": 3048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942247300926956e-05,
+      "loss": 1.1221,
+      "step": 3049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421990903477147e-05,
+      "loss": 1.0962,
+      "step": 3050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421508602531956e-05,
+      "loss": 1.1729,
+      "step": 3051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942102610644397e-05,
+      "loss": 1.209,
+      "step": 3052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9420543415223193e-05,
+      "loss": 1.2051,
+      "step": 3053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9420060528879608e-05,
+      "loss": 1.1587,
+      "step": 3054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9419577447423233e-05,
+      "loss": 1.1367,
+      "step": 3055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9419094170864064e-05,
+      "loss": 1.0664,
+      "step": 3056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941861069921211e-05,
+      "loss": 1.1533,
+      "step": 3057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941812703247739e-05,
+      "loss": 1.1011,
+      "step": 3058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941764317066992e-05,
+      "loss": 1.1084,
+      "step": 3059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9417159113799722e-05,
+      "loss": 1.1812,
+      "step": 3060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9416674861876822e-05,
+      "loss": 1.1328,
+      "step": 3061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941619041491125e-05,
+      "loss": 1.2373,
+      "step": 3062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941570577291304e-05,
+      "loss": 1.0435,
+      "step": 3063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941522093589223e-05,
+      "loss": 1.042,
+      "step": 3064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941473590385886e-05,
+      "loss": 1.0464,
+      "step": 3065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414250676822982e-05,
+      "loss": 1.2305,
+      "step": 3066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941376525479464e-05,
+      "loss": 1.0024,
+      "step": 3067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941327963778389e-05,
+      "loss": 1.1958,
+      "step": 3068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9412793825800795e-05,
+      "loss": 1.1562,
+      "step": 3069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941230781885541e-05,
+      "loss": 1.2148,
+      "step": 3070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941182161695781e-05,
+      "loss": 1.166,
+      "step": 3071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9411335220118056e-05,
+      "loss": 1.1489,
+      "step": 3072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941084862834623e-05,
+      "loss": 1.0444,
+      "step": 3073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9410361841652404e-05,
+      "loss": 1.0972,
+      "step": 3074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9409874860046667e-05,
+      "loss": 1.1587,
+      "step": 3075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9409387683539104e-05,
+      "loss": 1.3623,
+      "step": 3076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.94089003121398e-05,
+      "loss": 1.0259,
+      "step": 3077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9408412745858858e-05,
+      "loss": 1.0376,
+      "step": 3078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940792498470637e-05,
+      "loss": 1.125,
+      "step": 3079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407437028692443e-05,
+      "loss": 1.1846,
+      "step": 3080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9406948877827183e-05,
+      "loss": 1.0923,
+      "step": 3081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.94064605321207e-05,
+      "loss": 1.2031,
+      "step": 3082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940597199158311e-05,
+      "loss": 1.0442,
+      "step": 3083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9405483256224525e-05,
+      "loss": 1.2695,
+      "step": 3084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940499432605508e-05,
+      "loss": 1.1006,
+      "step": 3085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940450520108489e-05,
+      "loss": 1.1357,
+      "step": 3086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9404015881324097e-05,
+      "loss": 1.2266,
+      "step": 3087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940352636678283e-05,
+      "loss": 1.2393,
+      "step": 3088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9403036657471225e-05,
+      "loss": 1.147,
+      "step": 3089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940254675339943e-05,
+      "loss": 1.1685,
+      "step": 3090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9402056654577593e-05,
+      "loss": 1.1035,
+      "step": 3091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940156636101586e-05,
+      "loss": 1.1436,
+      "step": 3092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9401075872724396e-05,
+      "loss": 1.062,
+      "step": 3093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940058518971335e-05,
+      "loss": 1.1948,
+      "step": 3094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9400094311992888e-05,
+      "loss": 1.0908,
+      "step": 3095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9399603239573177e-05,
+      "loss": 1.1294,
+      "step": 3096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939911197246439e-05,
+      "loss": 1.1309,
+      "step": 3097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398620510676705e-05,
+      "loss": 1.1392,
+      "step": 3098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398128854220296e-05,
+      "loss": 1.1782,
+      "step": 3099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397637003105348e-05,
+      "loss": 1.0957,
+      "step": 3100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939714495734205e-05,
+      "loss": 1.2188,
+      "step": 3101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939665271694059e-05,
+      "loss": 1.2617,
+      "step": 3102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939616028191117e-05,
+      "loss": 1.019,
+      "step": 3103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939566765226398e-05,
+      "loss": 1.1724,
+      "step": 3104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9395174828009233e-05,
+      "loss": 0.9546,
+      "step": 3105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9394681809157134e-05,
+      "loss": 1.2236,
+      "step": 3106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9394188595717894e-05,
+      "loss": 1.2446,
+      "step": 3107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393695187701725e-05,
+      "loss": 1.2036,
+      "step": 3108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939320158511885e-05,
+      "loss": 1.0864,
+      "step": 3109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9392707787979495e-05,
+      "loss": 1.2412,
+      "step": 3110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9392213796293884e-05,
+      "loss": 1.2695,
+      "step": 3111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939171961007225e-05,
+      "loss": 1.1455,
+      "step": 3112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939122522932483e-05,
+      "loss": 1.041,
+      "step": 3113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9390730654061863e-05,
+      "loss": 1.3232,
+      "step": 3114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939023588429359e-05,
+      "loss": 1.2617,
+      "step": 3115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389740920030265e-05,
+      "loss": 1.0684,
+      "step": 3116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389245761282134e-05,
+      "loss": 1.1328,
+      "step": 3117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9388750408059457e-05,
+      "loss": 1.1846,
+      "step": 3118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938825486037249e-05,
+      "loss": 1.104,
+      "step": 3119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.93877591182315e-05,
+      "loss": 1.0889,
+      "step": 3120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938726318164676e-05,
+      "loss": 1.1138,
+      "step": 3121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938676705062853e-05,
+      "loss": 1.1826,
+      "step": 3122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386270725187094e-05,
+      "loss": 1.0425,
+      "step": 3123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9385774205332732e-05,
+      "loss": 1.1509,
+      "step": 3124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938527749107573e-05,
+      "loss": 1.1621,
+      "step": 3125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938478058242637e-05,
+      "loss": 1.0278,
+      "step": 3126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9384283479394946e-05,
+      "loss": 1.105,
+      "step": 3127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9383786181991758e-05,
+      "loss": 1.2656,
+      "step": 3128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9383288690227103e-05,
+      "loss": 1.1729,
+      "step": 3129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9382791004111287e-05,
+      "loss": 1.1763,
+      "step": 3130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938229312365462e-05,
+      "loss": 1.0122,
+      "step": 3131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381795048867405e-05,
+      "loss": 1.1152,
+      "step": 3132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381296779759967e-05,
+      "loss": 1.0923,
+      "step": 3133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9380798316342632e-05,
+      "loss": 1.2158,
+      "step": 3134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938029965862571e-05,
+      "loss": 1.1167,
+      "step": 3135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937980080661954e-05,
+      "loss": 1.0044,
+      "step": 3136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9379301760334445e-05,
+      "loss": 1.106,
+      "step": 3137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9378802519780772e-05,
+      "loss": 1.0371,
+      "step": 3138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937830308496886e-05,
+      "loss": 0.9199,
+      "step": 3139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9377803455909044e-05,
+      "loss": 1.2334,
+      "step": 3140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9377303632611683e-05,
+      "loss": 1.1484,
+      "step": 3141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9376803615087122e-05,
+      "loss": 1.0659,
+      "step": 3142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9376303403345723e-05,
+      "loss": 1.2417,
+      "step": 3143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375802997397847e-05,
+      "loss": 1.1104,
+      "step": 3144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375302397253854e-05,
+      "loss": 1.1191,
+      "step": 3145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9374801602924116e-05,
+      "loss": 1.1948,
+      "step": 3146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9374300614419006e-05,
+      "loss": 1.1821,
+      "step": 3147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9373799431748897e-05,
+      "loss": 1.2808,
+      "step": 3148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9373298054924175e-05,
+      "loss": 1.063,
+      "step": 3149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937279648395522e-05,
+      "loss": 1.1079,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937229471885242e-05,
+      "loss": 1.0986,
+      "step": 3151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9371792759626175e-05,
+      "loss": 1.1748,
+      "step": 3152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9371290606286876e-05,
+      "loss": 1.2422,
+      "step": 3153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9370788258844925e-05,
+      "loss": 1.209,
+      "step": 3154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9370285717310728e-05,
+      "loss": 1.209,
+      "step": 3155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9369782981694692e-05,
+      "loss": 1.1016,
+      "step": 3156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9369280052007233e-05,
+      "loss": 1.1758,
+      "step": 3157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9368776928258763e-05,
+      "loss": 1.0508,
+      "step": 3158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9368273610459706e-05,
+      "loss": 1.0913,
+      "step": 3159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936777009862049e-05,
+      "loss": 1.1777,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936726639275154e-05,
+      "loss": 1.1924,
+      "step": 3161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9366762492863286e-05,
+      "loss": 1.1978,
+      "step": 3162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9366258398966172e-05,
+      "loss": 1.1167,
+      "step": 3163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365754111070635e-05,
+      "loss": 1.127,
+      "step": 3164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365249629187118e-05,
+      "loss": 0.9932,
+      "step": 3165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9364744953326077e-05,
+      "loss": 1.1855,
+      "step": 3166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936424008349796e-05,
+      "loss": 1.0981,
+      "step": 3167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9363735019713223e-05,
+      "loss": 1.1382,
+      "step": 3168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936322976198233e-05,
+      "loss": 1.1313,
+      "step": 3169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9362724310315744e-05,
+      "loss": 1.2178,
+      "step": 3170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9362218664723937e-05,
+      "loss": 1.2007,
+      "step": 3171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9361712825217384e-05,
+      "loss": 1.1611,
+      "step": 3172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936120679180655e-05,
+      "loss": 1.0234,
+      "step": 3173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9360700564501936e-05,
+      "loss": 1.2236,
+      "step": 3174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.936019414331401e-05,
+      "loss": 1.1006,
+      "step": 3175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9359687528253265e-05,
+      "loss": 1.1782,
+      "step": 3176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9359180719330202e-05,
+      "loss": 1.1831,
+      "step": 3177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9358673716555312e-05,
+      "loss": 1.1187,
+      "step": 3178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9358166519939097e-05,
+      "loss": 1.2837,
+      "step": 3179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935765912949206e-05,
+      "loss": 1.1284,
+      "step": 3180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9357151545224714e-05,
+      "loss": 0.9585,
+      "step": 3181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9356643767147574e-05,
+      "loss": 1.1094,
+      "step": 3182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9356135795271155e-05,
+      "loss": 1.1006,
+      "step": 3183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935562762960597e-05,
+      "loss": 1.29,
+      "step": 3184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935511927016256e-05,
+      "loss": 1.2324,
+      "step": 3185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9354610716951448e-05,
+      "loss": 1.0073,
+      "step": 3186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9354101969983162e-05,
+      "loss": 1.0635,
+      "step": 3187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9353593029268242e-05,
+      "loss": 1.2329,
+      "step": 3188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9353083894817237e-05,
+      "loss": 1.2549,
+      "step": 3189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935257456664068e-05,
+      "loss": 1.123,
+      "step": 3190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9352065044749128e-05,
+      "loss": 1.1406,
+      "step": 3191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351555329153137e-05,
+      "loss": 1.1538,
+      "step": 3192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351045419863256e-05,
+      "loss": 1.0352,
+      "step": 3193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9350535316890055e-05,
+      "loss": 1.1797,
+      "step": 3194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9350025020244095e-05,
+      "loss": 1.165,
+      "step": 3195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9349514529935945e-05,
+      "loss": 0.9565,
+      "step": 3196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934900384597618e-05,
+      "loss": 1.1475,
+      "step": 3197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9348492968375375e-05,
+      "loss": 1.0737,
+      "step": 3198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9347981897144118e-05,
+      "loss": 1.2021,
+      "step": 3199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934747063229299e-05,
+      "loss": 1.0781,
+      "step": 3200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934695917383258e-05,
+      "loss": 0.906,
+      "step": 3201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934644752177348e-05,
+      "loss": 1.0513,
+      "step": 3202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9345935676126296e-05,
+      "loss": 1.2339,
+      "step": 3203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9345423636901616e-05,
+      "loss": 1.0503,
+      "step": 3204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934491140411006e-05,
+      "loss": 1.124,
+      "step": 3205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9344398977762225e-05,
+      "loss": 1.1196,
+      "step": 3206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9343886357868738e-05,
+      "loss": 1.1826,
+      "step": 3207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9343373544440203e-05,
+      "loss": 1.1235,
+      "step": 3208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934286053748725e-05,
+      "loss": 1.1396,
+      "step": 3209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9342347337020503e-05,
+      "loss": 1.1089,
+      "step": 3210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9341833943050588e-05,
+      "loss": 1.1963,
+      "step": 3211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9341320355588143e-05,
+      "loss": 1.106,
+      "step": 3212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9340806574643807e-05,
+      "loss": 1.1426,
+      "step": 3213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9340292600228217e-05,
+      "loss": 1.1655,
+      "step": 3214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.933977843235202e-05,
+      "loss": 1.085,
+      "step": 3215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.933926407102587e-05,
+      "loss": 1.1294,
+      "step": 3216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9338749516260414e-05,
+      "loss": 1.209,
+      "step": 3217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9338234768066316e-05,
+      "loss": 1.0815,
+      "step": 3218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9337719826454233e-05,
+      "loss": 1.0674,
+      "step": 3219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933720469143483e-05,
+      "loss": 1.1016,
+      "step": 3220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933668936301878e-05,
+      "loss": 1.1616,
+      "step": 3221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9336173841216754e-05,
+      "loss": 0.8413,
+      "step": 3222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9335658126039435e-05,
+      "loss": 1.084,
+      "step": 3223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.93351422174975e-05,
+      "loss": 0.9668,
+      "step": 3224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9334626115601634e-05,
+      "loss": 1.1787,
+      "step": 3225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9334109820362532e-05,
+      "loss": 1.1416,
+      "step": 3226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9333593331790884e-05,
+      "loss": 1.2271,
+      "step": 3227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9333076649897386e-05,
+      "loss": 1.1338,
+      "step": 3228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9332559774692745e-05,
+      "loss": 1.1841,
+      "step": 3229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933204270618766e-05,
+      "loss": 1.0942,
+      "step": 3230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9331525444392845e-05,
+      "loss": 0.9219,
+      "step": 3231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9331007989319016e-05,
+      "loss": 1.1753,
+      "step": 3232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9330490340976887e-05,
+      "loss": 1.209,
+      "step": 3233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932997249937718e-05,
+      "loss": 1.1665,
+      "step": 3234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329454464530625e-05,
+      "loss": 1.0249,
+      "step": 3235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328936236447945e-05,
+      "loss": 1.0386,
+      "step": 3236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328417815139876e-05,
+      "loss": 1.0977,
+      "step": 3237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932789920061716e-05,
+      "loss": 1.1519,
+      "step": 3238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9327380392890536e-05,
+      "loss": 1.2188,
+      "step": 3239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9326861391970747e-05,
+      "loss": 1.1865,
+      "step": 3240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932634219786855e-05,
+      "loss": 1.2119,
+      "step": 3241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932582281059469e-05,
+      "loss": 1.0762,
+      "step": 3242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9325303230159927e-05,
+      "loss": 1.0605,
+      "step": 3243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932478345657503e-05,
+      "loss": 1.2695,
+      "step": 3244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324263489850757e-05,
+      "loss": 1.0327,
+      "step": 3245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932374332999788e-05,
+      "loss": 1.2041,
+      "step": 3246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9323222977027173e-05,
+      "loss": 1.0034,
+      "step": 3247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9322702430949417e-05,
+      "loss": 1.0942,
+      "step": 3248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9322181691775387e-05,
+      "loss": 1.2842,
+      "step": 3249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321660759515876e-05,
+      "loss": 1.186,
+      "step": 3250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321139634181666e-05,
+      "loss": 1.0278,
+      "step": 3251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9320618315783558e-05,
+      "loss": 1.0835,
+      "step": 3252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9320096804332347e-05,
+      "loss": 1.2456,
+      "step": 3253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9319575099838834e-05,
+      "loss": 1.146,
+      "step": 3254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9319053202313827e-05,
+      "loss": 1.165,
+      "step": 3255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931853111176813e-05,
+      "loss": 1.2017,
+      "step": 3256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9318008828212565e-05,
+      "loss": 1.1411,
+      "step": 3257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317486351657947e-05,
+      "loss": 1.1089,
+      "step": 3258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9316963682115094e-05,
+      "loss": 1.001,
+      "step": 3259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9316440819594836e-05,
+      "loss": 1.229,
+      "step": 3260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315917764108e-05,
+      "loss": 1.0938,
+      "step": 3261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315394515665423e-05,
+      "loss": 1.1367,
+      "step": 3262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931487107427794e-05,
+      "loss": 1.0859,
+      "step": 3263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9314347439956395e-05,
+      "loss": 1.2871,
+      "step": 3264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9313823612711627e-05,
+      "loss": 1.165,
+      "step": 3265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9313299592554496e-05,
+      "loss": 1.1345,
+      "step": 3266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9312775379495847e-05,
+      "loss": 1.2441,
+      "step": 3267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931225097354655e-05,
+      "loss": 1.2119,
+      "step": 3268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9311726374717447e-05,
+      "loss": 1.1895,
+      "step": 3269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931120158301942e-05,
+      "loss": 1.1748,
+      "step": 3270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9310676598463335e-05,
+      "loss": 1.209,
+      "step": 3271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9310151421060067e-05,
+      "loss": 1.0044,
+      "step": 3272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309626050820486e-05,
+      "loss": 1.2231,
+      "step": 3273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930910048775548e-05,
+      "loss": 1.2188,
+      "step": 3274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9308574731875934e-05,
+      "loss": 1.1484,
+      "step": 3275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930804878319274e-05,
+      "loss": 1.1138,
+      "step": 3276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9307522641716787e-05,
+      "loss": 1.1133,
+      "step": 3277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9306996307458976e-05,
+      "loss": 1.083,
+      "step": 3278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9306469780430205e-05,
+      "loss": 1.0566,
+      "step": 3279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9305943060641384e-05,
+      "loss": 1.1885,
+      "step": 3280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9305416148103423e-05,
+      "loss": 1.1533,
+      "step": 3281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930488904282723e-05,
+      "loss": 1.0591,
+      "step": 3282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9304361744823725e-05,
+      "loss": 1.1553,
+      "step": 3283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9303834254103836e-05,
+      "loss": 1.0718,
+      "step": 3284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930330657067848e-05,
+      "loss": 1.1108,
+      "step": 3285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930277869455859e-05,
+      "loss": 1.0264,
+      "step": 3286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.93022506257551e-05,
+      "loss": 1.0576,
+      "step": 3287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930172236427895e-05,
+      "loss": 1.1313,
+      "step": 3288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301193910141078e-05,
+      "loss": 1.165,
+      "step": 3289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300665263352427e-05,
+      "loss": 1.2588,
+      "step": 3290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930013642392395e-05,
+      "loss": 1.0151,
+      "step": 3291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9299607391866606e-05,
+      "loss": 1.126,
+      "step": 3292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9299078167191344e-05,
+      "loss": 1.0002,
+      "step": 3293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929854874990913e-05,
+      "loss": 1.1201,
+      "step": 3294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9298019140030927e-05,
+      "loss": 1.0981,
+      "step": 3295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9297489337567708e-05,
+      "loss": 1.1309,
+      "step": 3296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929695934253044e-05,
+      "loss": 1.0498,
+      "step": 3297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929642915493011e-05,
+      "loss": 0.9766,
+      "step": 3298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9295898774777693e-05,
+      "loss": 1.2559,
+      "step": 3299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9295368202084176e-05,
+      "loss": 1.2402,
+      "step": 3300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9294837436860544e-05,
+      "loss": 1.2446,
+      "step": 3301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.92943064791178e-05,
+      "loss": 1.1621,
+      "step": 3302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929377532886693e-05,
+      "loss": 1.1748,
+      "step": 3303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929324398611895e-05,
+      "loss": 1.0757,
+      "step": 3304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9292712450884856e-05,
+      "loss": 1.0771,
+      "step": 3305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9292180723175656e-05,
+      "loss": 0.9519,
+      "step": 3306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9291648803002367e-05,
+      "loss": 0.9922,
+      "step": 3307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9291116690376002e-05,
+      "loss": 1.3242,
+      "step": 3308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.929058438530759e-05,
+      "loss": 1.1318,
+      "step": 3309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9290051887808147e-05,
+      "loss": 1.0952,
+      "step": 3310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928951919788871e-05,
+      "loss": 1.1836,
+      "step": 3311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928898631556031e-05,
+      "loss": 1.0869,
+      "step": 3312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9288453240833986e-05,
+      "loss": 1.251,
+      "step": 3313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9287919973720776e-05,
+      "loss": 1.0952,
+      "step": 3314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9287386514231726e-05,
+      "loss": 1.1636,
+      "step": 3315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9286852862377886e-05,
+      "loss": 1.0713,
+      "step": 3316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928631901817031e-05,
+      "loss": 1.2231,
+      "step": 3317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9285784981620056e-05,
+      "loss": 1.0659,
+      "step": 3318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928525075273818e-05,
+      "loss": 1.0796,
+      "step": 3319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284716331535752e-05,
+      "loss": 0.9756,
+      "step": 3320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928418171802384e-05,
+      "loss": 1.0117,
+      "step": 3321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928364691221352e-05,
+      "loss": 1.1113,
+      "step": 3322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9283111914115865e-05,
+      "loss": 1.166,
+      "step": 3323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9282576723741954e-05,
+      "loss": 1.0103,
+      "step": 3324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928204134110288e-05,
+      "loss": 1.1426,
+      "step": 3325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9281505766209726e-05,
+      "loss": 1.1436,
+      "step": 3326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9280969999073585e-05,
+      "loss": 1.0459,
+      "step": 3327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928043403970556e-05,
+      "loss": 1.1621,
+      "step": 3328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9279897888116742e-05,
+      "loss": 1.0728,
+      "step": 3329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9279361544318246e-05,
+      "loss": 1.23,
+      "step": 3330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9278825008321175e-05,
+      "loss": 1.105,
+      "step": 3331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9278288280136647e-05,
+      "loss": 1.0557,
+      "step": 3332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9277751359775773e-05,
+      "loss": 1.041,
+      "step": 3333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9277214247249677e-05,
+      "loss": 1.2236,
+      "step": 3334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9276676942569484e-05,
+      "loss": 1.2109,
+      "step": 3335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9276139445746325e-05,
+      "loss": 1.2446,
+      "step": 3336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927560175679133e-05,
+      "loss": 1.207,
+      "step": 3337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275063875715638e-05,
+      "loss": 1.106,
+      "step": 3338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9274525802530383e-05,
+      "loss": 1.1958,
+      "step": 3339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927398753724672e-05,
+      "loss": 1.1191,
+      "step": 3340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9273449079875788e-05,
+      "loss": 1.0679,
+      "step": 3341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272910430428747e-05,
+      "loss": 1.1416,
+      "step": 3342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272371588916752e-05,
+      "loss": 1.1865,
+      "step": 3343
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9271832555350963e-05,
+      "loss": 1.2163,
+      "step": 3344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9271293329742547e-05,
+      "loss": 1.1709,
+      "step": 3345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927075391210267e-05,
+      "loss": 1.27,
+      "step": 3346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.92702143024425e-05,
+      "loss": 1.1304,
+      "step": 3347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9269674500773226e-05,
+      "loss": 1.103,
+      "step": 3348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9269134507106016e-05,
+      "loss": 1.0908,
+      "step": 3349
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9268594321452064e-05,
+      "loss": 1.1191,
+      "step": 3350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9268053943822553e-05,
+      "loss": 1.0732,
+      "step": 3351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.926751337422868e-05,
+      "loss": 1.0542,
+      "step": 3352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9266972612681635e-05,
+      "loss": 1.0859,
+      "step": 3353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9266431659192624e-05,
+      "loss": 1.186,
+      "step": 3354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9265890513772853e-05,
+      "loss": 1.2368,
+      "step": 3355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9265349176433525e-05,
+      "loss": 1.1558,
+      "step": 3356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264807647185852e-05,
+      "loss": 1.2529,
+      "step": 3357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264265926041055e-05,
+      "loss": 1.1538,
+      "step": 3358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9263724013010354e-05,
+      "loss": 1.1333,
+      "step": 3359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926318190810497e-05,
+      "loss": 1.0769,
+      "step": 3360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262639611336137e-05,
+      "loss": 1.1406,
+      "step": 3361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262097122715082e-05,
+      "loss": 1.0122,
+      "step": 3362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9261554442253042e-05,
+      "loss": 1.2129,
+      "step": 3363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9261011569961258e-05,
+      "loss": 1.1362,
+      "step": 3364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9260468505850972e-05,
+      "loss": 1.0903,
+      "step": 3365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925992524993344e-05,
+      "loss": 1.1411,
+      "step": 3366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9259381802219902e-05,
+      "loss": 1.1875,
+      "step": 3367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9258838162721626e-05,
+      "loss": 1.1577,
+      "step": 3368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9258294331449866e-05,
+      "loss": 1.1353,
+      "step": 3369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9257750308415887e-05,
+      "loss": 1.1558,
+      "step": 3370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9257206093630957e-05,
+      "loss": 0.9788,
+      "step": 3371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9256661687106352e-05,
+      "loss": 1.0955,
+      "step": 3372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9256117088853336e-05,
+      "loss": 1.0464,
+      "step": 3373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255572298883205e-05,
+      "loss": 1.0229,
+      "step": 3374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925502731720723e-05,
+      "loss": 1.0811,
+      "step": 3375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925448214383671e-05,
+      "loss": 1.0703,
+      "step": 3376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9253936778782926e-05,
+      "loss": 1.0664,
+      "step": 3377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9253391222057178e-05,
+      "loss": 1.1846,
+      "step": 3378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9252845473670768e-05,
+      "loss": 1.1328,
+      "step": 3379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9252299533634998e-05,
+      "loss": 1.021,
+      "step": 3380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9251753401961178e-05,
+      "loss": 1.1592,
+      "step": 3381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9251207078660618e-05,
+      "loss": 1.189,
+      "step": 3382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925066056374463e-05,
+      "loss": 1.2524,
+      "step": 3383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9250113857224542e-05,
+      "loss": 1.1167,
+      "step": 3384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924956695911167e-05,
+      "loss": 1.1963,
+      "step": 3385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9249019869417348e-05,
+      "loss": 1.1118,
+      "step": 3386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.92484725881529e-05,
+      "loss": 1.083,
+      "step": 3387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924792511532967e-05,
+      "loss": 1.2676,
+      "step": 3388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9247377450958988e-05,
+      "loss": 1.2109,
+      "step": 3389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924682959505221e-05,
+      "loss": 1.083,
+      "step": 3390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9246281547620672e-05,
+      "loss": 1.0894,
+      "step": 3391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9245733308675728e-05,
+      "loss": 1.0938,
+      "step": 3392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9245184878228737e-05,
+      "loss": 1.0947,
+      "step": 3393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9244636256291052e-05,
+      "loss": 1.124,
+      "step": 3394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9244087442874048e-05,
+      "loss": 1.166,
+      "step": 3395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9243538437989082e-05,
+      "loss": 1.2026,
+      "step": 3396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9242989241647528e-05,
+      "loss": 1.2246,
+      "step": 3397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9242439853860762e-05,
+      "loss": 1.3301,
+      "step": 3398
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9241890274640163e-05,
+      "loss": 1.1509,
+      "step": 3399
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9241340503997114e-05,
+      "loss": 1.1777,
+      "step": 3400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9240790541942996e-05,
+      "loss": 1.2178,
+      "step": 3401
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9240240388489214e-05,
+      "loss": 0.9902,
+      "step": 3402
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923969004364715e-05,
+      "loss": 1.2441,
+      "step": 3403
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923913950742821e-05,
+      "loss": 1.2266,
+      "step": 3404
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9238588779843796e-05,
+      "loss": 1.1807,
+      "step": 3405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9238037860905313e-05,
+      "loss": 1.0703,
+      "step": 3406
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9237486750624174e-05,
+      "loss": 0.9614,
+      "step": 3407
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923693544901179e-05,
+      "loss": 1.0327,
+      "step": 3408
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9236383956079584e-05,
+      "loss": 0.9868,
+      "step": 3409
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9235832271838976e-05,
+      "loss": 1.1616,
+      "step": 3410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9235280396301394e-05,
+      "loss": 1.1401,
+      "step": 3411
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923472832947827e-05,
+      "loss": 1.2246,
+      "step": 3412
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923417607138104e-05,
+      "loss": 1.1206,
+      "step": 3413
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9233623622021136e-05,
+      "loss": 1.1162,
+      "step": 3414
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9233070981410007e-05,
+      "loss": 1.1353,
+      "step": 3415
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9232518149559095e-05,
+      "loss": 1.083,
+      "step": 3416
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9231965126479853e-05,
+      "loss": 1.0864,
+      "step": 3417
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9231411912183737e-05,
+      "loss": 1.0771,
+      "step": 3418
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9230858506682205e-05,
+      "loss": 1.1709,
+      "step": 3419
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9230304909986715e-05,
+      "loss": 1.1416,
+      "step": 3420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9229751122108737e-05,
+      "loss": 1.106,
+      "step": 3421
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922919714305974e-05,
+      "loss": 1.1543,
+      "step": 3422
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92286429728512e-05,
+      "loss": 1.1519,
+      "step": 3423
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9228088611494593e-05,
+      "loss": 1.0723,
+      "step": 3424
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9227534059001406e-05,
+      "loss": 1.1152,
+      "step": 3425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922697931538312e-05,
+      "loss": 1.1514,
+      "step": 3426
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9226424380651225e-05,
+      "loss": 1.0996,
+      "step": 3427
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9225869254817215e-05,
+      "loss": 1.1426,
+      "step": 3428
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9225313937892592e-05,
+      "loss": 1.1138,
+      "step": 3429
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922475842988886e-05,
+      "loss": 1.2061,
+      "step": 3430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9224202730817513e-05,
+      "loss": 1.145,
+      "step": 3431
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9223646840690072e-05,
+      "loss": 1.0737,
+      "step": 3432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922309075951805e-05,
+      "loss": 1.1973,
+      "step": 3433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9222534487312963e-05,
+      "loss": 1.0879,
+      "step": 3434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9221978024086327e-05,
+      "loss": 1.1763,
+      "step": 3435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9221421369849675e-05,
+      "loss": 1.085,
+      "step": 3436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220864524614534e-05,
+      "loss": 1.1655,
+      "step": 3437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922030748839244e-05,
+      "loss": 1.189,
+      "step": 3438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921975026119493e-05,
+      "loss": 1.0576,
+      "step": 3439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9219192843033544e-05,
+      "loss": 1.0864,
+      "step": 3440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9218635233919826e-05,
+      "loss": 1.1294,
+      "step": 3441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9218077433865324e-05,
+      "loss": 1.125,
+      "step": 3442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9217519442881597e-05,
+      "loss": 1.1362,
+      "step": 3443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9216961260980205e-05,
+      "loss": 1.1855,
+      "step": 3444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9216402888172697e-05,
+      "loss": 1.1094,
+      "step": 3445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921584432447065e-05,
+      "loss": 1.1016,
+      "step": 3446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9215285569885628e-05,
+      "loss": 1.2861,
+      "step": 3447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9214726624429208e-05,
+      "loss": 1.0444,
+      "step": 3448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921416748811296e-05,
+      "loss": 0.9773,
+      "step": 3449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9213608160948475e-05,
+      "loss": 1.1172,
+      "step": 3450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9213048642947326e-05,
+      "loss": 1.2207,
+      "step": 3451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921248893412111e-05,
+      "loss": 1.1816,
+      "step": 3452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921192903448142e-05,
+      "loss": 1.1621,
+      "step": 3453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9211368944039852e-05,
+      "loss": 1.2114,
+      "step": 3454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210808662808005e-05,
+      "loss": 0.9932,
+      "step": 3455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210248190797484e-05,
+      "loss": 1.21,
+      "step": 3456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.92096875280199e-05,
+      "loss": 1.1206,
+      "step": 3457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9209126674486864e-05,
+      "loss": 1.0527,
+      "step": 3458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9208565630209994e-05,
+      "loss": 1.1333,
+      "step": 3459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9208004395200908e-05,
+      "loss": 1.1475,
+      "step": 3460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9207442969471235e-05,
+      "loss": 1.1011,
+      "step": 3461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.92068813530326e-05,
+      "loss": 1.0781,
+      "step": 3462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9206319545896636e-05,
+      "loss": 1.1567,
+      "step": 3463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920575754807498e-05,
+      "loss": 1.085,
+      "step": 3464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920519535957927e-05,
+      "loss": 1.0527,
+      "step": 3465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9204632980421155e-05,
+      "loss": 1.2075,
+      "step": 3466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9204070410612278e-05,
+      "loss": 1.1963,
+      "step": 3467
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9203507650164295e-05,
+      "loss": 1.1201,
+      "step": 3468
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920294469908886e-05,
+      "loss": 1.1455,
+      "step": 3469
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9202381557397636e-05,
+      "loss": 1.1685,
+      "step": 3470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9201818225102288e-05,
+      "loss": 1.1489,
+      "step": 3471
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9201254702214477e-05,
+      "loss": 1.1353,
+      "step": 3472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920069098874588e-05,
+      "loss": 1.1831,
+      "step": 3473
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920012708470817e-05,
+      "loss": 1.1973,
+      "step": 3474
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9199562990113035e-05,
+      "loss": 1.1548,
+      "step": 3475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919899870497215e-05,
+      "loss": 1.083,
+      "step": 3476
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.91984342292972e-05,
+      "loss": 1.1938,
+      "step": 3477
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919786956309989e-05,
+      "loss": 1.2451,
+      "step": 3478
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9197304706391903e-05,
+      "loss": 1.1606,
+      "step": 3479
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9196739659184945e-05,
+      "loss": 1.0283,
+      "step": 3480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919617442149072e-05,
+      "loss": 1.0645,
+      "step": 3481
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9195608993320933e-05,
+      "loss": 1.1411,
+      "step": 3482
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919504337468729e-05,
+      "loss": 1.1011,
+      "step": 3483
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919447756560152e-05,
+      "loss": 1.0977,
+      "step": 3484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9193911566075327e-05,
+      "loss": 0.9814,
+      "step": 3485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9193345376120445e-05,
+      "loss": 1.2061,
+      "step": 3486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.91927789957486e-05,
+      "loss": 1.1479,
+      "step": 3487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919221242497152e-05,
+      "loss": 1.1689,
+      "step": 3488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.919164566380094e-05,
+      "loss": 1.1733,
+      "step": 3489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9191078712248602e-05,
+      "loss": 1.1519,
+      "step": 3490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9190511570326245e-05,
+      "loss": 1.2148,
+      "step": 3491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918994423804562e-05,
+      "loss": 1.0947,
+      "step": 3492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918937671541847e-05,
+      "loss": 1.2734,
+      "step": 3493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9188809002456565e-05,
+      "loss": 1.0791,
+      "step": 3494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918824109917165e-05,
+      "loss": 1.1387,
+      "step": 3495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918767300557549e-05,
+      "loss": 1.1167,
+      "step": 3496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918710472167986e-05,
+      "loss": 1.1436,
+      "step": 3497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918653624749652e-05,
+      "loss": 1.0645,
+      "step": 3498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918596758303725e-05,
+      "loss": 1.1367,
+      "step": 3499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9185398728313825e-05,
+      "loss": 1.0142,
+      "step": 3500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9184829683338032e-05,
+      "loss": 1.0938,
+      "step": 3501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9184260448121655e-05,
+      "loss": 1.2466,
+      "step": 3502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9183691022676484e-05,
+      "loss": 1.1875,
+      "step": 3503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9183121407014314e-05,
+      "loss": 1.1328,
+      "step": 3504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918255160114694e-05,
+      "loss": 1.0391,
+      "step": 3505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918198160508617e-05,
+      "loss": 1.0786,
+      "step": 3506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9181411418843802e-05,
+      "loss": 1.1245,
+      "step": 3507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9180841042431658e-05,
+      "loss": 1.1562,
+      "step": 3508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918027047586154e-05,
+      "loss": 1.1772,
+      "step": 3509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917969971914527e-05,
+      "loss": 1.0671,
+      "step": 3510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9179128772294674e-05,
+      "loss": 1.04,
+      "step": 3511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917855763532157e-05,
+      "loss": 1.083,
+      "step": 3512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9177986308237793e-05,
+      "loss": 1.0645,
+      "step": 3513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9177414791055174e-05,
+      "loss": 1.1147,
+      "step": 3514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176843083785558e-05,
+      "loss": 1.2412,
+      "step": 3515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176271186440773e-05,
+      "loss": 1.0386,
+      "step": 3516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9175699099032677e-05,
+      "loss": 1.3672,
+      "step": 3517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9175126821573112e-05,
+      "loss": 1.1597,
+      "step": 3518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9174554354073933e-05,
+      "loss": 1.1875,
+      "step": 3519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9173981696546998e-05,
+      "loss": 1.1392,
+      "step": 3520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9173408849004166e-05,
+      "loss": 1.0942,
+      "step": 3521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9172835811457306e-05,
+      "loss": 1.1118,
+      "step": 3522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9172262583918285e-05,
+      "loss": 1.0015,
+      "step": 3523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9171689166398973e-05,
+      "loss": 1.1089,
+      "step": 3524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917111555891125e-05,
+      "loss": 1.2368,
+      "step": 3525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9170541761466998e-05,
+      "loss": 0.9375,
+      "step": 3526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9169967774078102e-05,
+      "loss": 1.1387,
+      "step": 3527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9169393596756447e-05,
+      "loss": 1.0981,
+      "step": 3528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9168819229513928e-05,
+      "loss": 1.1592,
+      "step": 3529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9168244672362442e-05,
+      "loss": 1.2686,
+      "step": 3530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167669925313887e-05,
+      "loss": 1.1636,
+      "step": 3531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167094988380173e-05,
+      "loss": 1.2534,
+      "step": 3532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9166519861573202e-05,
+      "loss": 1.0117,
+      "step": 3533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9165944544904887e-05,
+      "loss": 1.1919,
+      "step": 3534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9165369038387147e-05,
+      "loss": 1.1157,
+      "step": 3535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.91647933420319e-05,
+      "loss": 1.1118,
+      "step": 3536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9164217455851074e-05,
+      "loss": 1.1855,
+      "step": 3537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9163641379856593e-05,
+      "loss": 1.104,
+      "step": 3538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916306511406039e-05,
+      "loss": 1.1904,
+      "step": 3539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9162488658474406e-05,
+      "loss": 1.2285,
+      "step": 3540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916191201311057e-05,
+      "loss": 1.1309,
+      "step": 3541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9161335177980834e-05,
+      "loss": 1.3174,
+      "step": 3542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9160758153097143e-05,
+      "loss": 1.1328,
+      "step": 3543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916018093847145e-05,
+      "loss": 1.0562,
+      "step": 3544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9159603534115707e-05,
+      "loss": 0.9695,
+      "step": 3545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9159025940041877e-05,
+      "loss": 1.0066,
+      "step": 3546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9158448156261923e-05,
+      "loss": 1.2065,
+      "step": 3547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9157870182787816e-05,
+      "loss": 1.1924,
+      "step": 3548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915729201963152e-05,
+      "loss": 1.2075,
+      "step": 3549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915671366680501e-05,
+      "loss": 0.9941,
+      "step": 3550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9156135124320273e-05,
+      "loss": 1.1113,
+      "step": 3551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9155556392189285e-05,
+      "loss": 1.1558,
+      "step": 3552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9154977470424036e-05,
+      "loss": 1.0571,
+      "step": 3553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9154398359036516e-05,
+      "loss": 1.1738,
+      "step": 3554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915381905803872e-05,
+      "loss": 1.0952,
+      "step": 3555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9153239567442647e-05,
+      "loss": 1.1318,
+      "step": 3556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9152659887260302e-05,
+      "loss": 1.0112,
+      "step": 3557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9152080017503686e-05,
+      "loss": 1.168,
+      "step": 3558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9151499958184817e-05,
+      "loss": 1.1885,
+      "step": 3559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.91509197093157e-05,
+      "loss": 1.2729,
+      "step": 3560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915033927090836e-05,
+      "loss": 1.1753,
+      "step": 3561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914975864297482e-05,
+      "loss": 1.1294,
+      "step": 3562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149177825527103e-05,
+      "loss": 1.1836,
+      "step": 3563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9148596818577242e-05,
+      "loss": 1.0127,
+      "step": 3564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9148015622137265e-05,
+      "loss": 1.0986,
+      "step": 3565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914743423621922e-05,
+      "loss": 1.1206,
+      "step": 3566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914685266083514e-05,
+      "loss": 1.1279,
+      "step": 3567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9146270895997074e-05,
+      "loss": 1.0481,
+      "step": 3568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145688941717074e-05,
+      "loss": 1.1616,
+      "step": 3569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145106798007192e-05,
+      "loss": 1.0815,
+      "step": 3570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9144524464879483e-05,
+      "loss": 1.0337,
+      "step": 3571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914394194234601e-05,
+      "loss": 1.1729,
+      "step": 3572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9143359230418845e-05,
+      "loss": 1.1865,
+      "step": 3573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9142776329110047e-05,
+      "loss": 1.1836,
+      "step": 3574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.91421932384317e-05,
+      "loss": 1.1929,
+      "step": 3575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9141609958395867e-05,
+      "loss": 1.1074,
+      "step": 3576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9141026489014644e-05,
+      "loss": 1.1226,
+      "step": 3577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914044283030011e-05,
+      "loss": 1.2314,
+      "step": 3578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913985898226435e-05,
+      "loss": 1.3926,
+      "step": 3579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9139274944919463e-05,
+      "loss": 1.0737,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9138690718277542e-05,
+      "loss": 1.2563,
+      "step": 3581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913810630235069e-05,
+      "loss": 1.2314,
+      "step": 3582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913752169715101e-05,
+      "loss": 1.0513,
+      "step": 3583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136936902690616e-05,
+      "loss": 1.1304,
+      "step": 3584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136351918981614e-05,
+      "loss": 1.0796,
+      "step": 3585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913576674603612e-05,
+      "loss": 1.1963,
+      "step": 3586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9135181383866256e-05,
+      "loss": 1.0591,
+      "step": 3587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9134595832484155e-05,
+      "loss": 1.0125,
+      "step": 3588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913401009190193e-05,
+      "loss": 1.0381,
+      "step": 3589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9133424162131723e-05,
+      "loss": 1.1538,
+      "step": 3590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9132838043185668e-05,
+      "loss": 1.1313,
+      "step": 3591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9132251735075906e-05,
+      "loss": 1.1606,
+      "step": 3592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913166523781458e-05,
+      "loss": 1.0479,
+      "step": 3593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9131078551413834e-05,
+      "loss": 1.0918,
+      "step": 3594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913049167588583e-05,
+      "loss": 1.2085,
+      "step": 3595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9129904611242707e-05,
+      "loss": 1.1777,
+      "step": 3596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912931735749664e-05,
+      "loss": 1.1543,
+      "step": 3597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128729914659788e-05,
+      "loss": 1.0532,
+      "step": 3598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128142282744316e-05,
+      "loss": 1.2666,
+      "step": 3599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9127554461762394e-05,
+      "loss": 1.0146,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9126966451726203e-05,
+      "loss": 1.1001,
+      "step": 3601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912637825264792e-05,
+      "loss": 1.0571,
+      "step": 3602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9125789864539725e-05,
+      "loss": 1.1611,
+      "step": 3603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9125201287413807e-05,
+      "loss": 1.1348,
+      "step": 3604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9124612521282354e-05,
+      "loss": 1.0356,
+      "step": 3605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912402356615757e-05,
+      "loss": 1.0928,
+      "step": 3606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9123434422051642e-05,
+      "loss": 1.1973,
+      "step": 3607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912284508897678e-05,
+      "loss": 1.1982,
+      "step": 3608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9122255566945188e-05,
+      "loss": 1.1299,
+      "step": 3609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9121665855969078e-05,
+      "loss": 1.1025,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912107595606066e-05,
+      "loss": 1.168,
+      "step": 3611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9120485867232162e-05,
+      "loss": 0.9878,
+      "step": 3612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119895589495798e-05,
+      "loss": 0.9863,
+      "step": 3613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119305122863795e-05,
+      "loss": 1.1519,
+      "step": 3614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9118714467348384e-05,
+      "loss": 1.1641,
+      "step": 3615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9118123622961798e-05,
+      "loss": 1.0996,
+      "step": 3616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9117532589716273e-05,
+      "loss": 1.1279,
+      "step": 3617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911694136762406e-05,
+      "loss": 1.2588,
+      "step": 3618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9116349956697394e-05,
+      "loss": 1.1084,
+      "step": 3619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9115758356948533e-05,
+      "loss": 1.2476,
+      "step": 3620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9115166568389726e-05,
+      "loss": 1.1646,
+      "step": 3621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9114574591033225e-05,
+      "loss": 1.0654,
+      "step": 3622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.91139824248913e-05,
+      "loss": 1.1938,
+      "step": 3623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9113390069976216e-05,
+      "loss": 1.1362,
+      "step": 3624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911279752630024e-05,
+      "loss": 1.1707,
+      "step": 3625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911220479387564e-05,
+      "loss": 1.0347,
+      "step": 3626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.91116118727147e-05,
+      "loss": 1.1177,
+      "step": 3627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9111018762829696e-05,
+      "loss": 1.21,
+      "step": 3628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9110425464232918e-05,
+      "loss": 1.1509,
+      "step": 3629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910983197693665e-05,
+      "loss": 1.103,
+      "step": 3630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910923830095319e-05,
+      "loss": 1.1768,
+      "step": 3631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9108644436294825e-05,
+      "loss": 1.1924,
+      "step": 3632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9108050382973866e-05,
+      "loss": 1.1592,
+      "step": 3633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910745614100261e-05,
+      "loss": 1.1943,
+      "step": 3634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9106861710393368e-05,
+      "loss": 1.1587,
+      "step": 3635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9106267091158455e-05,
+      "loss": 1.167,
+      "step": 3636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105672283310182e-05,
+      "loss": 1.1284,
+      "step": 3637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910507728686087e-05,
+      "loss": 1.1519,
+      "step": 3638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9104482101822844e-05,
+      "loss": 1.0481,
+      "step": 3639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910388672820843e-05,
+      "loss": 1.1025,
+      "step": 3640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9103291166029966e-05,
+      "loss": 1.0151,
+      "step": 3641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910269541529978e-05,
+      "loss": 1.0938,
+      "step": 3642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102099476030214e-05,
+      "loss": 1.1626,
+      "step": 3643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9101503348233613e-05,
+      "loss": 1.0571,
+      "step": 3644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910090703192232e-05,
+      "loss": 1.0874,
+      "step": 3645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9100310527108693e-05,
+      "loss": 1.1899,
+      "step": 3646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909971383380508e-05,
+      "loss": 1.001,
+      "step": 3647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9099116952023845e-05,
+      "loss": 1.0635,
+      "step": 3648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909851988177735e-05,
+      "loss": 0.9629,
+      "step": 3649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9097922623077957e-05,
+      "loss": 1.0269,
+      "step": 3650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9097325175938043e-05,
+      "loss": 1.1157,
+      "step": 3651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909672754036998e-05,
+      "loss": 1.1626,
+      "step": 3652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909612971638615e-05,
+      "loss": 1.2275,
+      "step": 3653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9095531703998927e-05,
+      "loss": 1.1641,
+      "step": 3654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094933503220705e-05,
+      "loss": 1.0181,
+      "step": 3655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094335114063873e-05,
+      "loss": 1.1445,
+      "step": 3656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909373653654082e-05,
+      "loss": 1.1055,
+      "step": 3657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909313777066395e-05,
+      "loss": 1.1704,
+      "step": 3658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9092538816445664e-05,
+      "loss": 1.0566,
+      "step": 3659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9091939673898365e-05,
+      "loss": 1.1533,
+      "step": 3660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9091340343034467e-05,
+      "loss": 1.0518,
+      "step": 3661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9090740823866375e-05,
+      "loss": 1.1196,
+      "step": 3662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909014111640652e-05,
+      "loss": 1.0728,
+      "step": 3663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9089541220667312e-05,
+      "loss": 1.1865,
+      "step": 3664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908894113666118e-05,
+      "loss": 1.0762,
+      "step": 3665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9088340864400556e-05,
+      "loss": 1.0449,
+      "step": 3666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087740403897865e-05,
+      "loss": 1.0996,
+      "step": 3667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087139755165552e-05,
+      "loss": 1.1074,
+      "step": 3668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9086538918216056e-05,
+      "loss": 1.0933,
+      "step": 3669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9085937893061826e-05,
+      "loss": 1.0767,
+      "step": 3670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.90853366797153e-05,
+      "loss": 1.0952,
+      "step": 3671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908473527818894e-05,
+      "loss": 1.1294,
+      "step": 3672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.90841336884952e-05,
+      "loss": 1.1846,
+      "step": 3673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9083531910646537e-05,
+      "loss": 1.2021,
+      "step": 3674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9082929944655418e-05,
+      "loss": 1.1646,
+      "step": 3675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908232779053431e-05,
+      "loss": 1.1362,
+      "step": 3676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9081725448295686e-05,
+      "loss": 0.9954,
+      "step": 3677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9081122917952027e-05,
+      "loss": 1.1875,
+      "step": 3678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9080520199515805e-05,
+      "loss": 1.1597,
+      "step": 3679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9079917292999508e-05,
+      "loss": 1.104,
+      "step": 3680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9079314198415622e-05,
+      "loss": 1.0464,
+      "step": 3681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907871091577664e-05,
+      "loss": 1.1074,
+      "step": 3682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9078107445095053e-05,
+      "loss": 1.0444,
+      "step": 3683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907750378638337e-05,
+      "loss": 1.1768,
+      "step": 3684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9076899939654083e-05,
+      "loss": 1.0381,
+      "step": 3685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9076295904919706e-05,
+      "loss": 1.1675,
+      "step": 3686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907569168219275e-05,
+      "loss": 1.0762,
+      "step": 3687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9075087271485726e-05,
+      "loss": 1.1274,
+      "step": 3688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9074482672811158e-05,
+      "loss": 1.1714,
+      "step": 3689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9073877886181568e-05,
+      "loss": 1.1299,
+      "step": 3690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9073272911609475e-05,
+      "loss": 1.1641,
+      "step": 3691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907266774910742e-05,
+      "loss": 1.041,
+      "step": 3692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9072062398687928e-05,
+      "loss": 1.0654,
+      "step": 3693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9071456860363547e-05,
+      "loss": 1.0435,
+      "step": 3694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9070851134146813e-05,
+      "loss": 1.0566,
+      "step": 3695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907024522005027e-05,
+      "loss": 1.1201,
+      "step": 3696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9069639118086474e-05,
+      "loss": 1.147,
+      "step": 3697
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9069032828267973e-05,
+      "loss": 1.1162,
+      "step": 3698
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9068426350607333e-05,
+      "loss": 1.1001,
+      "step": 3699
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9067819685117106e-05,
+      "loss": 1.0527,
+      "step": 3700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9067212831809863e-05,
+      "loss": 1.1528,
+      "step": 3701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9066605790698177e-05,
+      "loss": 1.1631,
+      "step": 3702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906599856179461e-05,
+      "loss": 0.9824,
+      "step": 3703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906539114511175e-05,
+      "loss": 1.1182,
+      "step": 3704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9064783540662175e-05,
+      "loss": 0.9854,
+      "step": 3705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9064175748458466e-05,
+      "loss": 1.1377,
+      "step": 3706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9063567768513218e-05,
+      "loss": 1.147,
+      "step": 3707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062959600839018e-05,
+      "loss": 1.0229,
+      "step": 3708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062351245448466e-05,
+      "loss": 1.1963,
+      "step": 3709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9061742702354166e-05,
+      "loss": 1.2227,
+      "step": 3710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9061133971568715e-05,
+      "loss": 1.0415,
+      "step": 3711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9060525053104724e-05,
+      "loss": 1.0112,
+      "step": 3712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9059915946974806e-05,
+      "loss": 1.1279,
+      "step": 3713
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9059306653191577e-05,
+      "loss": 1.2007,
+      "step": 3714
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9058697171767663e-05,
+      "loss": 1.1143,
+      "step": 3715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9058087502715675e-05,
+      "loss": 1.1689,
+      "step": 3716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9057477646048247e-05,
+      "loss": 1.0894,
+      "step": 3717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9056867601778017e-05,
+      "loss": 1.1313,
+      "step": 3718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9056257369917613e-05,
+      "loss": 1.1445,
+      "step": 3719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9055646950479675e-05,
+      "loss": 1.1177,
+      "step": 3720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9055036343476846e-05,
+      "loss": 1.0815,
+      "step": 3721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9054425548921775e-05,
+      "loss": 1.2432,
+      "step": 3722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9053814566827115e-05,
+      "loss": 1.2686,
+      "step": 3723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905320339720552e-05,
+      "loss": 1.0879,
+      "step": 3724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9052592040069645e-05,
+      "loss": 1.1006,
+      "step": 3725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9051980495432154e-05,
+      "loss": 1.166,
+      "step": 3726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905136876330572e-05,
+      "loss": 1.0703,
+      "step": 3727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9050756843703004e-05,
+      "loss": 1.1172,
+      "step": 3728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9050144736636683e-05,
+      "loss": 1.1309,
+      "step": 3729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9049532442119444e-05,
+      "loss": 1.0732,
+      "step": 3730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048919960163957e-05,
+      "loss": 1.1528,
+      "step": 3731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048307290782915e-05,
+      "loss": 1.1689,
+      "step": 3732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9047694433989007e-05,
+      "loss": 1.0977,
+      "step": 3733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9047081389794924e-05,
+      "loss": 1.2373,
+      "step": 3734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046468158213365e-05,
+      "loss": 1.1846,
+      "step": 3735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904585473925703e-05,
+      "loss": 1.1611,
+      "step": 3736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9045241132938634e-05,
+      "loss": 1.0781,
+      "step": 3737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9044627339270873e-05,
+      "loss": 1.2212,
+      "step": 3738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9044013358266468e-05,
+      "loss": 1.1318,
+      "step": 3739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9043399189938135e-05,
+      "loss": 1.0703,
+      "step": 3740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9042784834298593e-05,
+      "loss": 1.1763,
+      "step": 3741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904217029136057e-05,
+      "loss": 1.0801,
+      "step": 3742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9041555561136787e-05,
+      "loss": 1.1699,
+      "step": 3743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9040940643639986e-05,
+      "loss": 1.0522,
+      "step": 3744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90403255388829e-05,
+      "loss": 1.1074,
+      "step": 3745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9039710246878267e-05,
+      "loss": 1.252,
+      "step": 3746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9039094767638834e-05,
+      "loss": 1.1738,
+      "step": 3747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903847910117735e-05,
+      "loss": 1.064,
+      "step": 3748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903786324750656e-05,
+      "loss": 1.084,
+      "step": 3749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903724720663923e-05,
+      "loss": 1.2451,
+      "step": 3750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9036630978588113e-05,
+      "loss": 1.1392,
+      "step": 3751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9036014563365977e-05,
+      "loss": 0.981,
+      "step": 3752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9035397960985583e-05,
+      "loss": 1.0469,
+      "step": 3753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903478117145971e-05,
+      "loss": 1.2124,
+      "step": 3754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9034164194801128e-05,
+      "loss": 1.2612,
+      "step": 3755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9033547031022617e-05,
+      "loss": 1.1162,
+      "step": 3756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903292968013696e-05,
+      "loss": 1.1104,
+      "step": 3757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9032312142156947e-05,
+      "loss": 1.0298,
+      "step": 3758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9031694417095367e-05,
+      "loss": 1.2305,
+      "step": 3759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9031076504965013e-05,
+      "loss": 1.0361,
+      "step": 3760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.903045840577868e-05,
+      "loss": 1.0918,
+      "step": 3761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902984011954918e-05,
+      "loss": 1.0552,
+      "step": 3762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902922164628931e-05,
+      "loss": 1.0981,
+      "step": 3763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902860298601189e-05,
+      "loss": 1.2407,
+      "step": 3764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9027984138729724e-05,
+      "loss": 0.9722,
+      "step": 3765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9027365104455633e-05,
+      "loss": 1.1562,
+      "step": 3766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902674588320244e-05,
+      "loss": 1.2637,
+      "step": 3767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9026126474982973e-05,
+      "loss": 1.1807,
+      "step": 3768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9025506879810055e-05,
+      "loss": 1.0581,
+      "step": 3769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9024887097696525e-05,
+      "loss": 1.0415,
+      "step": 3770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902426712865522e-05,
+      "loss": 0.9878,
+      "step": 3771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902364697269898e-05,
+      "loss": 1.1021,
+      "step": 3772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9023026629840644e-05,
+      "loss": 1.1377,
+      "step": 3773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9022406100093073e-05,
+      "loss": 1.1733,
+      "step": 3774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9021785383469105e-05,
+      "loss": 0.9653,
+      "step": 3775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902116447998161e-05,
+      "loss": 1.103,
+      "step": 3776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9020543389643446e-05,
+      "loss": 1.0942,
+      "step": 3777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9019922112467467e-05,
+      "loss": 1.0918,
+      "step": 3778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9019300648466553e-05,
+      "loss": 1.1138,
+      "step": 3779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901867899765357e-05,
+      "loss": 1.0801,
+      "step": 3780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018057160041396e-05,
+      "loss": 1.1001,
+      "step": 3781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901743513564291e-05,
+      "loss": 1.1758,
+      "step": 3782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016812924470997e-05,
+      "loss": 1.1553,
+      "step": 3783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016190526538542e-05,
+      "loss": 1.1357,
+      "step": 3784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9015567941858436e-05,
+      "loss": 1.2646,
+      "step": 3785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901494517044358e-05,
+      "loss": 1.1343,
+      "step": 3786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901432221230687e-05,
+      "loss": 1.0234,
+      "step": 3787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9013699067461205e-05,
+      "loss": 1.1992,
+      "step": 3788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9013075735919498e-05,
+      "loss": 1.2251,
+      "step": 3789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012452217694654e-05,
+      "loss": 1.1616,
+      "step": 3790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901182851279959e-05,
+      "loss": 1.1055,
+      "step": 3791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9011204621247228e-05,
+      "loss": 1.21,
+      "step": 3792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9010580543050484e-05,
+      "loss": 1.1572,
+      "step": 3793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9009956278222292e-05,
+      "loss": 1.0039,
+      "step": 3794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9009331826775574e-05,
+      "loss": 1.0176,
+      "step": 3795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900870718872327e-05,
+      "loss": 1.1001,
+      "step": 3796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9008082364078314e-05,
+      "loss": 1.2197,
+      "step": 3797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9007457352853647e-05,
+      "loss": 1.1514,
+      "step": 3798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9006832155062222e-05,
+      "loss": 1.1509,
+      "step": 3799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900620677071698e-05,
+      "loss": 1.0107,
+      "step": 3800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900558119983088e-05,
+      "loss": 1.019,
+      "step": 3801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900495544241687e-05,
+      "loss": 1.1123,
+      "step": 3802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9004329498487922e-05,
+      "loss": 1.1445,
+      "step": 3803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9003703368056995e-05,
+      "loss": 1.1348,
+      "step": 3804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9003077051137054e-05,
+      "loss": 1.084,
+      "step": 3805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9002450547741085e-05,
+      "loss": 1.1641,
+      "step": 3806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900182385788205e-05,
+      "loss": 1.0259,
+      "step": 3807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900119698157294e-05,
+      "loss": 1.0557,
+      "step": 3808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900056991882673e-05,
+      "loss": 1.0381,
+      "step": 3809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8999942669656414e-05,
+      "loss": 1.0645,
+      "step": 3810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899931523407498e-05,
+      "loss": 1.1641,
+      "step": 3811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899868761209543e-05,
+      "loss": 1.1973,
+      "step": 3812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899805980373076e-05,
+      "loss": 1.167,
+      "step": 3813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899743180899397e-05,
+      "loss": 1.0991,
+      "step": 3814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899680362789807e-05,
+      "loss": 1.1865,
+      "step": 3815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8996175260456073e-05,
+      "loss": 1.1802,
+      "step": 3816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899554670668099e-05,
+      "loss": 1.0322,
+      "step": 3817
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8994917966585847e-05,
+      "loss": 1.2148,
+      "step": 3818
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899428904018366e-05,
+      "loss": 1.0845,
+      "step": 3819
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899365992748746e-05,
+      "loss": 1.1416,
+      "step": 3820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8993030628510273e-05,
+      "loss": 1.0137,
+      "step": 3821
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8992401143265134e-05,
+      "loss": 0.9653,
+      "step": 3822
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8991771471765087e-05,
+      "loss": 1.1548,
+      "step": 3823
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899114161402317e-05,
+      "loss": 1.2178,
+      "step": 3824
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899051157005243e-05,
+      "loss": 1.1602,
+      "step": 3825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898988133986591e-05,
+      "loss": 1.0552,
+      "step": 3826
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8989250923476672e-05,
+      "loss": 1.1323,
+      "step": 3827
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8988620320897773e-05,
+      "loss": 1.2314,
+      "step": 3828
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898798953214227e-05,
+      "loss": 1.0146,
+      "step": 3829
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8987358557223232e-05,
+      "loss": 1.1753,
+      "step": 3830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8986727396153724e-05,
+      "loss": 1.2285,
+      "step": 3831
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8986096048946826e-05,
+      "loss": 1.2041,
+      "step": 3832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8985464515615604e-05,
+      "loss": 1.1123,
+      "step": 3833
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8984832796173148e-05,
+      "loss": 1.0718,
+      "step": 3834
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898420089063254e-05,
+      "loss": 1.0698,
+      "step": 3835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8983568799006865e-05,
+      "loss": 1.0503,
+      "step": 3836
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8982936521309215e-05,
+      "loss": 1.1089,
+      "step": 3837
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8982304057552698e-05,
+      "loss": 1.2134,
+      "step": 3838
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981671407750398e-05,
+      "loss": 1.1807,
+      "step": 3839
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981038571915427e-05,
+      "loss": 1.1797,
+      "step": 3840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898040555006089e-05,
+      "loss": 1.0728,
+      "step": 3841
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89797723421999e-05,
+      "loss": 1.1562,
+      "step": 3842
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8979138948345572e-05,
+      "loss": 1.124,
+      "step": 3843
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8978505368511023e-05,
+      "loss": 1.1899,
+      "step": 3844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897787160270938e-05,
+      "loss": 1.1045,
+      "step": 3845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8977237650953764e-05,
+      "loss": 1.0776,
+      "step": 3846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8976603513257316e-05,
+      "loss": 1.1206,
+      "step": 3847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897596918963316e-05,
+      "loss": 1.1274,
+      "step": 3848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897533468009444e-05,
+      "loss": 1.0898,
+      "step": 3849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8974699984654297e-05,
+      "loss": 1.0371,
+      "step": 3850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8974065103325876e-05,
+      "loss": 1.1455,
+      "step": 3851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973430036122326e-05,
+      "loss": 1.0278,
+      "step": 3852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8972794783056806e-05,
+      "loss": 1.2617,
+      "step": 3853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8972159344142465e-05,
+      "loss": 1.2559,
+      "step": 3854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8971523719392474e-05,
+      "loss": 1.165,
+      "step": 3855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8970887908819995e-05,
+      "loss": 1.1426,
+      "step": 3856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8970251912438195e-05,
+      "loss": 1.0083,
+      "step": 3857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8969615730260246e-05,
+      "loss": 1.1675,
+      "step": 3858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8968979362299332e-05,
+      "loss": 1.1299,
+      "step": 3859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8968342808568626e-05,
+      "loss": 1.1152,
+      "step": 3860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8967706069081315e-05,
+      "loss": 1.063,
+      "step": 3861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896706914385059e-05,
+      "loss": 0.9966,
+      "step": 3862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896643203288964e-05,
+      "loss": 1.2656,
+      "step": 3863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8965794736211664e-05,
+      "loss": 1.124,
+      "step": 3864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8965157253829864e-05,
+      "loss": 1.1357,
+      "step": 3865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8964519585757432e-05,
+      "loss": 1.124,
+      "step": 3866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896388173200759e-05,
+      "loss": 1.125,
+      "step": 3867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896324369259354e-05,
+      "loss": 1.1621,
+      "step": 3868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8962605467528506e-05,
+      "loss": 1.0781,
+      "step": 3869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961967056825698e-05,
+      "loss": 1.1069,
+      "step": 3870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961328460498346e-05,
+      "loss": 1.25,
+      "step": 3871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8960689678559672e-05,
+      "loss": 1.1602,
+      "step": 3872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896005071102291e-05,
+      "loss": 1.0488,
+      "step": 3873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8959411557901293e-05,
+      "loss": 1.1118,
+      "step": 3874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8958772219208063e-05,
+      "loss": 1.1895,
+      "step": 3875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8958132694956454e-05,
+      "loss": 1.0723,
+      "step": 3876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8957492985159722e-05,
+      "loss": 1.1958,
+      "step": 3877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895685308983111e-05,
+      "loss": 1.3271,
+      "step": 3878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8956213008983874e-05,
+      "loss": 1.1416,
+      "step": 3879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8955572742631277e-05,
+      "loss": 1.0728,
+      "step": 3880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8954932290786568e-05,
+      "loss": 1.0083,
+      "step": 3881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8954291653463026e-05,
+      "loss": 1.1211,
+      "step": 3882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895365083067391e-05,
+      "loss": 1.0283,
+      "step": 3883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.89530098224325e-05,
+      "loss": 1.1484,
+      "step": 3884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8952368628752066e-05,
+      "loss": 1.1865,
+      "step": 3885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8951727249645896e-05,
+      "loss": 1.1582,
+      "step": 3886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895108568512727e-05,
+      "loss": 1.0059,
+      "step": 3887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8950443935209475e-05,
+      "loss": 1.1631,
+      "step": 3888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8949801999905808e-05,
+      "loss": 1.0876,
+      "step": 3889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8949159879229563e-05,
+      "loss": 1.0767,
+      "step": 3890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.894851757319404e-05,
+      "loss": 1.2212,
+      "step": 3891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.894787508181254e-05,
+      "loss": 0.9619,
+      "step": 3892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8947232405098374e-05,
+      "loss": 1.1631,
+      "step": 3893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8946589543064856e-05,
+      "loss": 1.0093,
+      "step": 3894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8945946495725294e-05,
+      "loss": 1.1997,
+      "step": 3895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8945303263093012e-05,
+      "loss": 1.0601,
+      "step": 3896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8944659845181333e-05,
+      "loss": 1.0264,
+      "step": 3897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894401624200358e-05,
+      "loss": 1.1338,
+      "step": 3898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8943372453573084e-05,
+      "loss": 1.0649,
+      "step": 3899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8942728479903187e-05,
+      "loss": 1.0747,
+      "step": 3900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894208432100722e-05,
+      "loss": 1.126,
+      "step": 3901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894143997689853e-05,
+      "loss": 1.0415,
+      "step": 3902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894079544759045e-05,
+      "loss": 1.0332,
+      "step": 3903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894015073309635e-05,
+      "loss": 1.1235,
+      "step": 3904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893950583342957e-05,
+      "loss": 1.1758,
+      "step": 3905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893886074860347e-05,
+      "loss": 0.9785,
+      "step": 3906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938215478631415e-05,
+      "loss": 1.0806,
+      "step": 3907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937570023526765e-05,
+      "loss": 1.1538,
+      "step": 3908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8936924383302894e-05,
+      "loss": 1.064,
+      "step": 3909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893627855797317e-05,
+      "loss": 1.0952,
+      "step": 3910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893563254755097e-05,
+      "loss": 1.0947,
+      "step": 3911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893498635204968e-05,
+      "loss": 1.1006,
+      "step": 3912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8934339971482676e-05,
+      "loss": 1.146,
+      "step": 3913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933693405863357e-05,
+      "loss": 1.001,
+      "step": 3914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933046655205103e-05,
+      "loss": 1.0254,
+      "step": 3915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8932399719521318e-05,
+      "loss": 1.1401,
+      "step": 3916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931752598825398e-05,
+      "loss": 1.1162,
+      "step": 3917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893110529313075e-05,
+      "loss": 1.019,
+      "step": 3918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8930457802450774e-05,
+      "loss": 1.0293,
+      "step": 3919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892981012679889e-05,
+      "loss": 1.1216,
+      "step": 3920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892916226618851e-05,
+      "loss": 1.0649,
+      "step": 3921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892851422063305e-05,
+      "loss": 1.0669,
+      "step": 3922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8927865990145936e-05,
+      "loss": 1.249,
+      "step": 3923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892721757474059e-05,
+      "loss": 1.1128,
+      "step": 3924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8926568974430445e-05,
+      "loss": 1.1221,
+      "step": 3925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892592018922894e-05,
+      "loss": 1.062,
+      "step": 3926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8925271219149504e-05,
+      "loss": 1.2002,
+      "step": 3927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8924622064205584e-05,
+      "loss": 1.1753,
+      "step": 3928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8923972724410627e-05,
+      "loss": 1.0669,
+      "step": 3929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8923323199778076e-05,
+      "loss": 1.0591,
+      "step": 3930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922673490321393e-05,
+      "loss": 1.1807,
+      "step": 3931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922023596054028e-05,
+      "loss": 1.2686,
+      "step": 3932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8921373516989444e-05,
+      "loss": 1.1011,
+      "step": 3933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8920723253141105e-05,
+      "loss": 1.0757,
+      "step": 3934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8920072804522483e-05,
+      "loss": 1.1831,
+      "step": 3935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.891942217114705e-05,
+      "loss": 1.1626,
+      "step": 3936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8918771353028272e-05,
+      "loss": 1.0898,
+      "step": 3937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8918120350179645e-05,
+      "loss": 1.082,
+      "step": 3938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8917469162614644e-05,
+      "loss": 1.1616,
+      "step": 3939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8916817790346756e-05,
+      "loss": 1.1934,
+      "step": 3940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8916166233389477e-05,
+      "loss": 1.1069,
+      "step": 3941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.89155144917563e-05,
+      "loss": 1.1284,
+      "step": 3942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891486256546072e-05,
+      "loss": 1.0493,
+      "step": 3943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8914210454516246e-05,
+      "loss": 1.1016,
+      "step": 3944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8913558158936384e-05,
+      "loss": 1.2075,
+      "step": 3945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8912905678734646e-05,
+      "loss": 1.0938,
+      "step": 3946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891225301392454e-05,
+      "loss": 1.168,
+      "step": 3947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891160016451959e-05,
+      "loss": 1.1514,
+      "step": 3948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8910947130533314e-05,
+      "loss": 1.1016,
+      "step": 3949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8910293911979244e-05,
+      "loss": 1.2441,
+      "step": 3950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8909640508870905e-05,
+      "loss": 1.02,
+      "step": 3951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8908986921221832e-05,
+      "loss": 1.0918,
+      "step": 3952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890833314904556e-05,
+      "loss": 1.207,
+      "step": 3953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8907679192355634e-05,
+      "loss": 1.0596,
+      "step": 3954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.89070250511656e-05,
+      "loss": 1.1406,
+      "step": 3955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8906370725489002e-05,
+      "loss": 1.0352,
+      "step": 3956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8905716215339395e-05,
+      "loss": 1.1704,
+      "step": 3957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8905061520730336e-05,
+      "loss": 1.0654,
+      "step": 3958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8904406641675384e-05,
+      "loss": 1.1865,
+      "step": 3959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8903751578188104e-05,
+      "loss": 1.2192,
+      "step": 3960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8903096330282064e-05,
+      "loss": 1.0156,
+      "step": 3961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890244089797084e-05,
+      "loss": 1.1499,
+      "step": 3962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8901785281267998e-05,
+      "loss": 1.0903,
+      "step": 3963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8901129480187125e-05,
+      "loss": 1.0908,
+      "step": 3964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8900473494741803e-05,
+      "loss": 1.25,
+      "step": 3965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889981732494562e-05,
+      "loss": 1.1494,
+      "step": 3966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8899160970812163e-05,
+      "loss": 1.2085,
+      "step": 3967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889850443235503e-05,
+      "loss": 1.0479,
+      "step": 3968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8897847709587817e-05,
+      "loss": 1.1064,
+      "step": 3969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889719080252413e-05,
+      "loss": 1.0327,
+      "step": 3970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889653371117757e-05,
+      "loss": 1.1504,
+      "step": 3971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8895876435561754e-05,
+      "loss": 1.0171,
+      "step": 3972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889521897569029e-05,
+      "loss": 1.02,
+      "step": 3973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8894561331576797e-05,
+      "loss": 1.0801,
+      "step": 3974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8893903503234895e-05,
+      "loss": 1.2798,
+      "step": 3975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8893245490678214e-05,
+      "loss": 1.1294,
+      "step": 3976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8892587293920382e-05,
+      "loss": 1.1753,
+      "step": 3977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8891928912975026e-05,
+      "loss": 1.0405,
+      "step": 3978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8891270347855792e-05,
+      "loss": 1.0586,
+      "step": 3979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8890611598576312e-05,
+      "loss": 1.0938,
+      "step": 3980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8889952665150232e-05,
+      "loss": 1.001,
+      "step": 3981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8889293547591205e-05,
+      "loss": 1.0239,
+      "step": 3982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.888863424591288e-05,
+      "loss": 1.0649,
+      "step": 3983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888797476012891e-05,
+      "loss": 1.0364,
+      "step": 3984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888731509025296e-05,
+      "loss": 1.0498,
+      "step": 3985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8886655236298686e-05,
+      "loss": 1.1753,
+      "step": 3986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8885995198279762e-05,
+      "loss": 1.0679,
+      "step": 3987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8885334976209856e-05,
+      "loss": 1.1299,
+      "step": 3988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8884674570102645e-05,
+      "loss": 1.2114,
+      "step": 3989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8884013979971807e-05,
+      "loss": 1.1348,
+      "step": 3990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8883353205831023e-05,
+      "loss": 1.2002,
+      "step": 3991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882692247693982e-05,
+      "loss": 1.0977,
+      "step": 3992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882031105574367e-05,
+      "loss": 1.2021,
+      "step": 3993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8881369779485882e-05,
+      "loss": 1.251,
+      "step": 3994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8880708269442216e-05,
+      "loss": 1.1157,
+      "step": 3995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8880046575457072e-05,
+      "loss": 1.0073,
+      "step": 3996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8879384697544165e-05,
+      "loss": 1.1299,
+      "step": 3997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8878722635717193e-05,
+      "loss": 1.1191,
+      "step": 3998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887806038998987e-05,
+      "loss": 1.0117,
+      "step": 3999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887739796037592e-05,
+      "loss": 0.9858,
+      "step": 4000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8876735346889056e-05,
+      "loss": 1.0903,
+      "step": 4001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8876072549543006e-05,
+      "loss": 1.1323,
+      "step": 4002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8875409568351498e-05,
+      "loss": 1.1265,
+      "step": 4003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887474640332826e-05,
+      "loss": 1.1353,
+      "step": 4004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8874083054487037e-05,
+      "loss": 1.2422,
+      "step": 4005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887341952184156e-05,
+      "loss": 1.186,
+      "step": 4006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887275580540557e-05,
+      "loss": 1.1396,
+      "step": 4007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8872091905192825e-05,
+      "loss": 1.1372,
+      "step": 4008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887142782121707e-05,
+      "loss": 1.0288,
+      "step": 4009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887076355349206e-05,
+      "loss": 1.2114,
+      "step": 4010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8870099102031557e-05,
+      "loss": 1.1045,
+      "step": 4011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8869434466849316e-05,
+      "loss": 1.0942,
+      "step": 4012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8868769647959107e-05,
+      "loss": 1.1021,
+      "step": 4013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88681046453747e-05,
+      "loss": 1.2061,
+      "step": 4014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867439459109874e-05,
+      "loss": 1.0454,
+      "step": 4015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88667740891784e-05,
+      "loss": 1.186,
+      "step": 4016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886610853559406e-05,
+      "loss": 1.1245,
+      "step": 4017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8865442798370642e-05,
+      "loss": 1.1538,
+      "step": 4018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8864776877521936e-05,
+      "loss": 1.04,
+      "step": 4019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886411077306173e-05,
+      "loss": 1.0903,
+      "step": 4020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8863444485003825e-05,
+      "loss": 0.9966,
+      "step": 4021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8862778013362017e-05,
+      "loss": 1.0005,
+      "step": 4022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8862111358150118e-05,
+      "loss": 1.2783,
+      "step": 4023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8861444519381927e-05,
+      "loss": 1.2578,
+      "step": 4024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8860777497071264e-05,
+      "loss": 1.2158,
+      "step": 4025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8860110291231936e-05,
+      "loss": 1.123,
+      "step": 4026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8859442901877774e-05,
+      "loss": 1.2339,
+      "step": 4027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8858775329022587e-05,
+      "loss": 1.126,
+      "step": 4028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8858107572680216e-05,
+      "loss": 1.1738,
+      "step": 4029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885743963286448e-05,
+      "loss": 1.0669,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8856771509589224e-05,
+      "loss": 1.1216,
+      "step": 4031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8856103202868283e-05,
+      "loss": 1.1184,
+      "step": 4032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8855434712715493e-05,
+      "loss": 1.1182,
+      "step": 4033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8854766039144708e-05,
+      "loss": 1.0073,
+      "step": 4034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885409718216977e-05,
+      "loss": 1.2554,
+      "step": 4035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8853428141804545e-05,
+      "loss": 1.1313,
+      "step": 4036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852758918062882e-05,
+      "loss": 1.2607,
+      "step": 4037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885208951095864e-05,
+      "loss": 1.1841,
+      "step": 4038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8851419920505692e-05,
+      "loss": 1.2329,
+      "step": 4039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8850750146717896e-05,
+      "loss": 1.062,
+      "step": 4040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8850080189609138e-05,
+      "loss": 1.2275,
+      "step": 4041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8849410049193285e-05,
+      "loss": 1.1157,
+      "step": 4042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884873972548422e-05,
+      "loss": 1.0347,
+      "step": 4043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8848069218495827e-05,
+      "loss": 1.1909,
+      "step": 4044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884739852824199e-05,
+      "loss": 1.0605,
+      "step": 4045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884672765473661e-05,
+      "loss": 1.0371,
+      "step": 4046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8846056597993577e-05,
+      "loss": 0.8994,
+      "step": 4047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8845385358026783e-05,
+      "loss": 1.2354,
+      "step": 4048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8844713934850147e-05,
+      "loss": 1.1533,
+      "step": 4049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8844042328477564e-05,
+      "loss": 1.0688,
+      "step": 4050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8843370538922942e-05,
+      "loss": 1.1221,
+      "step": 4051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8842698566200207e-05,
+      "loss": 1.1338,
+      "step": 4052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884202641032327e-05,
+      "loss": 1.1592,
+      "step": 4053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8841354071306056e-05,
+      "loss": 1.1689,
+      "step": 4054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8840681549162487e-05,
+      "loss": 1.0508,
+      "step": 4055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8840008843906496e-05,
+      "loss": 1.0034,
+      "step": 4056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8839335955552013e-05,
+      "loss": 1.1841,
+      "step": 4057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883866288411298e-05,
+      "loss": 1.1494,
+      "step": 4058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8837989629603335e-05,
+      "loss": 1.1719,
+      "step": 4059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8837316192037023e-05,
+      "loss": 1.0137,
+      "step": 4060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883664257142799e-05,
+      "loss": 0.9746,
+      "step": 4061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8835968767790192e-05,
+      "loss": 1.0488,
+      "step": 4062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8835294781137588e-05,
+      "loss": 1.1465,
+      "step": 4063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883462061148413e-05,
+      "loss": 1.207,
+      "step": 4064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8833946258843787e-05,
+      "loss": 1.1509,
+      "step": 4065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8833271723230522e-05,
+      "loss": 1.2031,
+      "step": 4066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8832597004658314e-05,
+      "loss": 1.2085,
+      "step": 4067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883192210314113e-05,
+      "loss": 1.1494,
+      "step": 4068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883124701869296e-05,
+      "loss": 1.1904,
+      "step": 4069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883057175132777e-05,
+      "loss": 1.1187,
+      "step": 4070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.882989630105956e-05,
+      "loss": 0.9805,
+      "step": 4071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.882922066790232e-05,
+      "loss": 1.0845,
+      "step": 4072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8828544851870032e-05,
+      "loss": 1.0547,
+      "step": 4073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882786885297671e-05,
+      "loss": 1.0898,
+      "step": 4074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8827192671236338e-05,
+      "loss": 1.1348,
+      "step": 4075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8826516306662936e-05,
+      "loss": 1.1108,
+      "step": 4076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882583975927051e-05,
+      "loss": 1.1108,
+      "step": 4077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8825163029073067e-05,
+      "loss": 1.0811,
+      "step": 4078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8824486116084633e-05,
+      "loss": 1.1602,
+      "step": 4079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8823809020319218e-05,
+      "loss": 1.0903,
+      "step": 4080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8823131741790853e-05,
+      "loss": 1.1733,
+      "step": 4081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8822454280513564e-05,
+      "loss": 1.1016,
+      "step": 4082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8821776636501385e-05,
+      "loss": 1.0679,
+      "step": 4083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8821098809768352e-05,
+      "loss": 1.2212,
+      "step": 4084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.88204208003285e-05,
+      "loss": 1.1108,
+      "step": 4085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881974260819588e-05,
+      "loss": 1.0422,
+      "step": 4086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881906423338453e-05,
+      "loss": 1.0054,
+      "step": 4087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8818385675908502e-05,
+      "loss": 1.1523,
+      "step": 4088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8817706935781855e-05,
+      "loss": 1.1411,
+      "step": 4089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881702801301865e-05,
+      "loss": 1.1357,
+      "step": 4090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8816348907632943e-05,
+      "loss": 1.0225,
+      "step": 4091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8815669619638803e-05,
+      "loss": 1.2222,
+      "step": 4092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8814990149050295e-05,
+      "loss": 1.1226,
+      "step": 4093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8814310495881498e-05,
+      "loss": 1.0076,
+      "step": 4094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881363066014649e-05,
+      "loss": 1.1401,
+      "step": 4095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8812950641859346e-05,
+      "loss": 1.0293,
+      "step": 4096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8812270441034156e-05,
+      "loss": 1.1538,
+      "step": 4097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881159005768501e-05,
+      "loss": 1.1245,
+      "step": 4098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8810909491825996e-05,
+      "loss": 1.0488,
+      "step": 4099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8810228743471214e-05,
+      "loss": 1.1958,
+      "step": 4100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8809547812634755e-05,
+      "loss": 1.0781,
+      "step": 4101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8808866699330736e-05,
+      "loss": 1.0435,
+      "step": 4102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8808185403573255e-05,
+      "loss": 1.1655,
+      "step": 4103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8807503925376428e-05,
+      "loss": 1.1621,
+      "step": 4104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880682226475437e-05,
+      "loss": 1.189,
+      "step": 4105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8806140421721193e-05,
+      "loss": 1.1465,
+      "step": 4106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880545839629103e-05,
+      "loss": 0.9736,
+      "step": 4107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8804776188477998e-05,
+      "loss": 1.0869,
+      "step": 4108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8804093798296236e-05,
+      "loss": 1.1704,
+      "step": 4109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8803411225759875e-05,
+      "loss": 1.1191,
+      "step": 4110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8802728470883046e-05,
+      "loss": 1.0522,
+      "step": 4111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.88020455336799e-05,
+      "loss": 1.0859,
+      "step": 4112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8801362414164578e-05,
+      "loss": 1.0479,
+      "step": 4113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8800679112351226e-05,
+      "loss": 1.2207,
+      "step": 4114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8799995628254003e-05,
+      "loss": 1.0415,
+      "step": 4115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8799311961887062e-05,
+      "loss": 1.0923,
+      "step": 4116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8798628113264568e-05,
+      "loss": 1.125,
+      "step": 4117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8797944082400675e-05,
+      "loss": 1.1528,
+      "step": 4118
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8797259869309563e-05,
+      "loss": 1.167,
+      "step": 4119
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8796575474005398e-05,
+      "loss": 1.1108,
+      "step": 4120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8795890896502355e-05,
+      "loss": 1.1274,
+      "step": 4121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8795206136814614e-05,
+      "loss": 1.1484,
+      "step": 4122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8794521194956358e-05,
+      "loss": 1.1704,
+      "step": 4123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8793836070941775e-05,
+      "loss": 1.0327,
+      "step": 4124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8793150764785057e-05,
+      "loss": 1.1514,
+      "step": 4125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879246527650039e-05,
+      "loss": 1.3711,
+      "step": 4126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8791779606101982e-05,
+      "loss": 1.0356,
+      "step": 4127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8791093753604032e-05,
+      "loss": 1.0591,
+      "step": 4128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879040771902074e-05,
+      "loss": 1.1147,
+      "step": 4129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878972150236632e-05,
+      "loss": 1.0967,
+      "step": 4130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8789035103654994e-05,
+      "loss": 1.0264,
+      "step": 4131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8788348522900966e-05,
+      "loss": 1.0928,
+      "step": 4132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878766176011846e-05,
+      "loss": 1.0884,
+      "step": 4133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8786974815321704e-05,
+      "loss": 1.0195,
+      "step": 4134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8786287688524925e-05,
+      "loss": 1.1704,
+      "step": 4135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878560037974235e-05,
+      "loss": 1.1958,
+      "step": 4136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784912888988224e-05,
+      "loss": 1.1401,
+      "step": 4137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878422521627678e-05,
+      "loss": 1.0791,
+      "step": 4138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878353736162226e-05,
+      "loss": 1.1343,
+      "step": 4139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878284932503892e-05,
+      "loss": 1.083,
+      "step": 4140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8782161106541005e-05,
+      "loss": 1.083,
+      "step": 4141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878147270614277e-05,
+      "loss": 1.1875,
+      "step": 4142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8780784123858468e-05,
+      "loss": 1.3096,
+      "step": 4143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8780095359702373e-05,
+      "loss": 1.0327,
+      "step": 4144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877940641368874e-05,
+      "loss": 1.0269,
+      "step": 4145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8778717285831848e-05,
+      "loss": 0.9761,
+      "step": 4146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8778027976145962e-05,
+      "loss": 1.124,
+      "step": 4147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8777338484645365e-05,
+      "loss": 1.1099,
+      "step": 4148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877664881134434e-05,
+      "loss": 1.0869,
+      "step": 4149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8775958956257168e-05,
+      "loss": 1.0269,
+      "step": 4150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8775268919398135e-05,
+      "loss": 0.9932,
+      "step": 4151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877457870078154e-05,
+      "loss": 1.2617,
+      "step": 4152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877388830042167e-05,
+      "loss": 1.1855,
+      "step": 4153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877319771833283e-05,
+      "loss": 1.0605,
+      "step": 4154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877250695452933e-05,
+      "loss": 1.0952,
+      "step": 4155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877181600902547e-05,
+      "loss": 1.0952,
+      "step": 4156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8771124881835562e-05,
+      "loss": 1.0444,
+      "step": 4157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8770433572973922e-05,
+      "loss": 1.1045,
+      "step": 4158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.876974208245487e-05,
+      "loss": 0.9958,
+      "step": 4159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8769050410292723e-05,
+      "loss": 0.9243,
+      "step": 4160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8768358556501812e-05,
+      "loss": 1.1621,
+      "step": 4161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.876766652109647e-05,
+      "loss": 1.0713,
+      "step": 4162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8766974304091022e-05,
+      "loss": 1.2183,
+      "step": 4163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876628190549981e-05,
+      "loss": 1.1206,
+      "step": 4164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876558932533718e-05,
+      "loss": 1.0972,
+      "step": 4165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876489656361747e-05,
+      "loss": 1.022,
+      "step": 4166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8764203620355032e-05,
+      "loss": 1.2461,
+      "step": 4167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876351049556422e-05,
+      "loss": 0.9922,
+      "step": 4168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8762817189259384e-05,
+      "loss": 1.0366,
+      "step": 4169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8762123701454896e-05,
+      "loss": 1.082,
+      "step": 4170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8761430032165106e-05,
+      "loss": 1.1396,
+      "step": 4171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8760736181404394e-05,
+      "loss": 1.1387,
+      "step": 4172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876004214918712e-05,
+      "loss": 1.0601,
+      "step": 4173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8759347935527667e-05,
+      "loss": 1.043,
+      "step": 4174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875865354044041e-05,
+      "loss": 1.1753,
+      "step": 4175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8757958963939734e-05,
+      "loss": 1.168,
+      "step": 4176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875726420604002e-05,
+      "loss": 1.0396,
+      "step": 4177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8756569266755667e-05,
+      "loss": 1.1201,
+      "step": 4178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8755874146101067e-05,
+      "loss": 1.2744,
+      "step": 4179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8755178844090612e-05,
+      "loss": 1.0845,
+      "step": 4180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8754483360738706e-05,
+      "loss": 1.3052,
+      "step": 4181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875378769605976e-05,
+      "loss": 1.1338,
+      "step": 4182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8753091850068172e-05,
+      "loss": 1.145,
+      "step": 4183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8752395822778365e-05,
+      "loss": 1.1851,
+      "step": 4184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8751699614204745e-05,
+      "loss": 1.0596,
+      "step": 4185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875100322436174e-05,
+      "loss": 1.1309,
+      "step": 4186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8750306653263775e-05,
+      "loss": 1.0742,
+      "step": 4187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8749609900925275e-05,
+      "loss": 0.957,
+      "step": 4188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8748912967360668e-05,
+      "loss": 1.0508,
+      "step": 4189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8748215852584393e-05,
+      "loss": 1.0029,
+      "step": 4190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874751855661089e-05,
+      "loss": 1.0464,
+      "step": 4191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8746821079454603e-05,
+      "loss": 1.0713,
+      "step": 4192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874612342112997e-05,
+      "loss": 1.1172,
+      "step": 4193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874542558165145e-05,
+      "loss": 1.0388,
+      "step": 4194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8744727561033493e-05,
+      "loss": 1.1021,
+      "step": 4195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8744029359290558e-05,
+      "loss": 1.0554,
+      "step": 4196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8743330976437107e-05,
+      "loss": 1.0527,
+      "step": 4197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8742632412487603e-05,
+      "loss": 1.106,
+      "step": 4198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8741933667456512e-05,
+      "loss": 1.2041,
+      "step": 4199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874123474135832e-05,
+      "loss": 1.1807,
+      "step": 4200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874053563420749e-05,
+      "loss": 1.0942,
+      "step": 4201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8739836346018507e-05,
+      "loss": 1.2402,
+      "step": 4202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8739136876805854e-05,
+      "loss": 1.1543,
+      "step": 4203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8738437226584022e-05,
+      "loss": 1.0205,
+      "step": 4204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87377373953675e-05,
+      "loss": 1.1729,
+      "step": 4205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8737037383170778e-05,
+      "loss": 1.1621,
+      "step": 4206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8736337190008362e-05,
+      "loss": 1.0957,
+      "step": 4207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8735636815894755e-05,
+      "loss": 1.1846,
+      "step": 4208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8734936260844465e-05,
+      "loss": 1.0171,
+      "step": 4209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8734235524871996e-05,
+      "loss": 1.0488,
+      "step": 4210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873353460799186e-05,
+      "loss": 1.0356,
+      "step": 4211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8732833510218585e-05,
+      "loss": 1.0527,
+      "step": 4212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8732132231566686e-05,
+      "loss": 1.1118,
+      "step": 4213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731430772050687e-05,
+      "loss": 1.1421,
+      "step": 4214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8730729131685122e-05,
+      "loss": 1.1841,
+      "step": 4215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8730027310484517e-05,
+      "loss": 1.0244,
+      "step": 4216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8729325308463415e-05,
+      "loss": 1.2041,
+      "step": 4217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872862312563635e-05,
+      "loss": 1.1118,
+      "step": 4218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8727920762017872e-05,
+      "loss": 1.0747,
+      "step": 4219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872721821762253e-05,
+      "loss": 1.1221,
+      "step": 4220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8726515492464866e-05,
+      "loss": 1.1401,
+      "step": 4221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8725812586559442e-05,
+      "loss": 1.1055,
+      "step": 4222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8725109499920817e-05,
+      "loss": 1.0977,
+      "step": 4223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872440623256355e-05,
+      "loss": 0.9248,
+      "step": 4224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8723702784502214e-05,
+      "loss": 1.043,
+      "step": 4225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8722999155751372e-05,
+      "loss": 1.2129,
+      "step": 4226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8722295346325597e-05,
+      "loss": 1.1858,
+      "step": 4227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8721591356239478e-05,
+      "loss": 1.0835,
+      "step": 4228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8720887185507585e-05,
+      "loss": 1.083,
+      "step": 4229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8720182834144506e-05,
+      "loss": 0.8767,
+      "step": 4230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8719478302164827e-05,
+      "loss": 1.1582,
+      "step": 4231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8718773589583152e-05,
+      "loss": 1.1504,
+      "step": 4232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8718068696414066e-05,
+      "loss": 0.9561,
+      "step": 4233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871736362267217e-05,
+      "loss": 1.1362,
+      "step": 4234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8716658368372076e-05,
+      "loss": 1.0811,
+      "step": 4235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8715952933528385e-05,
+      "loss": 1.0889,
+      "step": 4236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8715247318155707e-05,
+      "loss": 1.0469,
+      "step": 4237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871454152226866e-05,
+      "loss": 1.1665,
+      "step": 4238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8713835545881865e-05,
+      "loss": 1.0898,
+      "step": 4239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8713129389009942e-05,
+      "loss": 1.1851,
+      "step": 4240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8712423051667512e-05,
+      "loss": 1.0693,
+      "step": 4241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8711716533869216e-05,
+      "loss": 1.0552,
+      "step": 4242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8711009835629677e-05,
+      "loss": 1.1382,
+      "step": 4243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871030295696354e-05,
+      "loss": 1.1523,
+      "step": 4244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870959589788544e-05,
+      "loss": 1.2002,
+      "step": 4245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870888865841003e-05,
+      "loss": 1.2793,
+      "step": 4246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8708181238551952e-05,
+      "loss": 1.0161,
+      "step": 4247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870747363832586e-05,
+      "loss": 0.9854,
+      "step": 4248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8706765857746413e-05,
+      "loss": 1.0327,
+      "step": 4249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8706057896828265e-05,
+      "loss": 1.0786,
+      "step": 4250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8705349755586087e-05,
+      "loss": 0.9912,
+      "step": 4251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870464143403454e-05,
+      "loss": 1.1885,
+      "step": 4252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8703932932188296e-05,
+      "loss": 1.1094,
+      "step": 4253
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8703224250062035e-05,
+      "loss": 1.1572,
+      "step": 4254
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8702515387670433e-05,
+      "loss": 1.0547,
+      "step": 4255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.870180634502817e-05,
+      "loss": 1.207,
+      "step": 4256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.870109712214993e-05,
+      "loss": 1.1313,
+      "step": 4257
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8700387719050412e-05,
+      "loss": 0.9209,
+      "step": 4258
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8699678135744297e-05,
+      "loss": 0.9585,
+      "step": 4259
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8698968372246292e-05,
+      "loss": 1.0977,
+      "step": 4260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8698258428571097e-05,
+      "loss": 1.2148,
+      "step": 4261
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869754830473341e-05,
+      "loss": 1.1484,
+      "step": 4262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8696838000747948e-05,
+      "loss": 1.188,
+      "step": 4263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869612751662942e-05,
+      "loss": 1.0752,
+      "step": 4264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8695416852392536e-05,
+      "loss": 1.103,
+      "step": 4265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8694706008052023e-05,
+      "loss": 1.0591,
+      "step": 4266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.86939949836226e-05,
+      "loss": 1.1533,
+      "step": 4267
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8693283779119e-05,
+      "loss": 1.187,
+      "step": 4268
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8692572394555948e-05,
+      "loss": 1.1035,
+      "step": 4269
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869186082994818e-05,
+      "loss": 1.0854,
+      "step": 4270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8691149085310436e-05,
+      "loss": 1.0518,
+      "step": 4271
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8690437160657455e-05,
+      "loss": 1.0967,
+      "step": 4272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8689725056003986e-05,
+      "loss": 1.1387,
+      "step": 4273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8689012771364776e-05,
+      "loss": 1.0752,
+      "step": 4274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868830030675458e-05,
+      "loss": 1.1001,
+      "step": 4275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868758766218815e-05,
+      "loss": 1.1729,
+      "step": 4276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8686874837680255e-05,
+      "loss": 1.0229,
+      "step": 4277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8686161833245654e-05,
+      "loss": 1.0298,
+      "step": 4278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8685448648899114e-05,
+      "loss": 1.1719,
+      "step": 4279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868473528465541e-05,
+      "loss": 1.085,
+      "step": 4280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8684021740529315e-05,
+      "loss": 1.1567,
+      "step": 4281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8683308016535607e-05,
+      "loss": 0.9951,
+      "step": 4282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8682594112689077e-05,
+      "loss": 1.0415,
+      "step": 4283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8681880029004505e-05,
+      "loss": 1.1411,
+      "step": 4284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8681165765496683e-05,
+      "loss": 1.0532,
+      "step": 4285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8680451322180407e-05,
+      "loss": 1.1011,
+      "step": 4286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867973669907047e-05,
+      "loss": 1.0933,
+      "step": 4287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8679021896181674e-05,
+      "loss": 1.127,
+      "step": 4288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867830691352883e-05,
+      "loss": 0.9863,
+      "step": 4289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867759175112674e-05,
+      "loss": 1.1284,
+      "step": 4290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8676876408990226e-05,
+      "loss": 1.0264,
+      "step": 4291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8676160887134096e-05,
+      "loss": 1.2314,
+      "step": 4292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8675445185573176e-05,
+      "loss": 1.1191,
+      "step": 4293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8674729304322284e-05,
+      "loss": 0.9897,
+      "step": 4294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8674013243396252e-05,
+      "loss": 1.0562,
+      "step": 4295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867329700280991e-05,
+      "loss": 1.1025,
+      "step": 4296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8672580582578094e-05,
+      "loss": 1.0537,
+      "step": 4297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8671863982715647e-05,
+      "loss": 1.0786,
+      "step": 4298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.86711472032374e-05,
+      "loss": 1.1465,
+      "step": 4299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8670430244158208e-05,
+      "loss": 1.1162,
+      "step": 4300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8669713105492923e-05,
+      "loss": 1.0986,
+      "step": 4301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866899578725639e-05,
+      "loss": 1.0415,
+      "step": 4302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8668278289463476e-05,
+      "loss": 1.1748,
+      "step": 4303
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866756061212903e-05,
+      "loss": 1.0518,
+      "step": 4304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8666842755267933e-05,
+      "loss": 1.1958,
+      "step": 4305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866612471889504e-05,
+      "loss": 1.1284,
+      "step": 4306
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8665406503025226e-05,
+      "loss": 1.0542,
+      "step": 4307
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8664688107673376e-05,
+      "loss": 1.0283,
+      "step": 4308
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866396953285436e-05,
+      "loss": 1.0811,
+      "step": 4309
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8663250778583064e-05,
+      "loss": 0.998,
+      "step": 4310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8662531844874373e-05,
+      "loss": 1.2056,
+      "step": 4311
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8661812731743185e-05,
+      "loss": 1.1421,
+      "step": 4312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866109343920439e-05,
+      "loss": 1.186,
+      "step": 4313
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8660373967272884e-05,
+      "loss": 1.2461,
+      "step": 4314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8659654315963572e-05,
+      "loss": 1.1304,
+      "step": 4315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865893448529136e-05,
+      "loss": 0.9722,
+      "step": 4316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8658214475271158e-05,
+      "loss": 1.0679,
+      "step": 4317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8657494285917878e-05,
+      "loss": 1.0068,
+      "step": 4318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8656773917246432e-05,
+      "loss": 1.2178,
+      "step": 4319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8656053369271752e-05,
+      "loss": 1.0718,
+      "step": 4320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8655332642008753e-05,
+      "loss": 0.9521,
+      "step": 4321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8654611735472365e-05,
+      "loss": 1.1182,
+      "step": 4322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865389064967752e-05,
+      "loss": 1.0732,
+      "step": 4323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865316938463916e-05,
+      "loss": 1.1328,
+      "step": 4324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8652447940372214e-05,
+      "loss": 0.9688,
+      "step": 4325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8651726316891628e-05,
+      "loss": 1.147,
+      "step": 4326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8651004514212353e-05,
+      "loss": 1.1846,
+      "step": 4327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8650282532349332e-05,
+      "loss": 1.1377,
+      "step": 4328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864956037131753e-05,
+      "loss": 1.105,
+      "step": 4329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8648838031131894e-05,
+      "loss": 1.1382,
+      "step": 4330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8648115511807386e-05,
+      "loss": 1.106,
+      "step": 4331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864739281335898e-05,
+      "loss": 1.1685,
+      "step": 4332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8646669935801635e-05,
+      "loss": 0.9771,
+      "step": 4333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8645946879150332e-05,
+      "loss": 1.0869,
+      "step": 4334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8645223643420042e-05,
+      "loss": 1.1387,
+      "step": 4335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8644500228625747e-05,
+      "loss": 1.2061,
+      "step": 4336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864377663478243e-05,
+      "loss": 1.0674,
+      "step": 4337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8643052861905077e-05,
+      "loss": 0.9321,
+      "step": 4338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8642328910008677e-05,
+      "loss": 1.2021,
+      "step": 4339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8641604779108236e-05,
+      "loss": 1.1919,
+      "step": 4340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8640880469218738e-05,
+      "loss": 1.166,
+      "step": 4341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8640155980355196e-05,
+      "loss": 1.0527,
+      "step": 4342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8639431312532614e-05,
+      "loss": 1.2065,
+      "step": 4343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8638706465765997e-05,
+      "loss": 0.9827,
+      "step": 4344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8637981440070362e-05,
+      "loss": 1.1577,
+      "step": 4345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8637256235460724e-05,
+      "loss": 1.1729,
+      "step": 4346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8636530851952107e-05,
+      "loss": 1.0664,
+      "step": 4347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8635805289559533e-05,
+      "loss": 1.0776,
+      "step": 4348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8635079548298033e-05,
+      "loss": 1.0903,
+      "step": 4349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8634353628182636e-05,
+      "loss": 1.0645,
+      "step": 4350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8633627529228375e-05,
+      "loss": 1.0679,
+      "step": 4351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8632901251450294e-05,
+      "loss": 1.0635,
+      "step": 4352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8632174794863436e-05,
+      "loss": 1.0591,
+      "step": 4353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8631448159482847e-05,
+      "loss": 1.1235,
+      "step": 4354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863072134532357e-05,
+      "loss": 1.0679,
+      "step": 4355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8629994352400675e-05,
+      "loss": 1.0122,
+      "step": 4356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862926718072921e-05,
+      "loss": 1.1191,
+      "step": 4357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862853983032423e-05,
+      "loss": 1.0732,
+      "step": 4358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862781230120081e-05,
+      "loss": 1.0361,
+      "step": 4359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862708459337402e-05,
+      "loss": 1.1128,
+      "step": 4360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8626356706858927e-05,
+      "loss": 0.9453,
+      "step": 4361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862562864167061e-05,
+      "loss": 1.0913,
+      "step": 4362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8624900397824147e-05,
+      "loss": 1.1021,
+      "step": 4363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8624171975334625e-05,
+      "loss": 1.0483,
+      "step": 4364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8623443374217126e-05,
+      "loss": 1.0996,
+      "step": 4365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8622714594486746e-05,
+      "loss": 1.1528,
+      "step": 4366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862198563615858e-05,
+      "loss": 1.248,
+      "step": 4367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8621256499247725e-05,
+      "loss": 0.9683,
+      "step": 4368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8620527183769283e-05,
+      "loss": 1.1201,
+      "step": 4369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861979768973836e-05,
+      "loss": 1.0757,
+      "step": 4370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8619068017170064e-05,
+      "loss": 1.0454,
+      "step": 4371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8618338166079513e-05,
+      "loss": 1.1172,
+      "step": 4372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861760813648182e-05,
+      "loss": 1.292,
+      "step": 4373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8616877928392107e-05,
+      "loss": 0.998,
+      "step": 4374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8616147541825497e-05,
+      "loss": 1.0234,
+      "step": 4375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8615416976797123e-05,
+      "loss": 1.167,
+      "step": 4376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861468623332211e-05,
+      "loss": 1.1069,
+      "step": 4377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8613955311415595e-05,
+      "loss": 1.1782,
+      "step": 4378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8613224211092725e-05,
+      "loss": 1.0547,
+      "step": 4379
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8612492932368633e-05,
+      "loss": 1.1567,
+      "step": 4380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861176147525847e-05,
+      "loss": 1.0581,
+      "step": 4381
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861102983977739e-05,
+      "loss": 1.0547,
+      "step": 4382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8610298025940538e-05,
+      "loss": 0.9829,
+      "step": 4383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8609566033763075e-05,
+      "loss": 1.1113,
+      "step": 4384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8608833863260165e-05,
+      "loss": 1.1167,
+      "step": 4385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8608101514446973e-05,
+      "loss": 1.0962,
+      "step": 4386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8607368987338666e-05,
+      "loss": 1.0762,
+      "step": 4387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860663628195042e-05,
+      "loss": 1.1836,
+      "step": 4388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860590339829741e-05,
+      "loss": 1.2227,
+      "step": 4389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8605170336394806e-05,
+      "loss": 1.2178,
+      "step": 4390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8604437096257802e-05,
+      "loss": 1.1372,
+      "step": 4391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8603703677901587e-05,
+      "loss": 0.9697,
+      "step": 4392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8602970081341342e-05,
+      "loss": 1.02,
+      "step": 4393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860223630659227e-05,
+      "loss": 1.105,
+      "step": 4394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8601502353669568e-05,
+      "loss": 1.1182,
+      "step": 4395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8600768222588436e-05,
+      "loss": 1.0291,
+      "step": 4396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860003391336408e-05,
+      "loss": 1.1729,
+      "step": 4397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8599299426011708e-05,
+      "loss": 1.0962,
+      "step": 4398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859856476054654e-05,
+      "loss": 1.168,
+      "step": 4399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859782991698378e-05,
+      "loss": 1.0137,
+      "step": 4400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8597094895338662e-05,
+      "loss": 1.0361,
+      "step": 4401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85963596956264e-05,
+      "loss": 1.1104,
+      "step": 4402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8595624317862226e-05,
+      "loss": 1.1055,
+      "step": 4403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8594888762061374e-05,
+      "loss": 1.0991,
+      "step": 4404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8594153028239073e-05,
+      "loss": 1.1982,
+      "step": 4405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8593417116410567e-05,
+      "loss": 1.1812,
+      "step": 4406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85926810265911e-05,
+      "loss": 1.0366,
+      "step": 4407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859194475879591e-05,
+      "loss": 1.1177,
+      "step": 4408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8591208313040258e-05,
+      "loss": 1.0918,
+      "step": 4409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8590471689339385e-05,
+      "loss": 1.0928,
+      "step": 4410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8589734887708556e-05,
+      "loss": 0.9424,
+      "step": 4411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8588997908163037e-05,
+      "loss": 1.1489,
+      "step": 4412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858826075071808e-05,
+      "loss": 1.1333,
+      "step": 4413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8587523415388963e-05,
+      "loss": 1.1323,
+      "step": 4414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8586785902190955e-05,
+      "loss": 1.1709,
+      "step": 4415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8586048211139333e-05,
+      "loss": 1.1021,
+      "step": 4416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8585310342249375e-05,
+      "loss": 1.0547,
+      "step": 4417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858457229553636e-05,
+      "loss": 1.0884,
+      "step": 4418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8583834071015584e-05,
+      "loss": 1.0098,
+      "step": 4419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858309566870233e-05,
+      "loss": 1.0464,
+      "step": 4420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8582357088611897e-05,
+      "loss": 1.0938,
+      "step": 4421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8581618330759577e-05,
+      "loss": 1.2388,
+      "step": 4422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8580879395160676e-05,
+      "loss": 1.166,
+      "step": 4423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8580140281830496e-05,
+      "loss": 1.0449,
+      "step": 4424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8579400990784352e-05,
+      "loss": 1.0054,
+      "step": 4425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8578661522037555e-05,
+      "loss": 1.1143,
+      "step": 4426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8577921875605413e-05,
+      "loss": 1.1724,
+      "step": 4427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8577182051503252e-05,
+      "loss": 1.1582,
+      "step": 4428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85764420497464e-05,
+      "loss": 1.1475,
+      "step": 4429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8575701870350177e-05,
+      "loss": 1.103,
+      "step": 4430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8574961513329916e-05,
+      "loss": 1.0005,
+      "step": 4431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8574220978700953e-05,
+      "loss": 1.1372,
+      "step": 4432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8573480266478625e-05,
+      "loss": 1.0996,
+      "step": 4433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857273937667827e-05,
+      "loss": 1.2041,
+      "step": 4434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8571998309315245e-05,
+      "loss": 1.0503,
+      "step": 4435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8571257064404894e-05,
+      "loss": 1.2192,
+      "step": 4436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8570515641962563e-05,
+      "loss": 0.9395,
+      "step": 4437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8569774042003614e-05,
+      "loss": 1.1211,
+      "step": 4438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856903226454341e-05,
+      "loss": 1.1562,
+      "step": 4439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856829030959731e-05,
+      "loss": 0.9766,
+      "step": 4440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856754817718069e-05,
+      "loss": 0.9985,
+      "step": 4441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856680586730891e-05,
+      "loss": 1.1558,
+      "step": 4442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8566063379997354e-05,
+      "loss": 1.0308,
+      "step": 4443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85653207152614e-05,
+      "loss": 1.0938,
+      "step": 4444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856457787311642e-05,
+      "loss": 1.0537,
+      "step": 4445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8563834853577815e-05,
+      "loss": 1.0454,
+      "step": 4446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8563091656660965e-05,
+      "loss": 1.0942,
+      "step": 4447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8562348282381267e-05,
+      "loss": 1.0508,
+      "step": 4448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8561604730754114e-05,
+      "loss": 1.1626,
+      "step": 4449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8560861001794914e-05,
+      "loss": 1.1455,
+      "step": 4450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8560117095519067e-05,
+      "loss": 1.0781,
+      "step": 4451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8559373011941982e-05,
+      "loss": 1.0312,
+      "step": 4452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855862875107907e-05,
+      "loss": 1.1377,
+      "step": 4453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855788431294575e-05,
+      "loss": 1.2642,
+      "step": 4454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8557139697557435e-05,
+      "loss": 1.2344,
+      "step": 4455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8556394904929554e-05,
+      "loss": 1.0635,
+      "step": 4456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855564993507753e-05,
+      "loss": 1.1738,
+      "step": 4457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855490478801679e-05,
+      "loss": 1.2305,
+      "step": 4458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8554159463762774e-05,
+      "loss": 1.21,
+      "step": 4459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855341396233092e-05,
+      "loss": 1.0596,
+      "step": 4460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8552668283736663e-05,
+      "loss": 1.1196,
+      "step": 4461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8551922427995453e-05,
+      "loss": 1.0884,
+      "step": 4462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8551176395122734e-05,
+      "loss": 1.208,
+      "step": 4463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855043018513396e-05,
+      "loss": 1.1655,
+      "step": 4464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8549683798044597e-05,
+      "loss": 1.1396,
+      "step": 4465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8548937233870085e-05,
+      "loss": 1.207,
+      "step": 4466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85481904926259e-05,
+      "loss": 1.0771,
+      "step": 4467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8547443574327507e-05,
+      "loss": 1.0879,
+      "step": 4468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854669647899038e-05,
+      "loss": 0.9785,
+      "step": 4469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8545949206629986e-05,
+      "loss": 1.085,
+      "step": 4470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8545201757261803e-05,
+      "loss": 1.0269,
+      "step": 4471
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8544454130901318e-05,
+      "loss": 1.1406,
+      "step": 4472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8543706327564015e-05,
+      "loss": 1.1089,
+      "step": 4473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854295834726538e-05,
+      "loss": 1.1104,
+      "step": 4474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8542210190020906e-05,
+      "loss": 1.0977,
+      "step": 4475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8541461855846093e-05,
+      "loss": 1.1221,
+      "step": 4476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8540713344756435e-05,
+      "loss": 0.9561,
+      "step": 4477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8539964656767443e-05,
+      "loss": 1.1475,
+      "step": 4478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8539215791894617e-05,
+      "loss": 1.2397,
+      "step": 4479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853846675015347e-05,
+      "loss": 1.1616,
+      "step": 4480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8537717531559518e-05,
+      "loss": 1.1504,
+      "step": 4481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8536968136128276e-05,
+      "loss": 1.0146,
+      "step": 4482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853621856387527e-05,
+      "loss": 1.1348,
+      "step": 4483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8535468814816027e-05,
+      "loss": 1.0957,
+      "step": 4484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8534718888966065e-05,
+      "loss": 1.1235,
+      "step": 4485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853396878634093e-05,
+      "loss": 1.1328,
+      "step": 4486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8533218506956154e-05,
+      "loss": 1.1328,
+      "step": 4487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8532468050827274e-05,
+      "loss": 1.0298,
+      "step": 4488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8531717417969838e-05,
+      "loss": 1.1333,
+      "step": 4489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853096660839939e-05,
+      "loss": 1.1646,
+      "step": 4490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8530215622131482e-05,
+      "loss": 1.1636,
+      "step": 4491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852946445918167e-05,
+      "loss": 1.1982,
+      "step": 4492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8528713119565514e-05,
+      "loss": 1.1499,
+      "step": 4493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8527961603298572e-05,
+      "loss": 1.1372,
+      "step": 4494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8527209910396415e-05,
+      "loss": 1.0132,
+      "step": 4495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8526458040874606e-05,
+      "loss": 1.0317,
+      "step": 4496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8525705994748722e-05,
+      "loss": 1.1411,
+      "step": 4497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8524953772034337e-05,
+      "loss": 1.1064,
+      "step": 4498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8524201372747037e-05,
+      "loss": 0.9136,
+      "step": 4499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8523448796902398e-05,
+      "loss": 0.9868,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8522696044516017e-05,
+      "loss": 1.2114,
+      "step": 4501
+    },
+    {
+      "epoch": 1.0,
+      "eval_cIoU": 0.748431921005249,
+      "eval_gIoU": 0.7400373220443726,
+      "eval_runtime": 119.9898,
+      "eval_samples_per_second": 31.761,
+      "eval_steps_per_second": 3.975,
+      "step": 4501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8521943115603477e-05,
+      "loss": 0.9243,
+      "step": 4502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8521190010180378e-05,
+      "loss": 1.0479,
+      "step": 4503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8520436728262318e-05,
+      "loss": 1.103,
+      "step": 4504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.85196832698649e-05,
+      "loss": 0.9854,
+      "step": 4505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8518929635003723e-05,
+      "loss": 1.0449,
+      "step": 4506
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8518175823694408e-05,
+      "loss": 0.8813,
+      "step": 4507
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8517421835952558e-05,
+      "loss": 0.9263,
+      "step": 4508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.85166676717938e-05,
+      "loss": 1.0518,
+      "step": 4509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8515913331233744e-05,
+      "loss": 0.9614,
+      "step": 4510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8515158814288023e-05,
+      "loss": 0.9878,
+      "step": 4511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8514404120972257e-05,
+      "loss": 0.9917,
+      "step": 4512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8513649251302084e-05,
+      "loss": 0.9985,
+      "step": 4513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8512894205293136e-05,
+      "loss": 0.9209,
+      "step": 4514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8512138982961057e-05,
+      "loss": 1.0586,
+      "step": 4515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851138358432148e-05,
+      "loss": 1.0054,
+      "step": 4516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851062800939006e-05,
+      "loss": 0.9785,
+      "step": 4517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8509872258182443e-05,
+      "loss": 0.9946,
+      "step": 4518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8509116330714283e-05,
+      "loss": 0.9507,
+      "step": 4519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8508360227001236e-05,
+      "loss": 0.9414,
+      "step": 4520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8507603947058967e-05,
+      "loss": 0.9712,
+      "step": 4521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8506847490903134e-05,
+      "loss": 1.0093,
+      "step": 4522
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850609085854941e-05,
+      "loss": 0.9648,
+      "step": 4523
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8505334050013464e-05,
+      "loss": 0.9043,
+      "step": 4524
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8504577065310973e-05,
+      "loss": 0.9268,
+      "step": 4525
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8503819904457613e-05,
+      "loss": 0.9868,
+      "step": 4526
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850306256746907e-05,
+      "loss": 0.9907,
+      "step": 4527
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8502305054361034e-05,
+      "loss": 0.9624,
+      "step": 4528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850154736514919e-05,
+      "loss": 0.9321,
+      "step": 4529
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8500789499849227e-05,
+      "loss": 0.8667,
+      "step": 4530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850003145847685e-05,
+      "loss": 0.8877,
+      "step": 4531
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8499273241047756e-05,
+      "loss": 0.8452,
+      "step": 4532
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8498514847577655e-05,
+      "loss": 0.9287,
+      "step": 4533
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849775627808225e-05,
+      "loss": 0.999,
+      "step": 4534
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849699753257725e-05,
+      "loss": 0.9795,
+      "step": 4535
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8496238611078376e-05,
+      "loss": 0.938,
+      "step": 4536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8495479513601347e-05,
+      "loss": 0.9551,
+      "step": 4537
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8494720240161883e-05,
+      "loss": 0.9858,
+      "step": 4538
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8493960790775713e-05,
+      "loss": 0.978,
+      "step": 4539
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8493201165458563e-05,
+      "loss": 0.9565,
+      "step": 4540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8492441364226175e-05,
+      "loss": 0.8564,
+      "step": 4541
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8491681387094276e-05,
+      "loss": 0.9697,
+      "step": 4542
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8490921234078612e-05,
+      "loss": 0.9492,
+      "step": 4543
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849016090519493e-05,
+      "loss": 0.958,
+      "step": 4544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8489400400458976e-05,
+      "loss": 0.9683,
+      "step": 4545
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.84886397198865e-05,
+      "loss": 0.8823,
+      "step": 4546
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8487878863493263e-05,
+      "loss": 0.978,
+      "step": 4547
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8487117831295015e-05,
+      "loss": 1.0059,
+      "step": 4548
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848635662330753e-05,
+      "loss": 1.0825,
+      "step": 4549
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8485595239546567e-05,
+      "loss": 1.0352,
+      "step": 4550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8484833680027896e-05,
+      "loss": 0.8833,
+      "step": 4551
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8484071944767294e-05,
+      "loss": 0.9922,
+      "step": 4552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848331003378054e-05,
+      "loss": 0.9702,
+      "step": 4553
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848254794708341e-05,
+      "loss": 0.8867,
+      "step": 4554
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8481785684691692e-05,
+      "loss": 0.9834,
+      "step": 4555
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8481023246621172e-05,
+      "loss": 1.0107,
+      "step": 4556
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8480260632887647e-05,
+      "loss": 0.9731,
+      "step": 4557
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8479497843506905e-05,
+      "loss": 0.8638,
+      "step": 4558
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8478734878494752e-05,
+      "loss": 1.0254,
+      "step": 4559
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8477971737866988e-05,
+      "loss": 0.9595,
+      "step": 4560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847720842163942e-05,
+      "loss": 0.9346,
+      "step": 4561
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847644492982786e-05,
+      "loss": 0.9365,
+      "step": 4562
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8475681262448112e-05,
+      "loss": 0.9302,
+      "step": 4563
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8474917419516006e-05,
+      "loss": 1.041,
+      "step": 4564
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8474153401047358e-05,
+      "loss": 1.0391,
+      "step": 4565
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8473389207057995e-05,
+      "loss": 0.9683,
+      "step": 4566
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847262483756374e-05,
+      "loss": 0.9888,
+      "step": 4567
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847186029258043e-05,
+      "loss": 1.0513,
+      "step": 4568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8471095572123898e-05,
+      "loss": 1.0669,
+      "step": 4569
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8470330676209987e-05,
+      "loss": 0.9844,
+      "step": 4570
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8469565604854536e-05,
+      "loss": 0.9106,
+      "step": 4571
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8468800358073392e-05,
+      "loss": 1.0356,
+      "step": 4572
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8468034935882407e-05,
+      "loss": 0.939,
+      "step": 4573
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8467269338297428e-05,
+      "loss": 0.9648,
+      "step": 4574
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8466503565334324e-05,
+      "loss": 0.9463,
+      "step": 4575
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8465737617008946e-05,
+      "loss": 0.9751,
+      "step": 4576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8464971493337167e-05,
+      "loss": 1.105,
+      "step": 4577
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846420519433485e-05,
+      "loss": 1.0176,
+      "step": 4578
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8463438720017864e-05,
+      "loss": 1.0679,
+      "step": 4579
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8462672070402094e-05,
+      "loss": 1.0303,
+      "step": 4580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846190524550341e-05,
+      "loss": 0.8813,
+      "step": 4581
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8461138245337702e-05,
+      "loss": 0.9839,
+      "step": 4582
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846037106992085e-05,
+      "loss": 0.895,
+      "step": 4583
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845960371926875e-05,
+      "loss": 0.9355,
+      "step": 4584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8458836193397292e-05,
+      "loss": 1.0176,
+      "step": 4585
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845806849232237e-05,
+      "loss": 1.084,
+      "step": 4586
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8457300616059894e-05,
+      "loss": 1.0034,
+      "step": 4587
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8456532564625765e-05,
+      "loss": 1.0,
+      "step": 4588
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8455764338035886e-05,
+      "loss": 1.0032,
+      "step": 4589
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8454995936306176e-05,
+      "loss": 0.8975,
+      "step": 4590
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8454227359452547e-05,
+      "loss": 0.916,
+      "step": 4591
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845345860749092e-05,
+      "loss": 0.8862,
+      "step": 4592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8452689680437213e-05,
+      "loss": 1.0112,
+      "step": 4593
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8451920578307357e-05,
+      "loss": 0.9443,
+      "step": 4594
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8451151301117282e-05,
+      "loss": 1.0244,
+      "step": 4595
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845038184888292e-05,
+      "loss": 0.9785,
+      "step": 4596
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844961222162021e-05,
+      "loss": 1.0034,
+      "step": 4597
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844884241934509e-05,
+      "loss": 0.9106,
+      "step": 4598
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844807244207351e-05,
+      "loss": 0.9575,
+      "step": 4599
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8447302289821407e-05,
+      "loss": 1.0322,
+      "step": 4600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8446531962604743e-05,
+      "loss": 1.0693,
+      "step": 4601
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8445761460439472e-05,
+      "loss": 1.0063,
+      "step": 4602
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844499078334155e-05,
+      "loss": 0.9805,
+      "step": 4603
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8444219931326946e-05,
+      "loss": 0.938,
+      "step": 4604
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8443448904411615e-05,
+      "loss": 0.9858,
+      "step": 4605
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8442677702611535e-05,
+      "loss": 0.8965,
+      "step": 4606
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844190632594268e-05,
+      "loss": 1.0078,
+      "step": 4607
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8441134774421022e-05,
+      "loss": 0.9451,
+      "step": 4608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8440363048062546e-05,
+      "loss": 0.981,
+      "step": 4609
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8439591146883232e-05,
+      "loss": 0.9531,
+      "step": 4610
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8438819070899076e-05,
+      "loss": 0.98,
+      "step": 4611
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8438046820126063e-05,
+      "loss": 1.0254,
+      "step": 4612
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8437274394580188e-05,
+      "loss": 0.9321,
+      "step": 4613
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843650179427745e-05,
+      "loss": 1.0625,
+      "step": 4614
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8435729019233858e-05,
+      "loss": 1.0474,
+      "step": 4615
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843495606946541e-05,
+      "loss": 0.8921,
+      "step": 4616
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8434182944988118e-05,
+      "loss": 0.9365,
+      "step": 4617
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8433409645818e-05,
+      "loss": 0.9546,
+      "step": 4618
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8432636171971062e-05,
+      "loss": 1.1084,
+      "step": 4619
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843186252346334e-05,
+      "loss": 0.9561,
+      "step": 4620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8431088700310846e-05,
+      "loss": 1.0483,
+      "step": 4621
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843031470252961e-05,
+      "loss": 0.9783,
+      "step": 4622
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8429540530135668e-05,
+      "loss": 0.8818,
+      "step": 4623
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842876618314505e-05,
+      "loss": 0.9585,
+      "step": 4624
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.84279916615738e-05,
+      "loss": 0.9614,
+      "step": 4625
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8427216965437954e-05,
+      "loss": 0.9785,
+      "step": 4626
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8426442094753563e-05,
+      "loss": 0.9453,
+      "step": 4627
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8425667049536674e-05,
+      "loss": 1.0088,
+      "step": 4628
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8424891829803337e-05,
+      "loss": 0.8835,
+      "step": 4629
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8424116435569616e-05,
+      "loss": 0.9399,
+      "step": 4630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8423340866851566e-05,
+      "loss": 1.0259,
+      "step": 4631
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842256512366525e-05,
+      "loss": 1.0444,
+      "step": 4632
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8421789206026745e-05,
+      "loss": 1.0542,
+      "step": 4633
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8421013113952112e-05,
+      "loss": 0.8318,
+      "step": 4634
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8420236847457428e-05,
+      "loss": 0.9482,
+      "step": 4635
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8419460406558772e-05,
+      "loss": 0.9033,
+      "step": 4636
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8418683791272227e-05,
+      "loss": 1.0508,
+      "step": 4637
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8417907001613877e-05,
+      "loss": 1.1123,
+      "step": 4638
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8417130037599812e-05,
+      "loss": 0.9468,
+      "step": 4639
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841635289924613e-05,
+      "loss": 0.9487,
+      "step": 4640
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841557558656892e-05,
+      "loss": 0.9775,
+      "step": 4641
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841479809958428e-05,
+      "loss": 0.9492,
+      "step": 4642
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8414020438308323e-05,
+      "loss": 0.9653,
+      "step": 4643
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8413242602757153e-05,
+      "loss": 0.9214,
+      "step": 4644
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8412464592946877e-05,
+      "loss": 1.1689,
+      "step": 4645
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8411686408893612e-05,
+      "loss": 0.9409,
+      "step": 4646
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841090805061348e-05,
+      "loss": 0.9565,
+      "step": 4647
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8410129518122594e-05,
+      "loss": 0.9458,
+      "step": 4648
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8409350811437086e-05,
+      "loss": 0.9434,
+      "step": 4649
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8408571930573082e-05,
+      "loss": 1.0054,
+      "step": 4650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8407792875546717e-05,
+      "loss": 0.896,
+      "step": 4651
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8407013646374126e-05,
+      "loss": 1.0044,
+      "step": 4652
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840623424307145e-05,
+      "loss": 0.9092,
+      "step": 4653
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840545466565483e-05,
+      "loss": 0.9199,
+      "step": 4654
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8404674914140416e-05,
+      "loss": 1.1284,
+      "step": 4655
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8403894988544352e-05,
+      "loss": 1.0815,
+      "step": 4656
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.84031148888828e-05,
+      "loss": 1.0186,
+      "step": 4657
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8402334615171913e-05,
+      "loss": 0.9878,
+      "step": 4658
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8401554167427854e-05,
+      "loss": 0.9951,
+      "step": 4659
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8400773545666788e-05,
+      "loss": 1.1045,
+      "step": 4660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8399992749904883e-05,
+      "loss": 0.8809,
+      "step": 4661
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8399211780158315e-05,
+      "loss": 1.1289,
+      "step": 4662
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8398430636443255e-05,
+      "loss": 1.0386,
+      "step": 4663
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8397649318775885e-05,
+      "loss": 1.0552,
+      "step": 4664
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8396867827172386e-05,
+      "loss": 0.9104,
+      "step": 4665
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8396086161648947e-05,
+      "loss": 0.8545,
+      "step": 4666
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8395304322221756e-05,
+      "loss": 0.9834,
+      "step": 4667
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839452230890701e-05,
+      "loss": 0.9683,
+      "step": 4668
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.83937401217209e-05,
+      "loss": 1.0161,
+      "step": 4669
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8392957760679633e-05,
+      "loss": 1.0166,
+      "step": 4670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8392175225799412e-05,
+      "loss": 0.9629,
+      "step": 4671
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8391392517096446e-05,
+      "loss": 1.0229,
+      "step": 4672
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8390609634586948e-05,
+      "loss": 1.0337,
+      "step": 4673
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8389826578287126e-05,
+      "loss": 0.9253,
+      "step": 4674
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838904334821321e-05,
+      "loss": 0.8562,
+      "step": 4675
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8388259944381413e-05,
+      "loss": 0.9829,
+      "step": 4676
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8387476366807968e-05,
+      "loss": 0.981,
+      "step": 4677
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8386692615509098e-05,
+      "loss": 1.043,
+      "step": 4678
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8385908690501043e-05,
+      "loss": 0.9707,
+      "step": 4679
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838512459180004e-05,
+      "loss": 0.96,
+      "step": 4680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8384340319422324e-05,
+      "loss": 0.9648,
+      "step": 4681
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8383555873384144e-05,
+      "loss": 0.9824,
+      "step": 4682
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8382771253701747e-05,
+      "loss": 1.0664,
+      "step": 4683
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838198646039138e-05,
+      "loss": 0.9917,
+      "step": 4684
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838120149346931e-05,
+      "loss": 0.9956,
+      "step": 4685
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838041635295178e-05,
+      "loss": 0.9819,
+      "step": 4686
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8379631038855062e-05,
+      "loss": 0.9868,
+      "step": 4687
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837884555119542e-05,
+      "loss": 0.8691,
+      "step": 4688
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8378059889989123e-05,
+      "loss": 1.0508,
+      "step": 4689
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837727405525244e-05,
+      "loss": 1.0063,
+      "step": 4690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837648804700165e-05,
+      "loss": 0.9795,
+      "step": 4691
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8375701865253043e-05,
+      "loss": 1.0078,
+      "step": 4692
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8374915510022888e-05,
+      "loss": 1.0176,
+      "step": 4693
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837412898132748e-05,
+      "loss": 0.9722,
+      "step": 4694
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8373342279183107e-05,
+      "loss": 1.0986,
+      "step": 4695
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837255540360607e-05,
+      "loss": 0.9302,
+      "step": 4696
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837176835461266e-05,
+      "loss": 0.9639,
+      "step": 4697
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837098113221918e-05,
+      "loss": 0.9712,
+      "step": 4698
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8370193736441934e-05,
+      "loss": 0.8564,
+      "step": 4699
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8369406167297237e-05,
+      "loss": 0.9834,
+      "step": 4700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8368618424801397e-05,
+      "loss": 0.8682,
+      "step": 4701
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.836783050897073e-05,
+      "loss": 1.1445,
+      "step": 4702
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8367042419821557e-05,
+      "loss": 0.9229,
+      "step": 4703
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.83662541573702e-05,
+      "loss": 0.9443,
+      "step": 4704
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836546572163299e-05,
+      "loss": 1.0088,
+      "step": 4705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8364677112626252e-05,
+      "loss": 1.001,
+      "step": 4706
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8363888330366322e-05,
+      "loss": 0.8687,
+      "step": 4707
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8363099374869536e-05,
+      "loss": 0.9932,
+      "step": 4708
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836231024615224e-05,
+      "loss": 0.9678,
+      "step": 4709
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8361520944230772e-05,
+      "loss": 0.874,
+      "step": 4710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836073146912149e-05,
+      "loss": 1.0493,
+      "step": 4711
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8359941820840736e-05,
+      "loss": 1.0562,
+      "step": 4712
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8359151999404872e-05,
+      "loss": 0.9478,
+      "step": 4713
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8358362004830256e-05,
+      "loss": 0.9424,
+      "step": 4714
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8357571837133245e-05,
+      "loss": 1.0835,
+      "step": 4715
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8356781496330212e-05,
+      "loss": 0.8984,
+      "step": 4716
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8355990982437528e-05,
+      "loss": 0.9722,
+      "step": 4717
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8355200295471564e-05,
+      "loss": 1.0073,
+      "step": 4718
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835440943544869e-05,
+      "loss": 0.9268,
+      "step": 4719
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8353618402385298e-05,
+      "loss": 0.8701,
+      "step": 4720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835282719629777e-05,
+      "loss": 0.9258,
+      "step": 4721
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835203581720249e-05,
+      "loss": 1.0352,
+      "step": 4722
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8351244265115846e-05,
+      "loss": 0.9702,
+      "step": 4723
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835045254005424e-05,
+      "loss": 0.9927,
+      "step": 4724
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834966064203407e-05,
+      "loss": 1.1025,
+      "step": 4725
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8348868571071736e-05,
+      "loss": 1.0078,
+      "step": 4726
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8348076327183646e-05,
+      "loss": 0.9814,
+      "step": 4727
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8347283910386204e-05,
+      "loss": 0.9268,
+      "step": 4728
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834649132069583e-05,
+      "loss": 1.0645,
+      "step": 4729
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8345698558128932e-05,
+      "loss": 0.9429,
+      "step": 4730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834490562270194e-05,
+      "loss": 0.9443,
+      "step": 4731
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834411251443127e-05,
+      "loss": 0.9897,
+      "step": 4732
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8343319233333355e-05,
+      "loss": 1.147,
+      "step": 4733
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834252577942462e-05,
+      "loss": 1.0957,
+      "step": 4734
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8341732152721502e-05,
+      "loss": 0.9941,
+      "step": 4735
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8340938353240444e-05,
+      "loss": 0.8872,
+      "step": 4736
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834014438099788e-05,
+      "loss": 1.0874,
+      "step": 4737
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8339350236010256e-05,
+      "loss": 0.9121,
+      "step": 4738
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8338555918294023e-05,
+      "loss": 0.9531,
+      "step": 4739
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8337761427865635e-05,
+      "loss": 0.9673,
+      "step": 4740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8336966764741543e-05,
+      "loss": 1.0698,
+      "step": 4741
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833617192893821e-05,
+      "loss": 1.0469,
+      "step": 4742
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8335376920472098e-05,
+      "loss": 0.9707,
+      "step": 4743
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8334581739359678e-05,
+      "loss": 0.915,
+      "step": 4744
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8333786385617413e-05,
+      "loss": 1.0225,
+      "step": 4745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8332990859261777e-05,
+      "loss": 1.04,
+      "step": 4746
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8332195160309252e-05,
+      "loss": 0.9243,
+      "step": 4747
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833139928877632e-05,
+      "loss": 1.0269,
+      "step": 4748
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8330603244679458e-05,
+      "loss": 0.9951,
+      "step": 4749
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8329807028035162e-05,
+      "loss": 0.9966,
+      "step": 4750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8329010638859917e-05,
+      "loss": 0.9536,
+      "step": 4751
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832821407717022e-05,
+      "loss": 0.8789,
+      "step": 4752
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8327417342982573e-05,
+      "loss": 0.9717,
+      "step": 4753
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832662043631348e-05,
+      "loss": 0.9478,
+      "step": 4754
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832582335717944e-05,
+      "loss": 1.002,
+      "step": 4755
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8325026105596964e-05,
+      "loss": 1.0127,
+      "step": 4756
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832422868158257e-05,
+      "loss": 0.9961,
+      "step": 4757
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832343108515277e-05,
+      "loss": 0.9629,
+      "step": 4758
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8322633316324088e-05,
+      "loss": 0.9697,
+      "step": 4759
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832183537511304e-05,
+      "loss": 0.8447,
+      "step": 4760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8321037261536165e-05,
+      "loss": 1.0088,
+      "step": 4761
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8320238975609986e-05,
+      "loss": 1.0967,
+      "step": 4762
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8319440517351037e-05,
+      "loss": 1.0151,
+      "step": 4763
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8318641886775862e-05,
+      "loss": 1.1689,
+      "step": 4764
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8317843083900992e-05,
+      "loss": 1.041,
+      "step": 4765
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8317044108742986e-05,
+      "loss": 0.9209,
+      "step": 4766
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8316244961318383e-05,
+      "loss": 0.8516,
+      "step": 4767
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8315445641643738e-05,
+      "loss": 1.0435,
+      "step": 4768
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8314646149735606e-05,
+      "loss": 1.0405,
+      "step": 4769
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8313846485610548e-05,
+      "loss": 0.9482,
+      "step": 4770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8313046649285128e-05,
+      "loss": 0.9517,
+      "step": 4771
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831224664077591e-05,
+      "loss": 0.9834,
+      "step": 4772
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8311446460099465e-05,
+      "loss": 1.0039,
+      "step": 4773
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831064610727237e-05,
+      "loss": 0.8765,
+      "step": 4774
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8309845582311196e-05,
+      "loss": 0.917,
+      "step": 4775
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8309044885232525e-05,
+      "loss": 0.8975,
+      "step": 4776
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8308244016052947e-05,
+      "loss": 0.9731,
+      "step": 4777
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8307442974789045e-05,
+      "loss": 0.8882,
+      "step": 4778
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8306641761457413e-05,
+      "loss": 0.9141,
+      "step": 4779
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8305840376074642e-05,
+      "loss": 0.9526,
+      "step": 4780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830503881865734e-05,
+      "loss": 0.9502,
+      "step": 4781
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8304237089222098e-05,
+      "loss": 0.9434,
+      "step": 4782
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830343518778553e-05,
+      "loss": 0.9172,
+      "step": 4783
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8302633114364237e-05,
+      "loss": 0.9946,
+      "step": 4784
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8301830868974845e-05,
+      "loss": 1.001,
+      "step": 4785
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830102845163396e-05,
+      "loss": 0.8364,
+      "step": 4786
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8300225862358202e-05,
+      "loss": 1.001,
+      "step": 4787
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8299423101164197e-05,
+      "loss": 0.9761,
+      "step": 4788
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8298620168068578e-05,
+      "loss": 0.897,
+      "step": 4789
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8297817063087965e-05,
+      "loss": 1.0479,
+      "step": 4790
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8297013786239e-05,
+      "loss": 1.0581,
+      "step": 4791
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8296210337538313e-05,
+      "loss": 0.9438,
+      "step": 4792
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829540671700256e-05,
+      "loss": 1.0986,
+      "step": 4793
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.829460292464837e-05,
+      "loss": 1.1064,
+      "step": 4794
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82937989604924e-05,
+      "loss": 0.9556,
+      "step": 4795
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82929948245513e-05,
+      "loss": 0.9878,
+      "step": 4796
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8292190516841727e-05,
+      "loss": 0.8621,
+      "step": 4797
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8291386037380337e-05,
+      "loss": 0.9907,
+      "step": 4798
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82905813861838e-05,
+      "loss": 0.9658,
+      "step": 4799
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8289776563268775e-05,
+      "loss": 1.0225,
+      "step": 4800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828897156865193e-05,
+      "loss": 1.0654,
+      "step": 4801
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8288166402349946e-05,
+      "loss": 1.0366,
+      "step": 4802
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82873610643795e-05,
+      "loss": 0.8975,
+      "step": 4803
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8286555554757265e-05,
+      "loss": 1.0215,
+      "step": 4804
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8285749873499932e-05,
+      "loss": 1.0581,
+      "step": 4805
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8284944020624184e-05,
+      "loss": 0.9961,
+      "step": 4806
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8284137996146712e-05,
+      "loss": 0.958,
+      "step": 4807
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8283331800084218e-05,
+      "loss": 0.9302,
+      "step": 4808
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828252543245339e-05,
+      "loss": 0.9717,
+      "step": 4809
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828171889327094e-05,
+      "loss": 0.9702,
+      "step": 4810
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8280912182553567e-05,
+      "loss": 0.8584,
+      "step": 4811
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828010530031798e-05,
+      "loss": 0.9731,
+      "step": 4812
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8279298246580894e-05,
+      "loss": 0.9263,
+      "step": 4813
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8278491021359023e-05,
+      "loss": 0.9824,
+      "step": 4814
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827768362466909e-05,
+      "loss": 0.9521,
+      "step": 4815
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8276876056527813e-05,
+      "loss": 1.0186,
+      "step": 4816
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827606831695192e-05,
+      "loss": 1.0142,
+      "step": 4817
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827526040595815e-05,
+      "loss": 0.9421,
+      "step": 4818
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8274452323563226e-05,
+      "loss": 0.9521,
+      "step": 4819
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8273644069783888e-05,
+      "loss": 1.106,
+      "step": 4820
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827283564463688e-05,
+      "loss": 0.9429,
+      "step": 4821
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8272027048138943e-05,
+      "loss": 1.1182,
+      "step": 4822
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827121828030683e-05,
+      "loss": 0.8882,
+      "step": 4823
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8270409341157283e-05,
+      "loss": 1.0425,
+      "step": 4824
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826960023070707e-05,
+      "loss": 0.9277,
+      "step": 4825
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8268790948972942e-05,
+      "loss": 0.9717,
+      "step": 4826
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826798149597166e-05,
+      "loss": 1.0854,
+      "step": 4827
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826717187171999e-05,
+      "loss": 1.0068,
+      "step": 4828
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8266362076234708e-05,
+      "loss": 1.1108,
+      "step": 4829
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8265552109532577e-05,
+      "loss": 0.9536,
+      "step": 4830
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8264741971630383e-05,
+      "loss": 0.8984,
+      "step": 4831
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8263931662544904e-05,
+      "loss": 0.9771,
+      "step": 4832
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8263121182292917e-05,
+      "loss": 1.0059,
+      "step": 4833
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8262310530891212e-05,
+      "loss": 1.0171,
+      "step": 4834
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8261499708356583e-05,
+      "loss": 1.062,
+      "step": 4835
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826068871470582e-05,
+      "loss": 0.9614,
+      "step": 4836
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8259877549955726e-05,
+      "loss": 0.877,
+      "step": 4837
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8259066214123095e-05,
+      "loss": 1.0122,
+      "step": 4838
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825825470722474e-05,
+      "loss": 0.9644,
+      "step": 4839
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8257443029277463e-05,
+      "loss": 0.9502,
+      "step": 4840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825663118029808e-05,
+      "loss": 0.9595,
+      "step": 4841
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82558191603034e-05,
+      "loss": 1.0952,
+      "step": 4842
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8255006969310248e-05,
+      "loss": 1.0605,
+      "step": 4843
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8254194607335446e-05,
+      "loss": 1.0063,
+      "step": 4844
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8253382074395816e-05,
+      "loss": 0.9727,
+      "step": 4845
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8252569370508194e-05,
+      "loss": 0.998,
+      "step": 4846
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8251756495689403e-05,
+      "loss": 1.0576,
+      "step": 4847
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825094344995629e-05,
+      "loss": 0.9365,
+      "step": 4848
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8250130233325693e-05,
+      "loss": 0.9165,
+      "step": 4849
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8249316845814452e-05,
+      "loss": 0.9077,
+      "step": 4850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8248503287439416e-05,
+      "loss": 0.8564,
+      "step": 4851
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8247689558217438e-05,
+      "loss": 1.0703,
+      "step": 4852
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8246875658165367e-05,
+      "loss": 0.9658,
+      "step": 4853
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8246061587300067e-05,
+      "loss": 1.0293,
+      "step": 4854
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8245247345638396e-05,
+      "loss": 0.9575,
+      "step": 4855
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824443293319722e-05,
+      "loss": 0.9297,
+      "step": 4856
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8243618349993406e-05,
+      "loss": 0.9436,
+      "step": 4857
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8242803596043826e-05,
+      "loss": 0.937,
+      "step": 4858
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824198867136536e-05,
+      "loss": 0.9863,
+      "step": 4859
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8241173575974883e-05,
+      "loss": 0.9517,
+      "step": 4860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8240358309889278e-05,
+      "loss": 0.9177,
+      "step": 4861
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8239542873125433e-05,
+      "loss": 0.9995,
+      "step": 4862
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8238727265700235e-05,
+      "loss": 1.0259,
+      "step": 4863
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823791148763058e-05,
+      "loss": 1.0366,
+      "step": 4864
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8237095538933364e-05,
+      "loss": 0.9365,
+      "step": 4865
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823627941962548e-05,
+      "loss": 1.0029,
+      "step": 4866
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823546312972385e-05,
+      "loss": 0.9902,
+      "step": 4867
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8234646669245362e-05,
+      "loss": 0.9521,
+      "step": 4868
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8233830038206936e-05,
+      "loss": 0.9434,
+      "step": 4869
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8233013236625486e-05,
+      "loss": 1.0557,
+      "step": 4870
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823219626451793e-05,
+      "loss": 0.9287,
+      "step": 4871
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8231379121901186e-05,
+      "loss": 0.9844,
+      "step": 4872
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8230561808792186e-05,
+      "loss": 0.9319,
+      "step": 4873
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8229744325207854e-05,
+      "loss": 1.0791,
+      "step": 4874
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8228926671165118e-05,
+      "loss": 0.9092,
+      "step": 4875
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822810884668092e-05,
+      "loss": 0.8721,
+      "step": 4876
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8227290851772205e-05,
+      "loss": 1.0742,
+      "step": 4877
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82264726864559e-05,
+      "loss": 1.0,
+      "step": 4878
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8225654350748962e-05,
+      "loss": 1.0322,
+      "step": 4879
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8224835844668337e-05,
+      "loss": 1.062,
+      "step": 4880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8224017168230982e-05,
+      "loss": 0.825,
+      "step": 4881
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822319832145385e-05,
+      "loss": 1.1001,
+      "step": 4882
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82223793043539e-05,
+      "loss": 1.0947,
+      "step": 4883
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8221560116948103e-05,
+      "loss": 1.0059,
+      "step": 4884
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8220740759253418e-05,
+      "loss": 0.9014,
+      "step": 4885
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8219921231286824e-05,
+      "loss": 0.9756,
+      "step": 4886
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8219101533065284e-05,
+      "loss": 0.8604,
+      "step": 4887
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8218281664605787e-05,
+      "loss": 0.894,
+      "step": 4888
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8217461625925313e-05,
+      "loss": 0.916,
+      "step": 4889
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8216641417040842e-05,
+      "loss": 0.9785,
+      "step": 4890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8215821037969367e-05,
+      "loss": 0.9819,
+      "step": 4891
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821500048872788e-05,
+      "loss": 1.0239,
+      "step": 4892
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8214179769333373e-05,
+      "loss": 0.9507,
+      "step": 4893
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821335887980285e-05,
+      "loss": 0.9878,
+      "step": 4894
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8212537820153305e-05,
+      "loss": 1.0537,
+      "step": 4895
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8211716590401752e-05,
+      "loss": 0.9238,
+      "step": 4896
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.82108951905652e-05,
+      "loss": 1.0508,
+      "step": 4897
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821007362066066e-05,
+      "loss": 0.9995,
+      "step": 4898
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820925188070515e-05,
+      "loss": 1.0146,
+      "step": 4899
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8208429970715695e-05,
+      "loss": 1.0029,
+      "step": 4900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8207607890709304e-05,
+      "loss": 1.0381,
+      "step": 4901
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820678564070302e-05,
+      "loss": 0.9072,
+      "step": 4902
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8205963220713866e-05,
+      "loss": 0.9248,
+      "step": 4903
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820514063075888e-05,
+      "loss": 1.063,
+      "step": 4904
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.82043178708551e-05,
+      "loss": 0.8994,
+      "step": 4905
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820349494101956e-05,
+      "loss": 1.1089,
+      "step": 4906
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8202671841269313e-05,
+      "loss": 1.0625,
+      "step": 4907
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8201848571621405e-05,
+      "loss": 0.9575,
+      "step": 4908
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820102513209289e-05,
+      "loss": 0.981,
+      "step": 4909
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8200201522700818e-05,
+      "loss": 1.0054,
+      "step": 4910
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8199377743462252e-05,
+      "loss": 0.894,
+      "step": 4911
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8198553794394254e-05,
+      "loss": 0.939,
+      "step": 4912
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8197729675513888e-05,
+      "loss": 0.9878,
+      "step": 4913
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819690538683823e-05,
+      "loss": 1.061,
+      "step": 4914
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8196080928384345e-05,
+      "loss": 0.9521,
+      "step": 4915
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819525630016931e-05,
+      "loss": 1.0391,
+      "step": 4916
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8194431502210217e-05,
+      "loss": 1.1152,
+      "step": 4917
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8193606534524134e-05,
+      "loss": 0.9976,
+      "step": 4918
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8192781397128156e-05,
+      "loss": 1.001,
+      "step": 4919
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8191956090039373e-05,
+      "loss": 0.9658,
+      "step": 4920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8191130613274878e-05,
+      "loss": 0.9133,
+      "step": 4921
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819030496685177e-05,
+      "loss": 0.9961,
+      "step": 4922
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818947915078715e-05,
+      "loss": 1.0054,
+      "step": 4923
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818865316509812e-05,
+      "loss": 0.9917,
+      "step": 4924
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8187827009801794e-05,
+      "loss": 0.9185,
+      "step": 4925
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818700068491528e-05,
+      "loss": 0.918,
+      "step": 4926
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8186174190455688e-05,
+      "loss": 0.8792,
+      "step": 4927
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8185347526440147e-05,
+      "loss": 0.9546,
+      "step": 4928
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818452069288577e-05,
+      "loss": 0.9897,
+      "step": 4929
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818369368980969e-05,
+      "loss": 1.0122,
+      "step": 4930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8182866517229036e-05,
+      "loss": 1.0352,
+      "step": 4931
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8182039175160935e-05,
+      "loss": 0.9248,
+      "step": 4932
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8181211663622526e-05,
+      "loss": 1.0527,
+      "step": 4933
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818038398263095e-05,
+      "loss": 1.082,
+      "step": 4934
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817955613220335e-05,
+      "loss": 0.9404,
+      "step": 4935
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8178728112356873e-05,
+      "loss": 0.9429,
+      "step": 4936
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8177899923108667e-05,
+      "loss": 0.9673,
+      "step": 4937
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817707156447589e-05,
+      "loss": 0.8882,
+      "step": 4938
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8176243036475697e-05,
+      "loss": 0.9946,
+      "step": 4939
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8175414339125247e-05,
+      "loss": 1.0137,
+      "step": 4940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817458547244171e-05,
+      "loss": 0.9429,
+      "step": 4941
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8173756436442246e-05,
+      "loss": 0.9355,
+      "step": 4942
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8172927231144033e-05,
+      "loss": 0.9517,
+      "step": 4943
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817209785656424e-05,
+      "loss": 1.0566,
+      "step": 4944
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817126831272005e-05,
+      "loss": 0.9614,
+      "step": 4945
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8170438599628645e-05,
+      "loss": 0.8682,
+      "step": 4946
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816960871730721e-05,
+      "loss": 0.9224,
+      "step": 4947
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8168778665772936e-05,
+      "loss": 0.9561,
+      "step": 4948
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8167948445043007e-05,
+      "loss": 1.0288,
+      "step": 4949
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8167118055134625e-05,
+      "loss": 0.98,
+      "step": 4950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8166287496064988e-05,
+      "loss": 1.019,
+      "step": 4951
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8165456767851305e-05,
+      "loss": 0.936,
+      "step": 4952
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8164625870510773e-05,
+      "loss": 1.0269,
+      "step": 4953
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816379480406061e-05,
+      "loss": 0.9102,
+      "step": 4954
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816296356851802e-05,
+      "loss": 1.0371,
+      "step": 4955
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816213216390023e-05,
+      "loss": 1.0049,
+      "step": 4956
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8161300590224455e-05,
+      "loss": 0.8726,
+      "step": 4957
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816046884750792e-05,
+      "loss": 0.9351,
+      "step": 4958
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815963693576785e-05,
+      "loss": 1.0591,
+      "step": 4959
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815880485502148e-05,
+      "loss": 1.0625,
+      "step": 4960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8157972605286046e-05,
+      "loss": 0.9595,
+      "step": 4961
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815714018657878e-05,
+      "loss": 1.0425,
+      "step": 4962
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8156307598916926e-05,
+      "loss": 1.0112,
+      "step": 4963
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815547484231773e-05,
+      "loss": 0.978,
+      "step": 4964
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8154641916798436e-05,
+      "loss": 1.083,
+      "step": 4965
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8153808822376303e-05,
+      "loss": 0.9224,
+      "step": 4966
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8152975559068585e-05,
+      "loss": 0.9351,
+      "step": 4967
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815214212689254e-05,
+      "loss": 0.9312,
+      "step": 4968
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815130852586542e-05,
+      "loss": 0.897,
+      "step": 4969
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815047475600451e-05,
+      "loss": 0.9717,
+      "step": 4970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8149640817327067e-05,
+      "loss": 0.9385,
+      "step": 4971
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8148806709850365e-05,
+      "loss": 0.938,
+      "step": 4972
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8147972433591686e-05,
+      "loss": 0.9321,
+      "step": 4973
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8147137988568303e-05,
+      "loss": 0.978,
+      "step": 4974
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8146303374797504e-05,
+      "loss": 0.9976,
+      "step": 4975
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8145468592296574e-05,
+      "loss": 0.9951,
+      "step": 4976
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8144633641082805e-05,
+      "loss": 0.9834,
+      "step": 4977
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8143798521173486e-05,
+      "loss": 0.9175,
+      "step": 4978
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814296323258592e-05,
+      "loss": 0.9639,
+      "step": 4979
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814212777533741e-05,
+      "loss": 1.104,
+      "step": 4980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8141292149445252e-05,
+      "loss": 0.98,
+      "step": 4981
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814045635492676e-05,
+      "loss": 1.0098,
+      "step": 4982
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8139620391799243e-05,
+      "loss": 0.9546,
+      "step": 4983
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813878426008002e-05,
+      "loss": 1.0073,
+      "step": 4984
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8137947959786402e-05,
+      "loss": 0.9253,
+      "step": 4985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8137111490935715e-05,
+      "loss": 0.9526,
+      "step": 4986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8136274853545286e-05,
+      "loss": 0.8662,
+      "step": 4987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8135438047632444e-05,
+      "loss": 1.0088,
+      "step": 4988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8134601073214516e-05,
+      "loss": 0.8872,
+      "step": 4989
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813376393030884e-05,
+      "loss": 1.0903,
+      "step": 4990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813292661893276e-05,
+      "loss": 0.9507,
+      "step": 4991
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8132089139103612e-05,
+      "loss": 1.0576,
+      "step": 4992
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8131251490838747e-05,
+      "loss": 1.0864,
+      "step": 4993
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8130413674155512e-05,
+      "loss": 1.0034,
+      "step": 4994
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8129575689071265e-05,
+      "loss": 1.0591,
+      "step": 4995
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8128737535603357e-05,
+      "loss": 0.9375,
+      "step": 4996
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812789921376915e-05,
+      "loss": 0.9551,
+      "step": 4997
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8127060723586008e-05,
+      "loss": 1.0229,
+      "step": 4998
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8126222065071303e-05,
+      "loss": 0.9624,
+      "step": 4999
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8125383238242397e-05,
+      "loss": 0.9541,
+      "step": 5000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812454424311667e-05,
+      "loss": 1.0317,
+      "step": 5001
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8123705079711497e-05,
+      "loss": 1.0127,
+      "step": 5002
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812286574804426e-05,
+      "loss": 0.915,
+      "step": 5003
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8122026248132344e-05,
+      "loss": 0.9561,
+      "step": 5004
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8121186579993136e-05,
+      "loss": 0.9619,
+      "step": 5005
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812034674364403e-05,
+      "loss": 0.9316,
+      "step": 5006
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8119506739102418e-05,
+      "loss": 0.9438,
+      "step": 5007
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8118666566385704e-05,
+      "loss": 0.9683,
+      "step": 5008
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8117826225511283e-05,
+      "loss": 1.0137,
+      "step": 5009
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8116985716496566e-05,
+      "loss": 0.9272,
+      "step": 5010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8116145039358955e-05,
+      "loss": 0.9165,
+      "step": 5011
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8115304194115872e-05,
+      "loss": 0.915,
+      "step": 5012
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811446318078473e-05,
+      "loss": 0.9624,
+      "step": 5013
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8113621999382942e-05,
+      "loss": 1.0435,
+      "step": 5014
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8112780649927936e-05,
+      "loss": 1.0435,
+      "step": 5015
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811193913243714e-05,
+      "loss": 1.0176,
+      "step": 5016
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8111097446927982e-05,
+      "loss": 0.9673,
+      "step": 5017
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.81102555934179e-05,
+      "loss": 0.9585,
+      "step": 5018
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810941357192432e-05,
+      "loss": 0.9102,
+      "step": 5019
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8108571382464692e-05,
+      "loss": 0.9819,
+      "step": 5020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8107729025056453e-05,
+      "loss": 0.9463,
+      "step": 5021
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8106886499717058e-05,
+      "loss": 0.9531,
+      "step": 5022
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8106043806463952e-05,
+      "loss": 0.8794,
+      "step": 5023
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810520094531459e-05,
+      "loss": 0.9229,
+      "step": 5024
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810435791628643e-05,
+      "loss": 1.1079,
+      "step": 5025
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8103514719396936e-05,
+      "loss": 1.043,
+      "step": 5026
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810267135466357e-05,
+      "loss": 0.9902,
+      "step": 5027
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.81018278221038e-05,
+      "loss": 0.9932,
+      "step": 5028
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.81009841217351e-05,
+      "loss": 1.0869,
+      "step": 5029
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810014025357494e-05,
+      "loss": 0.9702,
+      "step": 5030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8099296217640806e-05,
+      "loss": 0.9238,
+      "step": 5031
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8098452013950174e-05,
+      "loss": 0.9382,
+      "step": 5032
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8097607642520535e-05,
+      "loss": 1.0308,
+      "step": 5033
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809676310336937e-05,
+      "loss": 0.9512,
+      "step": 5034
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809591839651418e-05,
+      "loss": 1.0039,
+      "step": 5035
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8095073521972458e-05,
+      "loss": 0.9106,
+      "step": 5036
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.80942284797617e-05,
+      "loss": 1.0195,
+      "step": 5037
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809338326989941e-05,
+      "loss": 1.0352,
+      "step": 5038
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.80925378924031e-05,
+      "loss": 0.9531,
+      "step": 5039
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8091692347290274e-05,
+      "loss": 0.9438,
+      "step": 5040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8090846634578448e-05,
+      "loss": 0.9404,
+      "step": 5041
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8090000754285135e-05,
+      "loss": 0.981,
+      "step": 5042
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808915470642786e-05,
+      "loss": 1.0225,
+      "step": 5043
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8088308491024145e-05,
+      "loss": 1.0425,
+      "step": 5044
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808746210809152e-05,
+      "loss": 0.9429,
+      "step": 5045
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808661555764751e-05,
+      "loss": 0.9097,
+      "step": 5046
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8085768839709654e-05,
+      "loss": 0.9819,
+      "step": 5047
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8084921954295485e-05,
+      "loss": 0.9526,
+      "step": 5048
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808407490142255e-05,
+      "loss": 1.0576,
+      "step": 5049
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8083227681108392e-05,
+      "loss": 0.9521,
+      "step": 5050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8082380293370553e-05,
+      "loss": 0.9424,
+      "step": 5051
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808153273822659e-05,
+      "loss": 0.9194,
+      "step": 5052
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8080685015694054e-05,
+      "loss": 0.9419,
+      "step": 5053
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8079837125790512e-05,
+      "loss": 1.041,
+      "step": 5054
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8078989068533518e-05,
+      "loss": 0.9722,
+      "step": 5055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807814084394064e-05,
+      "loss": 0.7549,
+      "step": 5056
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8077292452029444e-05,
+      "loss": 0.895,
+      "step": 5057
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8076443892817507e-05,
+      "loss": 0.8745,
+      "step": 5058
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.80755951663224e-05,
+      "loss": 1.0591,
+      "step": 5059
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8074746272561708e-05,
+      "loss": 0.979,
+      "step": 5060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807389721155301e-05,
+      "loss": 1.0884,
+      "step": 5061
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807304798331389e-05,
+      "loss": 0.9146,
+      "step": 5062
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8072198587861944e-05,
+      "loss": 0.9004,
+      "step": 5063
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8071349025214756e-05,
+      "loss": 1.0078,
+      "step": 5064
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8070499295389933e-05,
+      "loss": 0.9946,
+      "step": 5065
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806964939840507e-05,
+      "loss": 0.8726,
+      "step": 5066
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8068799334277764e-05,
+      "loss": 0.9702,
+      "step": 5067
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806794910302563e-05,
+      "loss": 0.8687,
+      "step": 5068
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8067098704666282e-05,
+      "loss": 0.9614,
+      "step": 5069
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8066248139217327e-05,
+      "loss": 0.925,
+      "step": 5070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806539740669638e-05,
+      "loss": 0.8892,
+      "step": 5071
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8064546507121067e-05,
+      "loss": 0.8765,
+      "step": 5072
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8063695440509013e-05,
+      "loss": 1.1143,
+      "step": 5073
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806284420687784e-05,
+      "loss": 0.916,
+      "step": 5074
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8061992806245186e-05,
+      "loss": 1.1514,
+      "step": 5075
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806114123862868e-05,
+      "loss": 0.9146,
+      "step": 5076
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8060289504045963e-05,
+      "loss": 0.9604,
+      "step": 5077
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8059437602514678e-05,
+      "loss": 1.0288,
+      "step": 5078
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8058585534052467e-05,
+      "loss": 0.999,
+      "step": 5079
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8057733298676975e-05,
+      "loss": 0.9683,
+      "step": 5080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8056880896405863e-05,
+      "loss": 0.8472,
+      "step": 5081
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8056028327256784e-05,
+      "loss": 0.9558,
+      "step": 5082
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805517559124739e-05,
+      "loss": 1.064,
+      "step": 5083
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8054322688395346e-05,
+      "loss": 1.0181,
+      "step": 5084
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805346961871832e-05,
+      "loss": 0.8979,
+      "step": 5085
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8052616382233986e-05,
+      "loss": 0.9927,
+      "step": 5086
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8051762978960008e-05,
+      "loss": 0.9766,
+      "step": 5087
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8050909408914064e-05,
+      "loss": 0.9114,
+      "step": 5088
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805005567211384e-05,
+      "loss": 0.9248,
+      "step": 5089
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8049201768577008e-05,
+      "loss": 0.9639,
+      "step": 5090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8048347698321263e-05,
+      "loss": 1.0107,
+      "step": 5091
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804749346136429e-05,
+      "loss": 0.9497,
+      "step": 5092
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804663905772379e-05,
+      "loss": 1.0332,
+      "step": 5093
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804578448741745e-05,
+      "loss": 0.9795,
+      "step": 5094
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8044929750462975e-05,
+      "loss": 0.9873,
+      "step": 5095
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8044074846878067e-05,
+      "loss": 1.0688,
+      "step": 5096
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804321977668044e-05,
+      "loss": 0.8867,
+      "step": 5097
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8042364539887796e-05,
+      "loss": 0.8721,
+      "step": 5098
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804150913651785e-05,
+      "loss": 0.9443,
+      "step": 5099
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8040653566588324e-05,
+      "loss": 1.1108,
+      "step": 5100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803979783011694e-05,
+      "loss": 1.0723,
+      "step": 5101
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8038941927121417e-05,
+      "loss": 0.8916,
+      "step": 5102
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8038085857619485e-05,
+      "loss": 0.9609,
+      "step": 5103
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8037229621628874e-05,
+      "loss": 0.936,
+      "step": 5104
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803637321916732e-05,
+      "loss": 1.125,
+      "step": 5105
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803551665025256e-05,
+      "loss": 1.0635,
+      "step": 5106
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8034659914902343e-05,
+      "loss": 0.9775,
+      "step": 5107
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8033803013134406e-05,
+      "loss": 0.9551,
+      "step": 5108
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8032945944966498e-05,
+      "loss": 1.0391,
+      "step": 5109
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8032088710416372e-05,
+      "loss": 1.0156,
+      "step": 5110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8031231309501786e-05,
+      "loss": 0.9473,
+      "step": 5111
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.80303737422405e-05,
+      "loss": 0.9873,
+      "step": 5112
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802951600865027e-05,
+      "loss": 1.0591,
+      "step": 5113
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8028658108748863e-05,
+      "loss": 0.9517,
+      "step": 5114
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8027800042554055e-05,
+      "loss": 0.9707,
+      "step": 5115
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8026941810083612e-05,
+      "loss": 1.085,
+      "step": 5116
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8026083411355314e-05,
+      "loss": 1.1279,
+      "step": 5117
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8025224846386936e-05,
+      "loss": 1.0288,
+      "step": 5118
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8024366115196264e-05,
+      "loss": 0.9785,
+      "step": 5119
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8023507217801085e-05,
+      "loss": 0.9041,
+      "step": 5120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802264815421919e-05,
+      "loss": 1.0244,
+      "step": 5121
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8021788924468372e-05,
+      "loss": 0.9517,
+      "step": 5122
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8020929528566423e-05,
+      "loss": 0.8882,
+      "step": 5123
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8020069966531145e-05,
+      "loss": 0.9375,
+      "step": 5124
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8019210238380346e-05,
+      "loss": 1.0024,
+      "step": 5125
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801835034413183e-05,
+      "loss": 0.9248,
+      "step": 5126
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8017490283803408e-05,
+      "loss": 1.0752,
+      "step": 5127
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8016630057412894e-05,
+      "loss": 0.8486,
+      "step": 5128
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8015769664978102e-05,
+      "loss": 0.9019,
+      "step": 5129
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801490910651686e-05,
+      "loss": 0.8799,
+      "step": 5130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8014048382046986e-05,
+      "loss": 1.0229,
+      "step": 5131
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801318749158631e-05,
+      "loss": 1.0381,
+      "step": 5132
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8012326435152668e-05,
+      "loss": 1.0308,
+      "step": 5133
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8011465212763885e-05,
+      "loss": 0.8931,
+      "step": 5134
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8010603824437806e-05,
+      "loss": 0.918,
+      "step": 5135
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800974227019227e-05,
+      "loss": 0.9951,
+      "step": 5136
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8008880550045122e-05,
+      "loss": 0.9478,
+      "step": 5137
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8008018664014213e-05,
+      "loss": 0.8828,
+      "step": 5138
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8007156612117388e-05,
+      "loss": 0.9595,
+      "step": 5139
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800629439437251e-05,
+      "loss": 0.8613,
+      "step": 5140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8005432010797434e-05,
+      "loss": 0.9326,
+      "step": 5141
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8004569461410023e-05,
+      "loss": 0.9512,
+      "step": 5142
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800370674622814e-05,
+      "loss": 1.022,
+      "step": 5143
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8002843865269655e-05,
+      "loss": 1.0542,
+      "step": 5144
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8001980818552447e-05,
+      "loss": 0.9888,
+      "step": 5145
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8001117606094384e-05,
+      "loss": 0.9595,
+      "step": 5146
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8000254227913346e-05,
+      "loss": 1.0312,
+      "step": 5147
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799939068402722e-05,
+      "loss": 1.0186,
+      "step": 5148
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7998526974453887e-05,
+      "loss": 0.916,
+      "step": 5149
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799766309921124e-05,
+      "loss": 0.958,
+      "step": 5150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7996799058317166e-05,
+      "loss": 0.9976,
+      "step": 5151
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7995934851789574e-05,
+      "loss": 1.082,
+      "step": 5152
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7995070479646353e-05,
+      "loss": 0.9033,
+      "step": 5153
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.799420594190541e-05,
+      "loss": 1.0034,
+      "step": 5154
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.799334123858465e-05,
+      "loss": 1.04,
+      "step": 5155
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7992476369701982e-05,
+      "loss": 0.9785,
+      "step": 5156
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7991611335275328e-05,
+      "loss": 0.9863,
+      "step": 5157
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7990746135322593e-05,
+      "loss": 1.001,
+      "step": 5158
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798988076986171e-05,
+      "loss": 0.9478,
+      "step": 5159
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798901523891059e-05,
+      "loss": 0.8296,
+      "step": 5160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7988149542487173e-05,
+      "loss": 1.0757,
+      "step": 5161
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7987283680609377e-05,
+      "loss": 0.9814,
+      "step": 5162
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7986417653295144e-05,
+      "loss": 1.0269,
+      "step": 5163
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7985551460562414e-05,
+      "loss": 0.9766,
+      "step": 5164
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798468510242912e-05,
+      "loss": 0.9868,
+      "step": 5165
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798381857891321e-05,
+      "loss": 1.0293,
+      "step": 5166
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7982951890032636e-05,
+      "loss": 1.0381,
+      "step": 5167
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798208503580534e-05,
+      "loss": 0.9487,
+      "step": 5168
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7981218016249286e-05,
+      "loss": 0.9707,
+      "step": 5169
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798035083138243e-05,
+      "loss": 0.9465,
+      "step": 5170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7979483481222727e-05,
+      "loss": 1.1099,
+      "step": 5171
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797861596578815e-05,
+      "loss": 0.9702,
+      "step": 5172
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7977748285096667e-05,
+      "loss": 0.8914,
+      "step": 5173
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7976880439166243e-05,
+      "loss": 0.9287,
+      "step": 5174
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797601242801486e-05,
+      "loss": 1.0845,
+      "step": 5175
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797514425166049e-05,
+      "loss": 1.0361,
+      "step": 5176
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7974275910121123e-05,
+      "loss": 1.0112,
+      "step": 5177
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7973407403414737e-05,
+      "loss": 0.8965,
+      "step": 5178
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797253873155933e-05,
+      "loss": 0.9658,
+      "step": 5179
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7971669894572883e-05,
+      "loss": 0.936,
+      "step": 5180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7970800892473403e-05,
+      "loss": 0.9165,
+      "step": 5181
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7969931725278877e-05,
+      "loss": 1.0146,
+      "step": 5182
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796906239300732e-05,
+      "loss": 1.0532,
+      "step": 5183
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796819289567673e-05,
+      "loss": 0.9326,
+      "step": 5184
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7967323233305118e-05,
+      "loss": 0.9702,
+      "step": 5185
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.79664534059105e-05,
+      "loss": 0.8257,
+      "step": 5186
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796558341351089e-05,
+      "loss": 0.9753,
+      "step": 5187
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7964713256124308e-05,
+      "loss": 0.9878,
+      "step": 5188
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7963842933768776e-05,
+      "loss": 1.0718,
+      "step": 5189
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7962972446462322e-05,
+      "loss": 1.0298,
+      "step": 5190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7962101794222975e-05,
+      "loss": 1.0303,
+      "step": 5191
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796123097706877e-05,
+      "loss": 0.9785,
+      "step": 5192
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7960359995017738e-05,
+      "loss": 0.7866,
+      "step": 5193
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7959488848087927e-05,
+      "loss": 0.978,
+      "step": 5194
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7958617536297377e-05,
+      "loss": 0.9624,
+      "step": 5195
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7957746059664135e-05,
+      "loss": 0.8452,
+      "step": 5196
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.795687441820625e-05,
+      "loss": 1.1084,
+      "step": 5197
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7956002611941783e-05,
+      "loss": 0.957,
+      "step": 5198
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.795513064088878e-05,
+      "loss": 1.002,
+      "step": 5199
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7954258505065313e-05,
+      "loss": 1.0005,
+      "step": 5200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7953386204489435e-05,
+      "loss": 1.0303,
+      "step": 5201
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7952513739179226e-05,
+      "loss": 0.9233,
+      "step": 5202
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7951641109152743e-05,
+      "loss": 0.9727,
+      "step": 5203
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.795076831442807e-05,
+      "loss": 0.9663,
+      "step": 5204
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7949895355023284e-05,
+      "loss": 0.8906,
+      "step": 5205
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794902223095646e-05,
+      "loss": 0.9937,
+      "step": 5206
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7948148942245695e-05,
+      "loss": 0.9307,
+      "step": 5207
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7947275488909062e-05,
+      "loss": 0.9473,
+      "step": 5208
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7946401870964658e-05,
+      "loss": 1.0273,
+      "step": 5209
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7945528088430585e-05,
+      "loss": 1.0845,
+      "step": 5210
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7944654141324932e-05,
+      "loss": 0.9053,
+      "step": 5211
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7943780029665802e-05,
+      "loss": 0.8911,
+      "step": 5212
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.79429057534713e-05,
+      "loss": 0.894,
+      "step": 5213
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7942031312759543e-05,
+      "loss": 0.9946,
+      "step": 5214
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7941156707548633e-05,
+      "loss": 1.002,
+      "step": 5215
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7940281937856688e-05,
+      "loss": 1.0098,
+      "step": 5216
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793940700370183e-05,
+      "loss": 0.9775,
+      "step": 5217
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7938531905102174e-05,
+      "loss": 1.0479,
+      "step": 5218
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7937656642075852e-05,
+      "loss": 1.0376,
+      "step": 5219
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7936781214640987e-05,
+      "loss": 0.9917,
+      "step": 5220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793590562281572e-05,
+      "loss": 1.0347,
+      "step": 5221
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7935029866618178e-05,
+      "loss": 0.8105,
+      "step": 5222
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7934153946066505e-05,
+      "loss": 0.8799,
+      "step": 5223
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7933277861178838e-05,
+      "loss": 0.8618,
+      "step": 5224
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793240161197333e-05,
+      "loss": 0.8906,
+      "step": 5225
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7931525198468125e-05,
+      "loss": 1.0664,
+      "step": 5226
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7930648620681382e-05,
+      "loss": 0.9443,
+      "step": 5227
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792977187863125e-05,
+      "loss": 0.9595,
+      "step": 5228
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7928894972335894e-05,
+      "loss": 0.8647,
+      "step": 5229
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792801790181347e-05,
+      "loss": 0.9292,
+      "step": 5230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792714066708215e-05,
+      "loss": 0.8848,
+      "step": 5231
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.79262632681601e-05,
+      "loss": 0.9956,
+      "step": 5232
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.79253857050655e-05,
+      "loss": 1.0454,
+      "step": 5233
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792450797781652e-05,
+      "loss": 0.9453,
+      "step": 5234
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792363008643134e-05,
+      "loss": 1.0981,
+      "step": 5235
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7922752030928147e-05,
+      "loss": 0.9893,
+      "step": 5236
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7921873811325126e-05,
+      "loss": 1.0142,
+      "step": 5237
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7920995427640467e-05,
+      "loss": 0.9014,
+      "step": 5238
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7920116879892366e-05,
+      "loss": 0.875,
+      "step": 5239
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7919238168099015e-05,
+      "loss": 0.937,
+      "step": 5240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7918359292278614e-05,
+      "loss": 1.0303,
+      "step": 5241
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7917480252449372e-05,
+      "loss": 0.8999,
+      "step": 5242
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7916601048629493e-05,
+      "loss": 0.8765,
+      "step": 5243
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7915721680837185e-05,
+      "loss": 0.8818,
+      "step": 5244
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791484214909067e-05,
+      "loss": 0.9165,
+      "step": 5245
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7913962453408156e-05,
+      "loss": 0.9819,
+      "step": 5246
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7913082593807867e-05,
+      "loss": 0.9785,
+      "step": 5247
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7912202570308032e-05,
+      "loss": 1.0127,
+      "step": 5248
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791132238292687e-05,
+      "loss": 0.9463,
+      "step": 5249
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7910442031682622e-05,
+      "loss": 1.1025,
+      "step": 5250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790956151659351e-05,
+      "loss": 0.8696,
+      "step": 5251
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7908680837677782e-05,
+      "loss": 0.9746,
+      "step": 5252
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7907799994953677e-05,
+      "loss": 1.0361,
+      "step": 5253
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790691898843943e-05,
+      "loss": 1.061,
+      "step": 5254
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7906037818153304e-05,
+      "loss": 0.9995,
+      "step": 5255
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7905156484113538e-05,
+      "loss": 0.8828,
+      "step": 5256
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7904274986338394e-05,
+      "loss": 0.886,
+      "step": 5257
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7903393324846127e-05,
+      "loss": 1.001,
+      "step": 5258
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7902511499654995e-05,
+      "loss": 1.0879,
+      "step": 5259
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790162951078327e-05,
+      "loss": 1.0029,
+      "step": 5260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7900747358249215e-05,
+      "loss": 0.978,
+      "step": 5261
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7899865042071102e-05,
+      "loss": 0.9238,
+      "step": 5262
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789898256226721e-05,
+      "loss": 0.9561,
+      "step": 5263
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7898099918855813e-05,
+      "loss": 0.8862,
+      "step": 5264
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7897217111855195e-05,
+      "loss": 0.9858,
+      "step": 5265
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789633414128364e-05,
+      "loss": 0.9507,
+      "step": 5266
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7895451007159435e-05,
+      "loss": 0.981,
+      "step": 5267
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7894567709500875e-05,
+      "loss": 0.9517,
+      "step": 5268
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7893684248326252e-05,
+      "loss": 0.9595,
+      "step": 5269
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7892800623653868e-05,
+      "loss": 0.9331,
+      "step": 5270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7891916835502025e-05,
+      "loss": 0.9453,
+      "step": 5271
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7891032883889027e-05,
+      "loss": 1.0347,
+      "step": 5272
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7890148768833183e-05,
+      "loss": 0.9634,
+      "step": 5273
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.78892644903528e-05,
+      "loss": 0.8472,
+      "step": 5274
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7888380048466207e-05,
+      "loss": 1.0825,
+      "step": 5275
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7887495443191708e-05,
+      "loss": 0.9917,
+      "step": 5276
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7886610674547637e-05,
+      "loss": 1.0288,
+      "step": 5277
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7885725742552313e-05,
+      "loss": 1.0317,
+      "step": 5278
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7884840647224066e-05,
+      "loss": 0.9458,
+      "step": 5279
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788395538858123e-05,
+      "loss": 1.0078,
+      "step": 5280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7883069966642145e-05,
+      "loss": 0.9058,
+      "step": 5281
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788218438142514e-05,
+      "loss": 0.9326,
+      "step": 5282
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7881298632948564e-05,
+      "loss": 1.1123,
+      "step": 5283
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7880412721230765e-05,
+      "loss": 0.9248,
+      "step": 5284
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7879526646290088e-05,
+      "loss": 0.9517,
+      "step": 5285
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7878640408144887e-05,
+      "loss": 0.917,
+      "step": 5286
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.787775400681352e-05,
+      "loss": 0.8638,
+      "step": 5287
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7876867442314342e-05,
+      "loss": 1.0034,
+      "step": 5288
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7875980714665724e-05,
+      "loss": 0.9307,
+      "step": 5289
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7875093823886026e-05,
+      "loss": 1.0122,
+      "step": 5290
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787420676999362e-05,
+      "loss": 1.0566,
+      "step": 5291
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7873319553006875e-05,
+      "loss": 0.9561,
+      "step": 5292
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7872432172944178e-05,
+      "loss": 0.894,
+      "step": 5293
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7871544629823894e-05,
+      "loss": 1.0322,
+      "step": 5294
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787065692366442e-05,
+      "loss": 0.9429,
+      "step": 5295
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7869769054484135e-05,
+      "loss": 0.8564,
+      "step": 5296
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7868881022301428e-05,
+      "loss": 0.9658,
+      "step": 5297
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7867992827134697e-05,
+      "loss": 0.9492,
+      "step": 5298
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7867104469002337e-05,
+      "loss": 0.8657,
+      "step": 5299
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7866215947922746e-05,
+      "loss": 0.998,
+      "step": 5300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7865327263914328e-05,
+      "loss": 0.999,
+      "step": 5301
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7864438416995494e-05,
+      "loss": 0.9512,
+      "step": 5302
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7863549407184652e-05,
+      "loss": 0.8486,
+      "step": 5303
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7862660234500212e-05,
+      "loss": 0.897,
+      "step": 5304
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7861770898960594e-05,
+      "loss": 0.856,
+      "step": 5305
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7860881400584224e-05,
+      "loss": 1.1235,
+      "step": 5306
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7859991739389514e-05,
+      "loss": 0.833,
+      "step": 5307
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7859101915394895e-05,
+      "loss": 0.9907,
+      "step": 5308
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7858211928618806e-05,
+      "loss": 1.0063,
+      "step": 5309
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785732177907967e-05,
+      "loss": 1.0469,
+      "step": 5310
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785643146679593e-05,
+      "loss": 0.9473,
+      "step": 5311
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785554099178602e-05,
+      "loss": 0.8896,
+      "step": 5312
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.78546503540684e-05,
+      "loss": 0.8506,
+      "step": 5313
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7853759553661497e-05,
+      "loss": 0.9255,
+      "step": 5314
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785286859058377e-05,
+      "loss": 0.936,
+      "step": 5315
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785197746485368e-05,
+      "loss": 1.0132,
+      "step": 5316
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785108617648968e-05,
+      "loss": 0.915,
+      "step": 5317
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7850194725510222e-05,
+      "loss": 0.9604,
+      "step": 5318
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784930311193378e-05,
+      "loss": 0.9976,
+      "step": 5319
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7848411335778822e-05,
+      "loss": 0.9639,
+      "step": 5320
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7847519397063814e-05,
+      "loss": 0.9917,
+      "step": 5321
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784662729580723e-05,
+      "loss": 1.0361,
+      "step": 5322
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7845735032027552e-05,
+      "loss": 0.9922,
+      "step": 5323
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7844842605743256e-05,
+      "loss": 0.9468,
+      "step": 5324
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7843950016972833e-05,
+      "loss": 0.9443,
+      "step": 5325
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7843057265734765e-05,
+      "loss": 0.9529,
+      "step": 5326
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7842164352047545e-05,
+      "loss": 0.9722,
+      "step": 5327
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784127127592967e-05,
+      "loss": 0.8784,
+      "step": 5328
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784037803739963e-05,
+      "loss": 1.0361,
+      "step": 5329
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7839484636475936e-05,
+      "loss": 0.9678,
+      "step": 5330
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.783859107317708e-05,
+      "loss": 0.9429,
+      "step": 5331
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7837697347521588e-05,
+      "loss": 0.9707,
+      "step": 5332
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7836803459527957e-05,
+      "loss": 0.8843,
+      "step": 5333
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7835909409214703e-05,
+      "loss": 0.9092,
+      "step": 5334
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.783501519660035e-05,
+      "loss": 1.0508,
+      "step": 5335
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7834120821703417e-05,
+      "loss": 0.8679,
+      "step": 5336
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7833226284542424e-05,
+      "loss": 0.8984,
+      "step": 5337
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7832331585135903e-05,
+      "loss": 1.0361,
+      "step": 5338
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7831436723502393e-05,
+      "loss": 0.9561,
+      "step": 5339
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7830541699660412e-05,
+      "loss": 0.9829,
+      "step": 5340
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7829646513628513e-05,
+      "loss": 1.0962,
+      "step": 5341
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7828751165425227e-05,
+      "loss": 1.022,
+      "step": 5342
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7827855655069108e-05,
+      "loss": 1.0273,
+      "step": 5343
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.78269599825787e-05,
+      "loss": 0.9961,
+      "step": 5344
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782606414797255e-05,
+      "loss": 1.0103,
+      "step": 5345
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7825168151269218e-05,
+      "loss": 0.9033,
+      "step": 5346
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7824271992487263e-05,
+      "loss": 1.0132,
+      "step": 5347
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782337567164525e-05,
+      "loss": 1.0083,
+      "step": 5348
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7822479188761733e-05,
+      "loss": 0.9639,
+      "step": 5349
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782158254385529e-05,
+      "loss": 1.0054,
+      "step": 5350
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7820685736944488e-05,
+      "loss": 0.938,
+      "step": 5351
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7819788768047906e-05,
+      "loss": 0.8872,
+      "step": 5352
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7818891637184118e-05,
+      "loss": 0.9595,
+      "step": 5353
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781799434437171e-05,
+      "loss": 0.9502,
+      "step": 5354
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7817096889629263e-05,
+      "loss": 0.9639,
+      "step": 5355
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781619927297537e-05,
+      "loss": 0.9941,
+      "step": 5356
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781530149442862e-05,
+      "loss": 1.0093,
+      "step": 5357
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7814403554007606e-05,
+      "loss": 0.999,
+      "step": 5358
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7813505451730933e-05,
+      "loss": 1.0684,
+      "step": 5359
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7812607187617197e-05,
+      "loss": 1.083,
+      "step": 5360
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7811708761685006e-05,
+      "loss": 0.978,
+      "step": 5361
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7810810173952967e-05,
+      "loss": 0.8975,
+      "step": 5362
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7809911424439693e-05,
+      "loss": 0.9961,
+      "step": 5363
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7809012513163803e-05,
+      "loss": 0.9365,
+      "step": 5364
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780811344014391e-05,
+      "loss": 0.8228,
+      "step": 5365
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780721420539863e-05,
+      "loss": 0.9702,
+      "step": 5366
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7806314808946606e-05,
+      "loss": 0.9014,
+      "step": 5367
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7805415250806457e-05,
+      "loss": 1.0835,
+      "step": 5368
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7804515530996816e-05,
+      "loss": 1.083,
+      "step": 5369
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7803615649536314e-05,
+      "loss": 0.9336,
+      "step": 5370
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7802715606443595e-05,
+      "loss": 0.9751,
+      "step": 5371
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.78018154017373e-05,
+      "loss": 1.0015,
+      "step": 5372
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7800915035436073e-05,
+      "loss": 1.1357,
+      "step": 5373
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780001450755856e-05,
+      "loss": 1.0586,
+      "step": 5374
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7799113818123427e-05,
+      "loss": 1.0176,
+      "step": 5375
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7798212967149315e-05,
+      "loss": 0.9077,
+      "step": 5376
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7797311954654887e-05,
+      "loss": 0.9736,
+      "step": 5377
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7796410780658804e-05,
+      "loss": 1.0283,
+      "step": 5378
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.779550944517973e-05,
+      "loss": 0.9722,
+      "step": 5379
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7794607948236345e-05,
+      "loss": 0.8667,
+      "step": 5380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.779370628984731e-05,
+      "loss": 1.0718,
+      "step": 5381
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7792804470031303e-05,
+      "loss": 1.0913,
+      "step": 5382
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7791902488807003e-05,
+      "loss": 0.9067,
+      "step": 5383
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7791000346193095e-05,
+      "loss": 0.8618,
+      "step": 5384
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7790098042208264e-05,
+      "loss": 0.9277,
+      "step": 5385
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7789195576871198e-05,
+      "loss": 0.8975,
+      "step": 5386
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7788292950200586e-05,
+      "loss": 0.9463,
+      "step": 5387
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7787390162215125e-05,
+      "loss": 1.0859,
+      "step": 5388
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778648721293352e-05,
+      "loss": 0.8513,
+      "step": 5389
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7785584102374466e-05,
+      "loss": 1.0381,
+      "step": 5390
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7784680830556674e-05,
+      "loss": 0.9055,
+      "step": 5391
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7783777397498848e-05,
+      "loss": 0.9292,
+      "step": 5392
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7782873803219705e-05,
+      "loss": 0.9429,
+      "step": 5393
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778197004773796e-05,
+      "loss": 0.835,
+      "step": 5394
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7781066131072323e-05,
+      "loss": 1.127,
+      "step": 5395
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7780162053241532e-05,
+      "loss": 0.9795,
+      "step": 5396
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7779257814264302e-05,
+      "loss": 1.0752,
+      "step": 5397
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7778353414159366e-05,
+      "loss": 1.0073,
+      "step": 5398
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7777448852945452e-05,
+      "loss": 0.9429,
+      "step": 5399
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77765441306413e-05,
+      "loss": 0.9619,
+      "step": 5400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7775639247265644e-05,
+      "loss": 1.0439,
+      "step": 5401
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7774734202837236e-05,
+      "loss": 0.926,
+      "step": 5402
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7773828997374813e-05,
+      "loss": 0.9658,
+      "step": 5403
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777292363089713e-05,
+      "loss": 1.0068,
+      "step": 5404
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777201810342293e-05,
+      "loss": 0.9294,
+      "step": 5405
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777111241497098e-05,
+      "loss": 0.9473,
+      "step": 5406
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7770206565560034e-05,
+      "loss": 0.9404,
+      "step": 5407
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7769300555208856e-05,
+      "loss": 1.0747,
+      "step": 5408
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7768394383936204e-05,
+      "loss": 0.8345,
+      "step": 5409
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7767488051760858e-05,
+      "loss": 0.9077,
+      "step": 5410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7766581558701582e-05,
+      "loss": 0.9526,
+      "step": 5411
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776567490477716e-05,
+      "loss": 0.8584,
+      "step": 5412
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7764768090006364e-05,
+      "loss": 1.0273,
+      "step": 5413
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7763861114407976e-05,
+      "loss": 0.9355,
+      "step": 5414
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776295397800079e-05,
+      "loss": 0.9927,
+      "step": 5415
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7762046680803585e-05,
+      "loss": 0.9014,
+      "step": 5416
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776113922283516e-05,
+      "loss": 1.0161,
+      "step": 5417
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7760231604114308e-05,
+      "loss": 0.9268,
+      "step": 5418
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7759323824659826e-05,
+      "loss": 1.0991,
+      "step": 5419
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7758415884490527e-05,
+      "loss": 0.9697,
+      "step": 5420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77575077836252e-05,
+      "loss": 0.9238,
+      "step": 5421
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7756599522082668e-05,
+      "loss": 1.0127,
+      "step": 5422
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7755691099881734e-05,
+      "loss": 0.9414,
+      "step": 5423
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775478251704122e-05,
+      "loss": 0.8342,
+      "step": 5424
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7753873773579943e-05,
+      "loss": 1.0347,
+      "step": 5425
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775296486951672e-05,
+      "loss": 0.9307,
+      "step": 5426
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7752055804870387e-05,
+      "loss": 1.0723,
+      "step": 5427
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7751146579659767e-05,
+      "loss": 0.9243,
+      "step": 5428
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7750237193903693e-05,
+      "loss": 0.9941,
+      "step": 5429
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7749327647621002e-05,
+      "loss": 0.8533,
+      "step": 5430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774841794083053e-05,
+      "loss": 0.8965,
+      "step": 5431
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7747508073551118e-05,
+      "loss": 1.0112,
+      "step": 5432
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774659804580162e-05,
+      "loss": 0.8696,
+      "step": 5433
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7745687857600876e-05,
+      "loss": 0.9819,
+      "step": 5434
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774477750896774e-05,
+      "loss": 0.9692,
+      "step": 5435
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7743866999921075e-05,
+      "loss": 1.1074,
+      "step": 5436
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774295633047973e-05,
+      "loss": 0.9858,
+      "step": 5437
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7742045500662568e-05,
+      "loss": 0.9219,
+      "step": 5438
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7741134510488463e-05,
+      "loss": 0.9897,
+      "step": 5439
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7740223359976276e-05,
+      "loss": 1.0645,
+      "step": 5440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7739312049144883e-05,
+      "loss": 0.9047,
+      "step": 5441
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7738400578013158e-05,
+      "loss": 1.0864,
+      "step": 5442
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7737488946599984e-05,
+      "loss": 1.1055,
+      "step": 5443
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7736577154924236e-05,
+      "loss": 1.1182,
+      "step": 5444
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7735665203004803e-05,
+      "loss": 0.9116,
+      "step": 5445
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7734753090860574e-05,
+      "loss": 0.8984,
+      "step": 5446
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773384081851044e-05,
+      "loss": 0.9707,
+      "step": 5447
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7732928385973298e-05,
+      "loss": 1.0391,
+      "step": 5448
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7732015793268046e-05,
+      "loss": 1.0093,
+      "step": 5449
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7731103040413584e-05,
+      "loss": 1.0864,
+      "step": 5450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773019012742882e-05,
+      "loss": 0.9502,
+      "step": 5451
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7729277054332664e-05,
+      "loss": 0.8821,
+      "step": 5452
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7728363821144026e-05,
+      "loss": 1.0132,
+      "step": 5453
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772745042788182e-05,
+      "loss": 1.0059,
+      "step": 5454
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7726536874564966e-05,
+      "loss": 1.0396,
+      "step": 5455
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7725623161212387e-05,
+      "loss": 0.855,
+      "step": 5456
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7724709287843006e-05,
+      "loss": 0.9961,
+      "step": 5457
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7723795254475753e-05,
+      "loss": 0.9453,
+      "step": 5458
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772288106112956e-05,
+      "loss": 0.9399,
+      "step": 5459
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772196670782336e-05,
+      "loss": 1.0,
+      "step": 5460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.77210521945761e-05,
+      "loss": 0.9927,
+      "step": 5461
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7720137521406707e-05,
+      "loss": 0.9736,
+      "step": 5462
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771922268833414e-05,
+      "loss": 0.8608,
+      "step": 5463
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7718307695377338e-05,
+      "loss": 0.9075,
+      "step": 5464
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771739254255526e-05,
+      "loss": 0.9131,
+      "step": 5465
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7716477229886855e-05,
+      "loss": 0.9717,
+      "step": 5466
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7715561757391088e-05,
+      "loss": 1.0322,
+      "step": 5467
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7714646125086912e-05,
+      "loss": 0.9854,
+      "step": 5468
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.77137303329933e-05,
+      "loss": 0.9331,
+      "step": 5469
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7712814381129216e-05,
+      "loss": 0.9946,
+      "step": 5470
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7711898269513632e-05,
+      "loss": 0.9966,
+      "step": 5471
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7710981998165528e-05,
+      "loss": 1.0527,
+      "step": 5472
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7710065567103872e-05,
+      "loss": 0.9561,
+      "step": 5473
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7709148976347656e-05,
+      "loss": 1.0166,
+      "step": 5474
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770823222591586e-05,
+      "loss": 0.8584,
+      "step": 5475
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770731531582747e-05,
+      "loss": 1.0039,
+      "step": 5476
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770639824610148e-05,
+      "loss": 1.0186,
+      "step": 5477
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7705481016756887e-05,
+      "loss": 1.0024,
+      "step": 5478
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7704563627812684e-05,
+      "loss": 0.9233,
+      "step": 5479
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7703646079287878e-05,
+      "loss": 1.0225,
+      "step": 5480
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7702728371201468e-05,
+      "loss": 0.9551,
+      "step": 5481
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770181050357247e-05,
+      "loss": 0.9819,
+      "step": 5482
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7700892476419888e-05,
+      "loss": 0.9941,
+      "step": 5483
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7699974289762736e-05,
+      "loss": 1.0156,
+      "step": 5484
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769905594362004e-05,
+      "loss": 0.9873,
+      "step": 5485
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7698137438010814e-05,
+      "loss": 1.0024,
+      "step": 5486
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7697218772954085e-05,
+      "loss": 0.9741,
+      "step": 5487
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7696299948468882e-05,
+      "loss": 0.9614,
+      "step": 5488
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769538096457423e-05,
+      "loss": 0.9414,
+      "step": 5489
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769446182128917e-05,
+      "loss": 0.9668,
+      "step": 5490
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769354251863274e-05,
+      "loss": 1.0635,
+      "step": 5491
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769262305662398e-05,
+      "loss": 0.9922,
+      "step": 5492
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7691703435281928e-05,
+      "loss": 1.0391,
+      "step": 5493
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769078365462564e-05,
+      "loss": 0.8867,
+      "step": 5494
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7689863714674166e-05,
+      "loss": 0.8936,
+      "step": 5495
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7688943615446554e-05,
+      "loss": 0.9634,
+      "step": 5496
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7688023356961866e-05,
+      "loss": 0.9194,
+      "step": 5497
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7687102939239163e-05,
+      "loss": 0.9585,
+      "step": 5498
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7686182362297507e-05,
+      "loss": 0.9736,
+      "step": 5499
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768526162615597e-05,
+      "loss": 0.8857,
+      "step": 5500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7684340730833615e-05,
+      "loss": 0.8862,
+      "step": 5501
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7683419676349526e-05,
+      "loss": 1.0073,
+      "step": 5502
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7682498462722773e-05,
+      "loss": 0.9092,
+      "step": 5503
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7681577089972437e-05,
+      "loss": 0.9443,
+      "step": 5504
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7680655558117604e-05,
+      "loss": 0.9956,
+      "step": 5505
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7679733867177362e-05,
+      "loss": 0.8398,
+      "step": 5506
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7678812017170798e-05,
+      "loss": 0.9761,
+      "step": 5507
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767789000811701e-05,
+      "loss": 0.8994,
+      "step": 5508
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7676967840035093e-05,
+      "loss": 0.9297,
+      "step": 5509
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7676045512944147e-05,
+      "loss": 1.0068,
+      "step": 5510
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7675123026863276e-05,
+      "loss": 0.9702,
+      "step": 5511
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7674200381811588e-05,
+      "loss": 1.0156,
+      "step": 5512
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767327757780819e-05,
+      "loss": 1.0249,
+      "step": 5513
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76723546148722e-05,
+      "loss": 1.0083,
+      "step": 5514
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7671431493022733e-05,
+      "loss": 0.9116,
+      "step": 5515
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.767050821227891e-05,
+      "loss": 0.9624,
+      "step": 5516
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7669584772659854e-05,
+      "loss": 0.9185,
+      "step": 5517
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7668661174184688e-05,
+      "loss": 0.9883,
+      "step": 5518
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7667737416872545e-05,
+      "loss": 0.8774,
+      "step": 5519
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7666813500742562e-05,
+      "loss": 0.9146,
+      "step": 5520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7665889425813867e-05,
+      "loss": 0.8511,
+      "step": 5521
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7664965192105607e-05,
+      "loss": 1.0308,
+      "step": 5522
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7664040799636922e-05,
+      "loss": 0.9814,
+      "step": 5523
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7663116248426965e-05,
+      "loss": 1.0312,
+      "step": 5524
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7662191538494874e-05,
+      "loss": 0.9424,
+      "step": 5525
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766126666985981e-05,
+      "loss": 1.1316,
+      "step": 5526
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766034164254093e-05,
+      "loss": 1.0479,
+      "step": 5527
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765941645655739e-05,
+      "loss": 1.0234,
+      "step": 5528
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7658491111928352e-05,
+      "loss": 0.9111,
+      "step": 5529
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765756560867299e-05,
+      "loss": 0.8838,
+      "step": 5530
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765663994681046e-05,
+      "loss": 0.9551,
+      "step": 5531
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7655714126359947e-05,
+      "loss": 1.041,
+      "step": 5532
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7654788147340625e-05,
+      "loss": 1.0293,
+      "step": 5533
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7653862009771667e-05,
+      "loss": 0.9844,
+      "step": 5534
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765293571367226e-05,
+      "loss": 0.999,
+      "step": 5535
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7652009259061592e-05,
+      "loss": 1.0435,
+      "step": 5536
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7651082645958847e-05,
+      "loss": 1.0083,
+      "step": 5537
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7650155874383217e-05,
+      "loss": 0.9536,
+      "step": 5538
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7649228944353906e-05,
+      "loss": 0.9287,
+      "step": 5539
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7648301855890104e-05,
+      "loss": 0.9668,
+      "step": 5540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764737460901102e-05,
+      "loss": 1.0312,
+      "step": 5541
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7646447203735855e-05,
+      "loss": 1.063,
+      "step": 5542
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7645519640083822e-05,
+      "loss": 0.8125,
+      "step": 5543
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7644591918074128e-05,
+      "loss": 1.0034,
+      "step": 5544
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764366403772599e-05,
+      "loss": 0.9775,
+      "step": 5545
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764273599905863e-05,
+      "loss": 0.9399,
+      "step": 5546
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764180780209127e-05,
+      "loss": 0.979,
+      "step": 5547
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764087944684313e-05,
+      "loss": 1.0459,
+      "step": 5548
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763995093333344e-05,
+      "loss": 0.96,
+      "step": 5549
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7639022261581435e-05,
+      "loss": 0.9888,
+      "step": 5550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763809343160635e-05,
+      "loss": 1.0347,
+      "step": 5551
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763716444342742e-05,
+      "loss": 0.9219,
+      "step": 5552
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7636235297063892e-05,
+      "loss": 0.9341,
+      "step": 5553
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7635305992535e-05,
+      "loss": 0.9702,
+      "step": 5554
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7634376529860006e-05,
+      "loss": 0.9775,
+      "step": 5555
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7633446909058153e-05,
+      "loss": 0.9673,
+      "step": 5556
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76325171301487e-05,
+      "loss": 1.0298,
+      "step": 5557
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76315871931509e-05,
+      "loss": 0.8745,
+      "step": 5558
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7630657098084016e-05,
+      "loss": 1.0049,
+      "step": 5559
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7629726844967315e-05,
+      "loss": 1.0635,
+      "step": 5560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7628796433820062e-05,
+      "loss": 1.0015,
+      "step": 5561
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7627865864661535e-05,
+      "loss": 1.041,
+      "step": 5562
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7626935137511e-05,
+      "loss": 0.9717,
+      "step": 5563
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7626004252387735e-05,
+      "loss": 1.0288,
+      "step": 5564
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762507320931103e-05,
+      "loss": 1.0771,
+      "step": 5565
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7624142008300158e-05,
+      "loss": 0.8853,
+      "step": 5566
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7623210649374412e-05,
+      "loss": 0.9307,
+      "step": 5567
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7622279132553085e-05,
+      "loss": 1.0142,
+      "step": 5568
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7621347457855466e-05,
+      "loss": 0.9229,
+      "step": 5569
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7620415625300853e-05,
+      "loss": 0.9482,
+      "step": 5570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7619483634908548e-05,
+      "loss": 1.0127,
+      "step": 5571
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7618551486697857e-05,
+      "loss": 1.0566,
+      "step": 5572
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7617619180688087e-05,
+      "loss": 0.9668,
+      "step": 5573
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761668671689854e-05,
+      "loss": 0.9829,
+      "step": 5574
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761575409534854e-05,
+      "loss": 1.0225,
+      "step": 5575
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.76148213160574e-05,
+      "loss": 1.0288,
+      "step": 5576
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761388837904444e-05,
+      "loss": 1.0088,
+      "step": 5577
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761295528432898e-05,
+      "loss": 0.9438,
+      "step": 5578
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7612022031930354e-05,
+      "loss": 0.978,
+      "step": 5579
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761108862186789e-05,
+      "loss": 1.0581,
+      "step": 5580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7610155054160913e-05,
+      "loss": 0.957,
+      "step": 5581
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7609221328828773e-05,
+      "loss": 1.0776,
+      "step": 5582
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7608287445890795e-05,
+      "loss": 0.9258,
+      "step": 5583
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7607353405366335e-05,
+      "loss": 0.9131,
+      "step": 5584
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7606419207274732e-05,
+      "loss": 1.0146,
+      "step": 5585
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760548485163534e-05,
+      "loss": 0.9683,
+      "step": 5586
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7604550338467505e-05,
+      "loss": 1.0273,
+      "step": 5587
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760361566779059e-05,
+      "loss": 0.9885,
+      "step": 5588
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760268083962395e-05,
+      "loss": 0.9653,
+      "step": 5589
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760174585398695e-05,
+      "loss": 0.9497,
+      "step": 5590
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760081071089896e-05,
+      "loss": 0.9644,
+      "step": 5591
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759987541037934e-05,
+      "loss": 0.9458,
+      "step": 5592
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7598939952447467e-05,
+      "loss": 0.998,
+      "step": 5593
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7598004337122715e-05,
+      "loss": 1.0015,
+      "step": 5594
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7597068564424464e-05,
+      "loss": 1.0049,
+      "step": 5595
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.75961326343721e-05,
+      "loss": 1.021,
+      "step": 5596
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7595196546985003e-05,
+      "loss": 0.9316,
+      "step": 5597
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7594260302282565e-05,
+      "loss": 1.1396,
+      "step": 5598
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7593323900284173e-05,
+      "loss": 0.9688,
+      "step": 5599
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7592387341009233e-05,
+      "loss": 1.0176,
+      "step": 5600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7591450624477127e-05,
+      "loss": 0.8628,
+      "step": 5601
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759051375070727e-05,
+      "loss": 1.0239,
+      "step": 5602
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7589576719719063e-05,
+      "loss": 1.0093,
+      "step": 5603
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7588639531531914e-05,
+      "loss": 0.9697,
+      "step": 5604
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7587702186165238e-05,
+      "loss": 1.0747,
+      "step": 5605
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7586764683638442e-05,
+      "loss": 1.1084,
+      "step": 5606
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758582702397095e-05,
+      "loss": 1.0586,
+      "step": 5607
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758488920718218e-05,
+      "loss": 1.0918,
+      "step": 5608
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758395123329156e-05,
+      "loss": 0.979,
+      "step": 5609
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7583013102318517e-05,
+      "loss": 1.0083,
+      "step": 5610
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758207481428248e-05,
+      "loss": 0.8682,
+      "step": 5611
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7581136369202883e-05,
+      "loss": 0.9268,
+      "step": 5612
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7580197767099166e-05,
+      "loss": 0.9404,
+      "step": 5613
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757925900799077e-05,
+      "loss": 0.9146,
+      "step": 5614
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7578320091897134e-05,
+      "loss": 0.8828,
+      "step": 5615
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757738101883771e-05,
+      "loss": 0.9688,
+      "step": 5616
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7576441788831947e-05,
+      "loss": 0.969,
+      "step": 5617
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7575502401899302e-05,
+      "loss": 1.0146,
+      "step": 5618
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7574562858059225e-05,
+      "loss": 0.917,
+      "step": 5619
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7573623157331183e-05,
+      "loss": 0.9731,
+      "step": 5620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7572683299734635e-05,
+      "loss": 1.1094,
+      "step": 5621
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7571743285289053e-05,
+      "loss": 0.9609,
+      "step": 5622
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7570803114013904e-05,
+      "loss": 0.8828,
+      "step": 5623
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7569862785928657e-05,
+      "loss": 1.0269,
+      "step": 5624
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7568922301052796e-05,
+      "loss": 0.9131,
+      "step": 5625
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7567981659405795e-05,
+      "loss": 0.9941,
+      "step": 5626
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756704086100714e-05,
+      "loss": 0.9761,
+      "step": 5627
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7566099905876318e-05,
+      "loss": 0.9785,
+      "step": 5628
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7565158794032816e-05,
+      "loss": 1.0615,
+      "step": 5629
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7564217525496127e-05,
+      "loss": 1.1123,
+      "step": 5630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7563276100285753e-05,
+      "loss": 0.938,
+      "step": 5631
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756233451842118e-05,
+      "loss": 0.9819,
+      "step": 5632
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7561392779921923e-05,
+      "loss": 1.0454,
+      "step": 5633
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756045088480748e-05,
+      "loss": 0.9668,
+      "step": 5634
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755950883309737e-05,
+      "loss": 0.9497,
+      "step": 5635
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7558566624811098e-05,
+      "loss": 0.8589,
+      "step": 5636
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7557624259968176e-05,
+      "loss": 0.9316,
+      "step": 5637
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7556681738588128e-05,
+      "loss": 1.0425,
+      "step": 5638
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7555739060690475e-05,
+      "loss": 0.9937,
+      "step": 5639
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755479622629474e-05,
+      "loss": 0.9692,
+      "step": 5640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7553853235420454e-05,
+      "loss": 1.0208,
+      "step": 5641
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755291008808715e-05,
+      "loss": 0.9731,
+      "step": 5642
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7551966784314363e-05,
+      "loss": 1.0186,
+      "step": 5643
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7551023324121624e-05,
+      "loss": 0.9783,
+      "step": 5644
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7550079707528483e-05,
+      "loss": 1.0752,
+      "step": 5645
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7549135934554477e-05,
+      "loss": 0.979,
+      "step": 5646
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7548192005219164e-05,
+      "loss": 0.9771,
+      "step": 5647
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7547247919542087e-05,
+      "loss": 0.9541,
+      "step": 5648
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75463036775428e-05,
+      "loss": 0.9937,
+      "step": 5649
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7545359279240865e-05,
+      "loss": 1.0532,
+      "step": 5650
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7544414724655842e-05,
+      "loss": 1.103,
+      "step": 5651
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754347001380729e-05,
+      "loss": 0.998,
+      "step": 5652
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7542525146714782e-05,
+      "loss": 0.9521,
+      "step": 5653
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7541580123397888e-05,
+      "loss": 1.0342,
+      "step": 5654
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754063494387618e-05,
+      "loss": 0.9614,
+      "step": 5655
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753968960816924e-05,
+      "loss": 1.0713,
+      "step": 5656
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7538744116296637e-05,
+      "loss": 1.0649,
+      "step": 5657
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7537798468277965e-05,
+      "loss": 1.0059,
+      "step": 5658
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7536852664132808e-05,
+      "loss": 0.9712,
+      "step": 5659
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7535906703880753e-05,
+      "loss": 0.9883,
+      "step": 5660
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7534960587541393e-05,
+      "loss": 1.0518,
+      "step": 5661
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753401431513433e-05,
+      "loss": 0.9932,
+      "step": 5662
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7533067886679157e-05,
+      "loss": 0.9736,
+      "step": 5663
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7532121302195486e-05,
+      "loss": 0.855,
+      "step": 5664
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7531174561702914e-05,
+      "loss": 1.0142,
+      "step": 5665
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7530227665221053e-05,
+      "loss": 0.9397,
+      "step": 5666
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7529280612769516e-05,
+      "loss": 0.9976,
+      "step": 5667
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7528333404367918e-05,
+      "loss": 0.9792,
+      "step": 5668
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7527386040035882e-05,
+      "loss": 0.9126,
+      "step": 5669
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7526438519793026e-05,
+      "loss": 0.9175,
+      "step": 5670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7525490843658977e-05,
+      "loss": 0.9839,
+      "step": 5671
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752454301165336e-05,
+      "loss": 0.8936,
+      "step": 5672
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7523595023795814e-05,
+      "loss": 0.9194,
+      "step": 5673
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752264688010597e-05,
+      "loss": 1.0142,
+      "step": 5674
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7521698580603467e-05,
+      "loss": 1.1069,
+      "step": 5675
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7520750125307947e-05,
+      "loss": 1.0186,
+      "step": 5676
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7519801514239053e-05,
+      "loss": 0.9204,
+      "step": 5677
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7518852747416436e-05,
+      "loss": 0.9878,
+      "step": 5678
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7517903824859744e-05,
+      "loss": 1.0596,
+      "step": 5679
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7516954746588636e-05,
+      "loss": 1.0288,
+      "step": 5680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7516005512622768e-05,
+      "loss": 0.876,
+      "step": 5681
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75150561229818e-05,
+      "loss": 0.9517,
+      "step": 5682
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7514106577685393e-05,
+      "loss": 1.042,
+      "step": 5683
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7513156876753216e-05,
+      "loss": 0.9141,
+      "step": 5684
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751220702020495e-05,
+      "loss": 1.0381,
+      "step": 5685
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7511257008060252e-05,
+      "loss": 0.9761,
+      "step": 5686
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7510306840338812e-05,
+      "loss": 0.9922,
+      "step": 5687
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7509356517060304e-05,
+      "loss": 1.019,
+      "step": 5688
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7508406038244415e-05,
+      "loss": 0.9058,
+      "step": 5689
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.750745540391083e-05,
+      "loss": 1.0098,
+      "step": 5690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7506504614079236e-05,
+      "loss": 1.0537,
+      "step": 5691
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7505553668769333e-05,
+      "loss": 1.0098,
+      "step": 5692
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7504602568000813e-05,
+      "loss": 0.9873,
+      "step": 5693
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7503651311793372e-05,
+      "loss": 1.0581,
+      "step": 5694
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7502699900166723e-05,
+      "loss": 1.0195,
+      "step": 5695
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7501748333140566e-05,
+      "loss": 1.022,
+      "step": 5696
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7500796610734607e-05,
+      "loss": 1.0288,
+      "step": 5697
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7499844732968565e-05,
+      "loss": 0.8862,
+      "step": 5698
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7498892699862153e-05,
+      "loss": 1.021,
+      "step": 5699
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749794051143509e-05,
+      "loss": 1.0366,
+      "step": 5700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7496988167707095e-05,
+      "loss": 0.9048,
+      "step": 5701
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.74960356686979e-05,
+      "loss": 0.9951,
+      "step": 5702
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749508301442723e-05,
+      "loss": 0.8938,
+      "step": 5703
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7494130204914815e-05,
+      "loss": 0.9414,
+      "step": 5704
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7493177240180394e-05,
+      "loss": 0.9585,
+      "step": 5705
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.74922241202437e-05,
+      "loss": 0.8652,
+      "step": 5706
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749127084512448e-05,
+      "loss": 1.0054,
+      "step": 5707
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749031741484248e-05,
+      "loss": 0.9502,
+      "step": 5708
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7489363829417438e-05,
+      "loss": 0.9409,
+      "step": 5709
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7488410088869115e-05,
+      "loss": 0.8936,
+      "step": 5710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748745619321726e-05,
+      "loss": 0.8828,
+      "step": 5711
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7486502142481633e-05,
+      "loss": 1.041,
+      "step": 5712
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7485547936681995e-05,
+      "loss": 0.9116,
+      "step": 5713
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748459357583811e-05,
+      "loss": 1.0791,
+      "step": 5714
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7483639059969743e-05,
+      "loss": 0.9629,
+      "step": 5715
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7482684389096664e-05,
+      "loss": 1.0762,
+      "step": 5716
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748172956323865e-05,
+      "loss": 1.0093,
+      "step": 5717
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7480774582415476e-05,
+      "loss": 0.9858,
+      "step": 5718
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747981944664692e-05,
+      "loss": 1.0054,
+      "step": 5719
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747886415595277e-05,
+      "loss": 0.8379,
+      "step": 5720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747790871035281e-05,
+      "loss": 0.9668,
+      "step": 5721
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7476953109866828e-05,
+      "loss": 0.9331,
+      "step": 5722
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7475997354514617e-05,
+      "loss": 0.9409,
+      "step": 5723
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7475041444315978e-05,
+      "loss": 0.9702,
+      "step": 5724
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7474085379290703e-05,
+      "loss": 0.9097,
+      "step": 5725
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7473129159458597e-05,
+      "loss": 0.9565,
+      "step": 5726
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747217278483947e-05,
+      "loss": 0.9849,
+      "step": 5727
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7471216255453126e-05,
+      "loss": 0.9175,
+      "step": 5728
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7470259571319378e-05,
+      "loss": 0.8906,
+      "step": 5729
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7469302732458044e-05,
+      "loss": 1.0063,
+      "step": 5730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746834573888894e-05,
+      "loss": 0.9946,
+      "step": 5731
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746738859063189e-05,
+      "loss": 0.8955,
+      "step": 5732
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746643128770671e-05,
+      "loss": 0.9917,
+      "step": 5733
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7465473830133244e-05,
+      "loss": 0.9927,
+      "step": 5734
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746451621793131e-05,
+      "loss": 1.0347,
+      "step": 5735
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7463558451120747e-05,
+      "loss": 0.8198,
+      "step": 5736
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7462600529721393e-05,
+      "loss": 1.0146,
+      "step": 5737
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746164245375309e-05,
+      "loss": 0.9419,
+      "step": 5738
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.746068422323568e-05,
+      "loss": 0.8906,
+      "step": 5739
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7459725838189012e-05,
+      "loss": 0.9453,
+      "step": 5740
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7458767298632934e-05,
+      "loss": 0.8926,
+      "step": 5741
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7457808604587302e-05,
+      "loss": 1.0117,
+      "step": 5742
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7456849756071976e-05,
+      "loss": 0.8955,
+      "step": 5743
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7455890753106808e-05,
+      "loss": 0.907,
+      "step": 5744
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7454931595711668e-05,
+      "loss": 1.0454,
+      "step": 5745
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745397228390642e-05,
+      "loss": 1.0176,
+      "step": 5746
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7453012817710933e-05,
+      "loss": 0.9158,
+      "step": 5747
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745205319714508e-05,
+      "loss": 0.9897,
+      "step": 5748
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745109342222874e-05,
+      "loss": 1.043,
+      "step": 5749
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7450133492981792e-05,
+      "loss": 0.8882,
+      "step": 5750
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7449173409424114e-05,
+      "loss": 0.9126,
+      "step": 5751
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7448213171575596e-05,
+      "loss": 0.9058,
+      "step": 5752
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7447252779456124e-05,
+      "loss": 0.9478,
+      "step": 5753
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7446292233085592e-05,
+      "loss": 1.0039,
+      "step": 5754
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7445331532483896e-05,
+      "loss": 0.9438,
+      "step": 5755
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7444370677670935e-05,
+      "loss": 1.0615,
+      "step": 5756
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7443409668666607e-05,
+      "loss": 1.0649,
+      "step": 5757
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744244850549082e-05,
+      "loss": 0.9785,
+      "step": 5758
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744148718816348e-05,
+      "loss": 1.0327,
+      "step": 5759
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.74405257167045e-05,
+      "loss": 0.8201,
+      "step": 5760
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7439564091133792e-05,
+      "loss": 0.998,
+      "step": 5761
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743860231147128e-05,
+      "loss": 0.895,
+      "step": 5762
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7437640377736874e-05,
+      "loss": 0.9165,
+      "step": 5763
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743667828995051e-05,
+      "loss": 1.0195,
+      "step": 5764
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7435716048132105e-05,
+      "loss": 0.9873,
+      "step": 5765
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7434753652301597e-05,
+      "loss": 0.9136,
+      "step": 5766
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7433791102478916e-05,
+      "loss": 0.874,
+      "step": 5767
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7432828398684e-05,
+      "loss": 0.9814,
+      "step": 5768
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7431865540936788e-05,
+      "loss": 1.0513,
+      "step": 5769
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7430902529257222e-05,
+      "loss": 0.9253,
+      "step": 5770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742993936366525e-05,
+      "loss": 1.0093,
+      "step": 5771
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742897604418082e-05,
+      "loss": 0.9741,
+      "step": 5772
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742801257082389e-05,
+      "loss": 1.0605,
+      "step": 5773
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742704894361441e-05,
+      "loss": 0.9199,
+      "step": 5774
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7426085162572343e-05,
+      "loss": 1.0269,
+      "step": 5775
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7425121227717646e-05,
+      "loss": 1.1074,
+      "step": 5776
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742415713907029e-05,
+      "loss": 0.9932,
+      "step": 5777
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7423192896650244e-05,
+      "loss": 1.0063,
+      "step": 5778
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7422228500477474e-05,
+      "loss": 0.9214,
+      "step": 5779
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7421263950571962e-05,
+      "loss": 0.9688,
+      "step": 5780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742029924695368e-05,
+      "loss": 1.0435,
+      "step": 5781
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.741933438964261e-05,
+      "loss": 0.8701,
+      "step": 5782
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.741836937865874e-05,
+      "loss": 0.9487,
+      "step": 5783
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741740421402206e-05,
+      "loss": 1.0732,
+      "step": 5784
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7416438895752557e-05,
+      "loss": 1.0132,
+      "step": 5785
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741547342387022e-05,
+      "loss": 1.0288,
+      "step": 5786
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7414507798395055e-05,
+      "loss": 1.0947,
+      "step": 5787
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7413542019347063e-05,
+      "loss": 0.9878,
+      "step": 5788
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7412576086746238e-05,
+      "loss": 1.0791,
+      "step": 5789
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7411610000612596e-05,
+      "loss": 0.9448,
+      "step": 5790
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7410643760966146e-05,
+      "loss": 0.9536,
+      "step": 5791
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7409677367826897e-05,
+      "loss": 0.9873,
+      "step": 5792
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7408710821214864e-05,
+      "loss": 1.0156,
+      "step": 5793
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7407744121150075e-05,
+      "loss": 1.0405,
+      "step": 5794
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7406777267652546e-05,
+      "loss": 1.0454,
+      "step": 5795
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7405810260742307e-05,
+      "loss": 0.9702,
+      "step": 5796
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7404843100439382e-05,
+      "loss": 1.0068,
+      "step": 5797
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7403875786763808e-05,
+      "loss": 1.0615,
+      "step": 5798
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7402908319735616e-05,
+      "loss": 0.9165,
+      "step": 5799
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740194069937485e-05,
+      "loss": 0.9858,
+      "step": 5800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740097292570155e-05,
+      "loss": 0.9844,
+      "step": 5801
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740000499873576e-05,
+      "loss": 0.939,
+      "step": 5802
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7399036918497526e-05,
+      "loss": 0.8643,
+      "step": 5803
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7398068685006903e-05,
+      "loss": 1.0259,
+      "step": 5804
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7397100298283943e-05,
+      "loss": 0.9761,
+      "step": 5805
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7396131758348702e-05,
+      "loss": 1.0952,
+      "step": 5806
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7395163065221247e-05,
+      "loss": 1.1748,
+      "step": 5807
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739419421892164e-05,
+      "loss": 1.0747,
+      "step": 5808
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7393225219469946e-05,
+      "loss": 0.8789,
+      "step": 5809
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7392256066886237e-05,
+      "loss": 0.9692,
+      "step": 5810
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7391286761190585e-05,
+      "loss": 0.9556,
+      "step": 5811
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739031730240307e-05,
+      "loss": 1.0254,
+      "step": 5812
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7389347690543764e-05,
+      "loss": 1.0894,
+      "step": 5813
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7388377925632758e-05,
+      "loss": 0.8916,
+      "step": 5814
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738740800769014e-05,
+      "loss": 0.9834,
+      "step": 5815
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738643793673599e-05,
+      "loss": 1.0186,
+      "step": 5816
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7385467712790407e-05,
+      "loss": 0.9053,
+      "step": 5817
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7384497335873484e-05,
+      "loss": 0.9565,
+      "step": 5818
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7383526806005326e-05,
+      "loss": 0.9448,
+      "step": 5819
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7382556123206026e-05,
+      "loss": 0.9536,
+      "step": 5820
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7381585287495693e-05,
+      "loss": 1.0083,
+      "step": 5821
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7380614298894443e-05,
+      "loss": 1.0156,
+      "step": 5822
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7379643157422377e-05,
+      "loss": 0.9922,
+      "step": 5823
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737867186309961e-05,
+      "loss": 0.9961,
+      "step": 5824
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7377700415946267e-05,
+      "loss": 1.0234,
+      "step": 5825
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7376728815982463e-05,
+      "loss": 0.9971,
+      "step": 5826
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737575706322833e-05,
+      "loss": 0.9692,
+      "step": 5827
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7374785157703983e-05,
+      "loss": 0.9351,
+      "step": 5828
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7373813099429564e-05,
+      "loss": 0.9844,
+      "step": 5829
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.73728408884252e-05,
+      "loss": 1.0098,
+      "step": 5830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7371868524711034e-05,
+      "loss": 0.9883,
+      "step": 5831
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.73708960083072e-05,
+      "loss": 0.875,
+      "step": 5832
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7369923339233847e-05,
+      "loss": 1.0151,
+      "step": 5833
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7368950517511112e-05,
+      "loss": 0.8325,
+      "step": 5834
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7367977543159153e-05,
+      "loss": 0.9673,
+      "step": 5835
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736700441619812e-05,
+      "loss": 1.0649,
+      "step": 5836
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736603113664817e-05,
+      "loss": 0.9609,
+      "step": 5837
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7365057704529462e-05,
+      "loss": 0.9565,
+      "step": 5838
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736408411986216e-05,
+      "loss": 1.0537,
+      "step": 5839
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7363110382666423e-05,
+      "loss": 1.0225,
+      "step": 5840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736213649296242e-05,
+      "loss": 1.0708,
+      "step": 5841
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736116245077033e-05,
+      "loss": 0.9092,
+      "step": 5842
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7360188256110323e-05,
+      "loss": 0.8428,
+      "step": 5843
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735921390900258e-05,
+      "loss": 0.9507,
+      "step": 5844
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7358239409467276e-05,
+      "loss": 0.9395,
+      "step": 5845
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.73572647575246e-05,
+      "loss": 0.9624,
+      "step": 5846
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735628995319474e-05,
+      "loss": 0.9888,
+      "step": 5847
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7355314996497882e-05,
+      "loss": 1.0127,
+      "step": 5848
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7354339887454224e-05,
+      "loss": 1.0034,
+      "step": 5849
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735336462608397e-05,
+      "loss": 1.0381,
+      "step": 5850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7352389212407305e-05,
+      "loss": 1.0044,
+      "step": 5851
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7351413646444436e-05,
+      "loss": 0.9893,
+      "step": 5852
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7350437928215577e-05,
+      "loss": 0.9702,
+      "step": 5853
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7349462057740935e-05,
+      "loss": 1.0376,
+      "step": 5854
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7348486035040716e-05,
+      "loss": 0.9624,
+      "step": 5855
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734750986013514e-05,
+      "loss": 1.0767,
+      "step": 5856
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7346533533044428e-05,
+      "loss": 0.9277,
+      "step": 5857
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7345557053788806e-05,
+      "loss": 0.959,
+      "step": 5858
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7344580422388492e-05,
+      "loss": 0.894,
+      "step": 5859
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7343603638863714e-05,
+      "loss": 0.8735,
+      "step": 5860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734262670323471e-05,
+      "loss": 0.9473,
+      "step": 5861
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7341649615521706e-05,
+      "loss": 1.0933,
+      "step": 5862
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7340672375744945e-05,
+      "loss": 0.8506,
+      "step": 5863
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7339694983924672e-05,
+      "loss": 0.8904,
+      "step": 5864
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733871744008112e-05,
+      "loss": 1.0259,
+      "step": 5865
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7337739744234554e-05,
+      "loss": 0.8564,
+      "step": 5866
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7336761896405207e-05,
+      "loss": 0.9517,
+      "step": 5867
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7335783896613344e-05,
+      "loss": 1.0244,
+      "step": 5868
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7334805744879214e-05,
+      "loss": 0.9922,
+      "step": 5869
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733382744122308e-05,
+      "loss": 1.043,
+      "step": 5870
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7332848985665206e-05,
+      "loss": 1.0586,
+      "step": 5871
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7331870378225863e-05,
+      "loss": 1.0034,
+      "step": 5872
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7330891618925307e-05,
+      "loss": 0.9194,
+      "step": 5873
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7329912707783825e-05,
+      "loss": 1.0376,
+      "step": 5874
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7328933644821683e-05,
+      "loss": 0.9756,
+      "step": 5875
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7327954430059164e-05,
+      "loss": 1.0332,
+      "step": 5876
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7326975063516548e-05,
+      "loss": 0.8555,
+      "step": 5877
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732599554521412e-05,
+      "loss": 1.0063,
+      "step": 5878
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7325015875172173e-05,
+      "loss": 0.8433,
+      "step": 5879
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7324036053410995e-05,
+      "loss": 0.9722,
+      "step": 5880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732305607995088e-05,
+      "loss": 1.0938,
+      "step": 5881
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7322075954812123e-05,
+      "loss": 1.0674,
+      "step": 5882
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7321095678015033e-05,
+      "loss": 0.9751,
+      "step": 5883
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7320115249579903e-05,
+      "loss": 0.8804,
+      "step": 5884
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731913466952705e-05,
+      "loss": 1.0127,
+      "step": 5885
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7318153937876778e-05,
+      "loss": 0.9907,
+      "step": 5886
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7317173054649404e-05,
+      "loss": 0.9648,
+      "step": 5887
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7316192019865244e-05,
+      "loss": 0.8237,
+      "step": 5888
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7315210833544616e-05,
+      "loss": 0.9316,
+      "step": 5889
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7314229495707842e-05,
+      "loss": 0.8906,
+      "step": 5890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7313248006375248e-05,
+      "loss": 1.0181,
+      "step": 5891
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731226636556717e-05,
+      "loss": 0.9775,
+      "step": 5892
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731128457330393e-05,
+      "loss": 0.957,
+      "step": 5893
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731030262960587e-05,
+      "loss": 1.0142,
+      "step": 5894
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730932053449333e-05,
+      "loss": 1.0332,
+      "step": 5895
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730833828798664e-05,
+      "loss": 0.9272,
+      "step": 5896
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730735589010616e-05,
+      "loss": 0.9434,
+      "step": 5897
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730637334087223e-05,
+      "loss": 0.9551,
+      "step": 5898
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.73053906403052e-05,
+      "loss": 1.0801,
+      "step": 5899
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730440778842543e-05,
+      "loss": 0.9658,
+      "step": 5900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7303424785253268e-05,
+      "loss": 0.9497,
+      "step": 5901
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7302441630809083e-05,
+      "loss": 0.9575,
+      "step": 5902
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7301458325113233e-05,
+      "loss": 0.9824,
+      "step": 5903
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730047486818609e-05,
+      "loss": 0.8828,
+      "step": 5904
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729949126004802e-05,
+      "loss": 0.9575,
+      "step": 5905
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7298507500719396e-05,
+      "loss": 0.8813,
+      "step": 5906
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7297523590220596e-05,
+      "loss": 1.0083,
+      "step": 5907
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7296539528571998e-05,
+      "loss": 0.8975,
+      "step": 5908
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7295555315793982e-05,
+      "loss": 0.9165,
+      "step": 5909
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7294570951906942e-05,
+      "loss": 1.0498,
+      "step": 5910
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7293586436931255e-05,
+      "loss": 0.8833,
+      "step": 5911
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7292601770887322e-05,
+      "loss": 1.0557,
+      "step": 5912
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729161695379553e-05,
+      "loss": 1.0049,
+      "step": 5913
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7290631985676288e-05,
+      "loss": 1.0093,
+      "step": 5914
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7289646866549986e-05,
+      "loss": 0.8862,
+      "step": 5915
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7288661596437033e-05,
+      "loss": 0.939,
+      "step": 5916
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7287676175357834e-05,
+      "loss": 0.9653,
+      "step": 5917
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7286690603332808e-05,
+      "loss": 0.9551,
+      "step": 5918
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7285704880382356e-05,
+      "loss": 0.9666,
+      "step": 5919
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7284719006526903e-05,
+      "loss": 0.998,
+      "step": 5920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7283732981786866e-05,
+      "loss": 0.916,
+      "step": 5921
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728274680618267e-05,
+      "loss": 1.0195,
+      "step": 5922
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7281760479734743e-05,
+      "loss": 0.8438,
+      "step": 5923
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728077400246351e-05,
+      "loss": 0.9182,
+      "step": 5924
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.72797873743894e-05,
+      "loss": 1.0415,
+      "step": 5925
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7278800595532854e-05,
+      "loss": 0.9751,
+      "step": 5926
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7277813665914313e-05,
+      "loss": 0.9819,
+      "step": 5927
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7276826585554217e-05,
+      "loss": 0.9729,
+      "step": 5928
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7275839354473004e-05,
+      "loss": 0.8867,
+      "step": 5929
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7274851972691134e-05,
+      "loss": 0.9521,
+      "step": 5930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7273864440229046e-05,
+      "loss": 0.8975,
+      "step": 5931
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7272876757107203e-05,
+      "loss": 0.978,
+      "step": 5932
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727188892334606e-05,
+      "loss": 0.9834,
+      "step": 5933
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7270900938966072e-05,
+      "loss": 0.9775,
+      "step": 5934
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726991280398771e-05,
+      "loss": 0.9553,
+      "step": 5935
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7268924518431437e-05,
+      "loss": 0.8403,
+      "step": 5936
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7267936082317728e-05,
+      "loss": 0.8857,
+      "step": 5937
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726694749566705e-05,
+      "loss": 1.1787,
+      "step": 5938
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726595875849988e-05,
+      "loss": 0.7939,
+      "step": 5939
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7264969870836698e-05,
+      "loss": 0.9697,
+      "step": 5940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7263980832697983e-05,
+      "loss": 1.0747,
+      "step": 5941
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726299164410423e-05,
+      "loss": 0.9692,
+      "step": 5942
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7262002305075918e-05,
+      "loss": 0.9326,
+      "step": 5943
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7261012815633543e-05,
+      "loss": 0.9419,
+      "step": 5944
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.72600231757976e-05,
+      "loss": 1.0205,
+      "step": 5945
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725903338558858e-05,
+      "loss": 1.0146,
+      "step": 5946
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7258043445026995e-05,
+      "loss": 1.0088,
+      "step": 5947
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725705335413334e-05,
+      "loss": 0.9907,
+      "step": 5948
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725606311292813e-05,
+      "loss": 1.0786,
+      "step": 5949
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725507272143187e-05,
+      "loss": 1.022,
+      "step": 5950
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725408217966508e-05,
+      "loss": 1.0137,
+      "step": 5951
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7253091487648267e-05,
+      "loss": 0.9946,
+      "step": 5952
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7252100645401952e-05,
+      "loss": 0.9795,
+      "step": 5953
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7251109652946663e-05,
+      "loss": 0.8882,
+      "step": 5954
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7250118510302926e-05,
+      "loss": 0.9639,
+      "step": 5955
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7249127217491265e-05,
+      "loss": 1.0366,
+      "step": 5956
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724813577453222e-05,
+      "loss": 1.0303,
+      "step": 5957
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7247144181446316e-05,
+      "loss": 1.0708,
+      "step": 5958
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7246152438254098e-05,
+      "loss": 0.9702,
+      "step": 5959
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724516054497611e-05,
+      "loss": 0.9985,
+      "step": 5960
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7244168501632887e-05,
+      "loss": 0.9233,
+      "step": 5961
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7243176308244985e-05,
+      "loss": 0.9097,
+      "step": 5962
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7242183964832952e-05,
+      "loss": 0.894,
+      "step": 5963
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.724119147141734e-05,
+      "loss": 0.9897,
+      "step": 5964
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.724019882801871e-05,
+      "loss": 1.0625,
+      "step": 5965
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723920603465762e-05,
+      "loss": 1.0181,
+      "step": 5966
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7238213091354634e-05,
+      "loss": 1.0503,
+      "step": 5967
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723721999813032e-05,
+      "loss": 0.8984,
+      "step": 5968
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7236226755005237e-05,
+      "loss": 0.9824,
+      "step": 5969
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7235233361999975e-05,
+      "loss": 0.8628,
+      "step": 5970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7234239819135092e-05,
+      "loss": 0.9985,
+      "step": 5971
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723324612643118e-05,
+      "loss": 1.0581,
+      "step": 5972
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7232252283908815e-05,
+      "loss": 0.9678,
+      "step": 5973
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7231258291588582e-05,
+      "loss": 1.1035,
+      "step": 5974
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723026414949107e-05,
+      "loss": 1.0552,
+      "step": 5975
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722926985763687e-05,
+      "loss": 0.9478,
+      "step": 5976
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7228275416046575e-05,
+      "loss": 0.9375,
+      "step": 5977
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7227280824740784e-05,
+      "loss": 0.9397,
+      "step": 5978
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.72262860837401e-05,
+      "loss": 0.9336,
+      "step": 5979
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7225291193065122e-05,
+      "loss": 1.001,
+      "step": 5980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7224296152736455e-05,
+      "loss": 0.9619,
+      "step": 5981
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7223300962774712e-05,
+      "loss": 0.897,
+      "step": 5982
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722230562320051e-05,
+      "loss": 0.8872,
+      "step": 5983
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7221310134034458e-05,
+      "loss": 1.0923,
+      "step": 5984
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722031449529718e-05,
+      "loss": 0.9819,
+      "step": 5985
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7219318707009293e-05,
+      "loss": 0.9634,
+      "step": 5986
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7218322769191423e-05,
+      "loss": 0.9858,
+      "step": 5987
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721732668186421e-05,
+      "loss": 0.8345,
+      "step": 5988
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721633044504827e-05,
+      "loss": 1.1113,
+      "step": 5989
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7215334058764243e-05,
+      "loss": 0.9956,
+      "step": 5990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7214337523032765e-05,
+      "loss": 0.9844,
+      "step": 5991
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721334083787448e-05,
+      "loss": 0.9958,
+      "step": 5992
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721234400331003e-05,
+      "loss": 1.0503,
+      "step": 5993
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7211347019360064e-05,
+      "loss": 0.9224,
+      "step": 5994
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7210349886045233e-05,
+      "loss": 0.9648,
+      "step": 5995
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7209352603386187e-05,
+      "loss": 1.0405,
+      "step": 5996
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7208355171403578e-05,
+      "loss": 0.9609,
+      "step": 5997
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7207357590118076e-05,
+      "loss": 1.0474,
+      "step": 5998
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7206359859550335e-05,
+      "loss": 0.8696,
+      "step": 5999
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7205361979721024e-05,
+      "loss": 1.021,
+      "step": 6000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720436395065081e-05,
+      "loss": 0.9824,
+      "step": 6001
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7203365772360365e-05,
+      "loss": 0.9868,
+      "step": 6002
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720236744487036e-05,
+      "loss": 1.0439,
+      "step": 6003
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720136896820148e-05,
+      "loss": 0.9609,
+      "step": 6004
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7200370342374407e-05,
+      "loss": 1.0703,
+      "step": 6005
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7199371567409815e-05,
+      "loss": 0.9883,
+      "step": 6006
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.71983726433284e-05,
+      "loss": 1.0576,
+      "step": 6007
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7197373570150852e-05,
+      "loss": 0.9297,
+      "step": 6008
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7196374347897854e-05,
+      "loss": 1.0132,
+      "step": 6009
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7195374976590113e-05,
+      "loss": 1.0737,
+      "step": 6010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7194375456248325e-05,
+      "loss": 0.8877,
+      "step": 6011
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7193375786893193e-05,
+      "loss": 0.9956,
+      "step": 6012
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719237596854542e-05,
+      "loss": 1.0312,
+      "step": 6013
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719137600122572e-05,
+      "loss": 1.0239,
+      "step": 6014
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71903758849548e-05,
+      "loss": 0.9229,
+      "step": 6015
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718937561975338e-05,
+      "loss": 1.0991,
+      "step": 6016
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7188375205642172e-05,
+      "loss": 0.9922,
+      "step": 6017
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7187374642641898e-05,
+      "loss": 1.0513,
+      "step": 6018
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718637393077329e-05,
+      "loss": 0.9849,
+      "step": 6019
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7185373070057064e-05,
+      "loss": 0.8843,
+      "step": 6020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7184372060513953e-05,
+      "loss": 0.9585,
+      "step": 6021
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71833709021647e-05,
+      "loss": 0.9739,
+      "step": 6022
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7182369595030032e-05,
+      "loss": 0.9658,
+      "step": 6023
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7181368139130688e-05,
+      "loss": 0.8356,
+      "step": 6024
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7180366534487415e-05,
+      "loss": 0.9629,
+      "step": 6025
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7179364781120958e-05,
+      "loss": 1.0435,
+      "step": 6026
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717836287905206e-05,
+      "loss": 0.9692,
+      "step": 6027
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7177360828301484e-05,
+      "loss": 0.9026,
+      "step": 6028
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7176358628889978e-05,
+      "loss": 1.0405,
+      "step": 6029
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7175356280838298e-05,
+      "loss": 0.8813,
+      "step": 6030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717435378416721e-05,
+      "loss": 0.9385,
+      "step": 6031
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7173351138897476e-05,
+      "loss": 1.0391,
+      "step": 6032
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717234834504986e-05,
+      "loss": 0.9556,
+      "step": 6033
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7171345402645144e-05,
+      "loss": 0.9199,
+      "step": 6034
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7170342311704085e-05,
+      "loss": 0.9438,
+      "step": 6035
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716933907224747e-05,
+      "loss": 0.9404,
+      "step": 6036
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7168335684296077e-05,
+      "loss": 0.9565,
+      "step": 6037
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7167332147870687e-05,
+      "loss": 0.9604,
+      "step": 6038
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7166328462992087e-05,
+      "loss": 1.0332,
+      "step": 6039
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716532462968107e-05,
+      "loss": 0.9219,
+      "step": 6040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7164320647958418e-05,
+      "loss": 0.8955,
+      "step": 6041
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716331651784493e-05,
+      "loss": 1.0156,
+      "step": 6042
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716231223936141e-05,
+      "loss": 0.9312,
+      "step": 6043
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716130781252865e-05,
+      "loss": 0.9541,
+      "step": 6044
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7160303237367462e-05,
+      "loss": 0.8809,
+      "step": 6045
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7159298513898648e-05,
+      "loss": 0.9971,
+      "step": 6046
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715829364214302e-05,
+      "loss": 0.9917,
+      "step": 6047
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715728862212139e-05,
+      "loss": 1.0112,
+      "step": 6048
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715628345385458e-05,
+      "loss": 0.9668,
+      "step": 6049
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7155278137363402e-05,
+      "loss": 0.9521,
+      "step": 6050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7154272672668684e-05,
+      "loss": 0.9521,
+      "step": 6051
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715326705979125e-05,
+      "loss": 0.8979,
+      "step": 6052
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7152261298751927e-05,
+      "loss": 0.8855,
+      "step": 6053
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7151255389571547e-05,
+      "loss": 0.9758,
+      "step": 6054
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.715024933227095e-05,
+      "loss": 1.1216,
+      "step": 6055
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7149243126870966e-05,
+      "loss": 0.9153,
+      "step": 6056
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7148236773392443e-05,
+      "loss": 0.9014,
+      "step": 6057
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714723027185622e-05,
+      "loss": 0.9453,
+      "step": 6058
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7146223622283148e-05,
+      "loss": 0.8384,
+      "step": 6059
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7145216824694076e-05,
+      "loss": 0.8755,
+      "step": 6060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7144209879109853e-05,
+      "loss": 0.9849,
+      "step": 6061
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7143202785551346e-05,
+      "loss": 0.9536,
+      "step": 6062
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7142195544039402e-05,
+      "loss": 1.0083,
+      "step": 6063
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714118815459489e-05,
+      "loss": 1.0229,
+      "step": 6064
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7140180617238673e-05,
+      "loss": 0.9414,
+      "step": 6065
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7139172931991623e-05,
+      "loss": 0.9902,
+      "step": 6066
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7138165098874612e-05,
+      "loss": 0.9336,
+      "step": 6067
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713715711790851e-05,
+      "loss": 0.9624,
+      "step": 6068
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7136148989114198e-05,
+      "loss": 0.9253,
+      "step": 6069
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7135140712512552e-05,
+      "loss": 1.0083,
+      "step": 6070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7134132288124464e-05,
+      "loss": 0.9067,
+      "step": 6071
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7133123715970818e-05,
+      "loss": 0.9844,
+      "step": 6072
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.71321149960725e-05,
+      "loss": 0.9229,
+      "step": 6073
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713110612845041e-05,
+      "loss": 0.937,
+      "step": 6074
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7130097113125437e-05,
+      "loss": 0.9268,
+      "step": 6075
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7129087950118486e-05,
+      "loss": 0.9458,
+      "step": 6076
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7128078639450453e-05,
+      "loss": 1.0952,
+      "step": 6077
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712706918114225e-05,
+      "loss": 0.9644,
+      "step": 6078
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7126059575214782e-05,
+      "loss": 0.9902,
+      "step": 6079
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712504982168896e-05,
+      "loss": 1.0371,
+      "step": 6080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7124039920585698e-05,
+      "loss": 0.8994,
+      "step": 6081
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712302987192592e-05,
+      "loss": 0.9243,
+      "step": 6082
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7122019675730537e-05,
+      "loss": 0.9478,
+      "step": 6083
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712100933202048e-05,
+      "loss": 1.1138,
+      "step": 6084
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711999884081667e-05,
+      "loss": 0.9565,
+      "step": 6085
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7118988202140043e-05,
+      "loss": 0.9268,
+      "step": 6086
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7117977416011523e-05,
+      "loss": 0.9971,
+      "step": 6087
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7116966482452056e-05,
+      "loss": 1.1045,
+      "step": 6088
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711595540148258e-05,
+      "loss": 1.0469,
+      "step": 6089
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7114944173124025e-05,
+      "loss": 0.916,
+      "step": 6090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7113932797397348e-05,
+      "loss": 0.998,
+      "step": 6091
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7112921274323496e-05,
+      "loss": 0.9463,
+      "step": 6092
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7111909603923415e-05,
+      "loss": 1.0508,
+      "step": 6093
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7110897786218063e-05,
+      "loss": 0.9268,
+      "step": 6094
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7109885821228393e-05,
+      "loss": 0.8589,
+      "step": 6095
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710887370897537e-05,
+      "loss": 0.8545,
+      "step": 6096
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7107861449479957e-05,
+      "loss": 1.0298,
+      "step": 6097
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710684904276312e-05,
+      "loss": 0.9658,
+      "step": 6098
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7105836488845824e-05,
+      "loss": 0.9302,
+      "step": 6099
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710482378774905e-05,
+      "loss": 1.0933,
+      "step": 6100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7103810939493764e-05,
+      "loss": 0.9937,
+      "step": 6101
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710279794410095e-05,
+      "loss": 0.9375,
+      "step": 6102
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7101784801591594e-05,
+      "loss": 0.998,
+      "step": 6103
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710077151198667e-05,
+      "loss": 0.958,
+      "step": 6104
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7099758075307172e-05,
+      "loss": 0.9062,
+      "step": 6105
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7098744491574094e-05,
+      "loss": 0.8965,
+      "step": 6106
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7097730760808425e-05,
+      "loss": 1.0239,
+      "step": 6107
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709671688303116e-05,
+      "loss": 1.0112,
+      "step": 6108
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70957028582633e-05,
+      "loss": 0.9678,
+      "step": 6109
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7094688686525853e-05,
+      "loss": 0.9941,
+      "step": 6110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7093674367839823e-05,
+      "loss": 0.958,
+      "step": 6111
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7092659902226215e-05,
+      "loss": 1.0308,
+      "step": 6112
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7091645289706048e-05,
+      "loss": 1.0317,
+      "step": 6113
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7090630530300326e-05,
+      "loss": 0.8569,
+      "step": 6114
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708961562403008e-05,
+      "loss": 1.0942,
+      "step": 6115
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7088600570916325e-05,
+      "loss": 0.938,
+      "step": 6116
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7087585370980083e-05,
+      "loss": 1.0669,
+      "step": 6117
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7086570024242382e-05,
+      "loss": 0.9985,
+      "step": 6118
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708555453072426e-05,
+      "loss": 0.9648,
+      "step": 6119
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7084538890446746e-05,
+      "loss": 0.8613,
+      "step": 6120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708352310343087e-05,
+      "loss": 0.9937,
+      "step": 6121
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7082507169697678e-05,
+      "loss": 1.0132,
+      "step": 6122
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708149108926821e-05,
+      "loss": 0.8464,
+      "step": 6123
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7080474862163513e-05,
+      "loss": 0.9321,
+      "step": 6124
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7079458488404635e-05,
+      "loss": 0.8926,
+      "step": 6125
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707844196801263e-05,
+      "loss": 0.8765,
+      "step": 6126
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707742530100855e-05,
+      "loss": 0.894,
+      "step": 6127
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7076408487413454e-05,
+      "loss": 0.9263,
+      "step": 6128
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7075391527248402e-05,
+      "loss": 1.0024,
+      "step": 6129
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7074374420534456e-05,
+      "loss": 0.8848,
+      "step": 6130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7073357167292684e-05,
+      "loss": 1.0327,
+      "step": 6131
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7072339767544156e-05,
+      "loss": 1.0303,
+      "step": 6132
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7071322221309947e-05,
+      "loss": 1.0718,
+      "step": 6133
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707030452861113e-05,
+      "loss": 0.8447,
+      "step": 6134
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7069286689468788e-05,
+      "loss": 0.9243,
+      "step": 6135
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7068268703903996e-05,
+      "loss": 0.8782,
+      "step": 6136
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7067250571937847e-05,
+      "loss": 0.9302,
+      "step": 6137
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706623229359142e-05,
+      "loss": 0.8877,
+      "step": 6138
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7065213868885817e-05,
+      "loss": 1.0889,
+      "step": 6139
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7064195297842122e-05,
+      "loss": 1.0215,
+      "step": 6140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7063176580481437e-05,
+      "loss": 1.0029,
+      "step": 6141
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7062157716824864e-05,
+      "loss": 0.9932,
+      "step": 6142
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70611387068935e-05,
+      "loss": 0.9487,
+      "step": 6143
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7060119550708457e-05,
+      "loss": 1.0791,
+      "step": 6144
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705910024829084e-05,
+      "loss": 0.9458,
+      "step": 6145
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7058080799661766e-05,
+      "loss": 0.9648,
+      "step": 6146
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7057061204842348e-05,
+      "loss": 1.0337,
+      "step": 6147
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7056041463853703e-05,
+      "loss": 1.04,
+      "step": 6148
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7055021576716954e-05,
+      "loss": 0.9453,
+      "step": 6149
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7054001543453224e-05,
+      "loss": 1.0566,
+      "step": 6150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7052981364083643e-05,
+      "loss": 0.98,
+      "step": 6151
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705196103862934e-05,
+      "loss": 0.9243,
+      "step": 6152
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7050940567111447e-05,
+      "loss": 0.9717,
+      "step": 6153
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7049919949551103e-05,
+      "loss": 0.9487,
+      "step": 6154
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7048899185969444e-05,
+      "loss": 1.0576,
+      "step": 6155
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7047878276387616e-05,
+      "loss": 1.0439,
+      "step": 6156
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704685722082676e-05,
+      "loss": 0.9438,
+      "step": 6157
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7045836019308032e-05,
+      "loss": 1.1084,
+      "step": 6158
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704481467185258e-05,
+      "loss": 0.9141,
+      "step": 6159
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7043793178481557e-05,
+      "loss": 0.9463,
+      "step": 6160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7042771539216122e-05,
+      "loss": 0.9854,
+      "step": 6161
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704174975407743e-05,
+      "loss": 0.9619,
+      "step": 6162
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7040727823086657e-05,
+      "loss": 1.001,
+      "step": 6163
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7039705746264964e-05,
+      "loss": 1.0435,
+      "step": 6164
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7038683523633515e-05,
+      "loss": 0.959,
+      "step": 6165
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703766115521349e-05,
+      "loss": 1.001,
+      "step": 6166
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7036638641026064e-05,
+      "loss": 0.9607,
+      "step": 6167
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703561598109241e-05,
+      "loss": 0.9341,
+      "step": 6168
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7034593175433712e-05,
+      "loss": 0.8247,
+      "step": 6169
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7033570224071162e-05,
+      "loss": 0.9849,
+      "step": 6170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703254712702594e-05,
+      "loss": 0.9961,
+      "step": 6171
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703152388431924e-05,
+      "loss": 1.0762,
+      "step": 6172
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7030500495972254e-05,
+      "loss": 0.9561,
+      "step": 6173
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7029476962006182e-05,
+      "loss": 1.0693,
+      "step": 6174
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702845328244222e-05,
+      "loss": 0.9175,
+      "step": 6175
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7027429457301577e-05,
+      "loss": 1.0244,
+      "step": 6176
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7026405486605453e-05,
+      "loss": 1.0166,
+      "step": 6177
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702538137037506e-05,
+      "loss": 0.8618,
+      "step": 6178
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7024357108631604e-05,
+      "loss": 0.9795,
+      "step": 6179
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7023332701396312e-05,
+      "loss": 0.8989,
+      "step": 6180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702230814869039e-05,
+      "loss": 0.9106,
+      "step": 6181
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7021283450535066e-05,
+      "loss": 0.9751,
+      "step": 6182
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702025860695156e-05,
+      "loss": 0.9131,
+      "step": 6183
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7019233617961104e-05,
+      "loss": 0.8691,
+      "step": 6184
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7018208483584927e-05,
+      "loss": 0.957,
+      "step": 6185
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7017183203844253e-05,
+      "loss": 1.0166,
+      "step": 6186
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7016157778760334e-05,
+      "loss": 0.8982,
+      "step": 6187
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701513220835439e-05,
+      "loss": 0.9946,
+      "step": 6188
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7014106492647684e-05,
+      "loss": 0.9111,
+      "step": 6189
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7013080631661444e-05,
+      "loss": 0.9609,
+      "step": 6190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701205462541693e-05,
+      "loss": 1.0234,
+      "step": 6191
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701102847393538e-05,
+      "loss": 0.9834,
+      "step": 6192
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7010002177238067e-05,
+      "loss": 1.0117,
+      "step": 6193
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700897573534623e-05,
+      "loss": 0.9136,
+      "step": 6194
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7007949148281135e-05,
+      "loss": 0.9951,
+      "step": 6195
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700692241606405e-05,
+      "loss": 0.9912,
+      "step": 6196
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7005895538716236e-05,
+      "loss": 0.9111,
+      "step": 6197
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7004868516258968e-05,
+      "loss": 0.9897,
+      "step": 6198
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700384134871351e-05,
+      "loss": 1.0098,
+      "step": 6199
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700281403610114e-05,
+      "loss": 1.0601,
+      "step": 6200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700178657844314e-05,
+      "loss": 1.0146,
+      "step": 6201
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700075897576079e-05,
+      "loss": 1.0405,
+      "step": 6202
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6999731228075372e-05,
+      "loss": 0.9824,
+      "step": 6203
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699870333540817e-05,
+      "loss": 0.9736,
+      "step": 6204
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6997675297780484e-05,
+      "loss": 0.9141,
+      "step": 6205
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6996647115213595e-05,
+      "loss": 0.9927,
+      "step": 6206
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699561878772881e-05,
+      "loss": 1.0454,
+      "step": 6207
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6994590315347423e-05,
+      "loss": 0.9155,
+      "step": 6208
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6993561698090734e-05,
+      "loss": 0.9575,
+      "step": 6209
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6992532935980058e-05,
+      "loss": 1.0391,
+      "step": 6210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6991504029036692e-05,
+      "loss": 0.9248,
+      "step": 6211
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699047497728195e-05,
+      "loss": 0.8193,
+      "step": 6212
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6989445780737152e-05,
+      "loss": 0.9897,
+      "step": 6213
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698841643942361e-05,
+      "loss": 0.9849,
+      "step": 6214
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6987386953362644e-05,
+      "loss": 0.9375,
+      "step": 6215
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698635732257558e-05,
+      "loss": 1.0488,
+      "step": 6216
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6985327547083743e-05,
+      "loss": 1.0,
+      "step": 6217
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698429762690846e-05,
+      "loss": 0.8518,
+      "step": 6218
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6983267562071067e-05,
+      "loss": 0.9404,
+      "step": 6219
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6982237352592893e-05,
+      "loss": 1.0063,
+      "step": 6220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6981206998495285e-05,
+      "loss": 0.9609,
+      "step": 6221
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6980176499799578e-05,
+      "loss": 0.9429,
+      "step": 6222
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697914585652712e-05,
+      "loss": 0.9946,
+      "step": 6223
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6978115068699255e-05,
+      "loss": 0.8936,
+      "step": 6224
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6977084136337332e-05,
+      "loss": 0.9639,
+      "step": 6225
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697605305946271e-05,
+      "loss": 1.0088,
+      "step": 6226
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697502183809674e-05,
+      "loss": 0.9668,
+      "step": 6227
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6973990472260783e-05,
+      "loss": 1.0361,
+      "step": 6228
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.69729589619762e-05,
+      "loss": 0.9448,
+      "step": 6229
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697192730726436e-05,
+      "loss": 0.8696,
+      "step": 6230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6970895508146622e-05,
+      "loss": 0.8901,
+      "step": 6231
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6969863564644367e-05,
+      "loss": 0.9985,
+      "step": 6232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6968831476778963e-05,
+      "loss": 0.937,
+      "step": 6233
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6967799244571793e-05,
+      "loss": 0.9126,
+      "step": 6234
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696676686804423e-05,
+      "loss": 0.9976,
+      "step": 6235
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6965734347217662e-05,
+      "loss": 1.0151,
+      "step": 6236
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6964701682113477e-05,
+      "loss": 0.8745,
+      "step": 6237
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6963668872753056e-05,
+      "loss": 0.9438,
+      "step": 6238
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6962635919157798e-05,
+      "loss": 0.9053,
+      "step": 6239
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6961602821349095e-05,
+      "loss": 1.0078,
+      "step": 6240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6960569579348347e-05,
+      "loss": 0.958,
+      "step": 6241
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6959536193176954e-05,
+      "loss": 0.8552,
+      "step": 6242
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6958502662856318e-05,
+      "loss": 1.0229,
+      "step": 6243
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695746898840785e-05,
+      "loss": 0.9365,
+      "step": 6244
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6956435169852954e-05,
+      "loss": 0.9893,
+      "step": 6245
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6955401207213048e-05,
+      "loss": 0.9839,
+      "step": 6246
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695436710050955e-05,
+      "loss": 0.9492,
+      "step": 6247
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6953332849763873e-05,
+      "loss": 1.0337,
+      "step": 6248
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695229845499744e-05,
+      "loss": 0.9395,
+      "step": 6249
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695126391623168e-05,
+      "loss": 1.0195,
+      "step": 6250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6950229233488015e-05,
+      "loss": 1.0278,
+      "step": 6251
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6949194406787882e-05,
+      "loss": 0.8472,
+      "step": 6252
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694815943615271e-05,
+      "loss": 0.8503,
+      "step": 6253
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6947124321603937e-05,
+      "loss": 0.9683,
+      "step": 6254
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6946089063163006e-05,
+      "loss": 1.0332,
+      "step": 6255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6945053660851358e-05,
+      "loss": 0.8809,
+      "step": 6256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694401811469043e-05,
+      "loss": 0.98,
+      "step": 6257
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6942982424701687e-05,
+      "loss": 0.7932,
+      "step": 6258
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6941946590906573e-05,
+      "loss": 0.9517,
+      "step": 6259
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6940910613326538e-05,
+      "loss": 0.999,
+      "step": 6260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6939874491983044e-05,
+      "loss": 1.0552,
+      "step": 6261
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693883822689755e-05,
+      "loss": 0.9272,
+      "step": 6262
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6937801818091527e-05,
+      "loss": 0.9453,
+      "step": 6263
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693676526558643e-05,
+      "loss": 0.9849,
+      "step": 6264
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6935728569403734e-05,
+      "loss": 0.9932,
+      "step": 6265
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6934691729564915e-05,
+      "loss": 0.9453,
+      "step": 6266
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6933654746091442e-05,
+      "loss": 0.9106,
+      "step": 6267
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6932617619004798e-05,
+      "loss": 0.9756,
+      "step": 6268
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693158034832646e-05,
+      "loss": 0.8022,
+      "step": 6269
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693054293407792e-05,
+      "loss": 0.9946,
+      "step": 6270
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6929505376280663e-05,
+      "loss": 0.9443,
+      "step": 6271
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6928467674956173e-05,
+      "loss": 0.9829,
+      "step": 6272
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692742983012595e-05,
+      "loss": 1.001,
+      "step": 6273
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6926391841811487e-05,
+      "loss": 0.9209,
+      "step": 6274
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6925353710034286e-05,
+      "loss": 1.0635,
+      "step": 6275
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6924315434815847e-05,
+      "loss": 0.9058,
+      "step": 6276
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6923277016177674e-05,
+      "loss": 1.0659,
+      "step": 6277
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6922238454141278e-05,
+      "loss": 0.9253,
+      "step": 6278
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.692119974872817e-05,
+      "loss": 0.98,
+      "step": 6279
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6920160899959864e-05,
+      "loss": 0.8545,
+      "step": 6280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6919121907857874e-05,
+      "loss": 1.0166,
+      "step": 6281
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6918082772443728e-05,
+      "loss": 0.9463,
+      "step": 6282
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691704349373894e-05,
+      "loss": 0.9453,
+      "step": 6283
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691600407176504e-05,
+      "loss": 0.9287,
+      "step": 6284
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6914964506543556e-05,
+      "loss": 0.9829,
+      "step": 6285
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6913924798096025e-05,
+      "loss": 1.1338,
+      "step": 6286
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6912884946443978e-05,
+      "loss": 0.9795,
+      "step": 6287
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691184495160895e-05,
+      "loss": 0.9175,
+      "step": 6288
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691080481361248e-05,
+      "loss": 0.9751,
+      "step": 6289
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6909764532476126e-05,
+      "loss": 0.8018,
+      "step": 6290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690872410822142e-05,
+      "loss": 0.9478,
+      "step": 6291
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6907683540869915e-05,
+      "loss": 1.0098,
+      "step": 6292
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690664283044317e-05,
+      "loss": 0.9995,
+      "step": 6293
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690560197696273e-05,
+      "loss": 0.9133,
+      "step": 6294
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690456098045017e-05,
+      "loss": 0.8823,
+      "step": 6295
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6903519840927033e-05,
+      "loss": 0.9907,
+      "step": 6296
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6902478558414896e-05,
+      "loss": 1.0483,
+      "step": 6297
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690143713293532e-05,
+      "loss": 0.9053,
+      "step": 6298
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6900395564509883e-05,
+      "loss": 0.9731,
+      "step": 6299
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689935385316015e-05,
+      "loss": 1.106,
+      "step": 6300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6898311998907704e-05,
+      "loss": 0.9854,
+      "step": 6301
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689727000177412e-05,
+      "loss": 1.0444,
+      "step": 6302
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6896227861780983e-05,
+      "loss": 1.0498,
+      "step": 6303
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6895185578949875e-05,
+      "loss": 0.9863,
+      "step": 6304
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6894143153302388e-05,
+      "loss": 0.9189,
+      "step": 6305
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689310058486011e-05,
+      "loss": 0.8398,
+      "step": 6306
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6892057873644638e-05,
+      "loss": 0.9756,
+      "step": 6307
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689101501967757e-05,
+      "loss": 1.0366,
+      "step": 6308
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.68899720229805e-05,
+      "loss": 1.1411,
+      "step": 6309
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6888928883575036e-05,
+      "loss": 0.9644,
+      "step": 6310
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688788560148279e-05,
+      "loss": 1.0474,
+      "step": 6311
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6886842176725354e-05,
+      "loss": 0.8823,
+      "step": 6312
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6885798609324357e-05,
+      "loss": 0.9565,
+      "step": 6313
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6884754899301404e-05,
+      "loss": 1.0107,
+      "step": 6314
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6883711046678118e-05,
+      "loss": 1.0513,
+      "step": 6315
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6882667051476117e-05,
+      "loss": 0.9185,
+      "step": 6316
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6881622913717023e-05,
+      "loss": 0.9355,
+      "step": 6317
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6880578633422465e-05,
+      "loss": 1.0493,
+      "step": 6318
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6879534210614074e-05,
+      "loss": 1.0249,
+      "step": 6319
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687848964531348e-05,
+      "loss": 0.9468,
+      "step": 6320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6877444937542323e-05,
+      "loss": 0.9795,
+      "step": 6321
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6876400087322232e-05,
+      "loss": 0.9629,
+      "step": 6322
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687535509467486e-05,
+      "loss": 1.0596,
+      "step": 6323
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6874309959621844e-05,
+      "loss": 0.9795,
+      "step": 6324
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6873264682184833e-05,
+      "loss": 0.916,
+      "step": 6325
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6872219262385477e-05,
+      "loss": 1.0068,
+      "step": 6326
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687117370024543e-05,
+      "loss": 0.999,
+      "step": 6327
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6870127995786346e-05,
+      "loss": 1.0112,
+      "step": 6328
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6869082149029892e-05,
+      "loss": 0.9214,
+      "step": 6329
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686803615999772e-05,
+      "loss": 0.9712,
+      "step": 6330
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6866990028711502e-05,
+      "loss": 0.9019,
+      "step": 6331
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.68659437551929e-05,
+      "loss": 1.0825,
+      "step": 6332
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686489733946359e-05,
+      "loss": 1.0664,
+      "step": 6333
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6863850781545245e-05,
+      "loss": 0.9458,
+      "step": 6334
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6862804081459536e-05,
+      "loss": 0.8713,
+      "step": 6335
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6861757239228153e-05,
+      "loss": 0.9648,
+      "step": 6336
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686071025487277e-05,
+      "loss": 1.0869,
+      "step": 6337
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6859663128415076e-05,
+      "loss": 0.9673,
+      "step": 6338
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6858615859876758e-05,
+      "loss": 0.9185,
+      "step": 6339
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6857568449279514e-05,
+      "loss": 1.0454,
+      "step": 6340
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685652089664503e-05,
+      "loss": 0.9883,
+      "step": 6341
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6855473201995007e-05,
+      "loss": 0.9614,
+      "step": 6342
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6854425365351145e-05,
+      "loss": 0.9717,
+      "step": 6343
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685337738673515e-05,
+      "loss": 0.9805,
+      "step": 6344
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6852329266168724e-05,
+      "loss": 1.064,
+      "step": 6345
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685128100367358e-05,
+      "loss": 0.9985,
+      "step": 6346
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6850232599271423e-05,
+      "loss": 0.8706,
+      "step": 6347
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684918405298398e-05,
+      "loss": 0.9453,
+      "step": 6348
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6848135364832955e-05,
+      "loss": 0.8501,
+      "step": 6349
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684708653484008e-05,
+      "loss": 0.9668,
+      "step": 6350
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6846037563027075e-05,
+      "loss": 1.0107,
+      "step": 6351
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684498844941566e-05,
+      "loss": 1.0439,
+      "step": 6352
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684393919402758e-05,
+      "loss": 1.1191,
+      "step": 6353
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6842889796884556e-05,
+      "loss": 1.0527,
+      "step": 6354
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6841840258008328e-05,
+      "loss": 0.9551,
+      "step": 6355
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684079057742063e-05,
+      "loss": 0.9834,
+      "step": 6356
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6839740755143212e-05,
+      "loss": 1.02,
+      "step": 6357
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683869079119781e-05,
+      "loss": 0.8979,
+      "step": 6358
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6837640685606173e-05,
+      "loss": 1.0195,
+      "step": 6359
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6836590438390054e-05,
+      "loss": 0.9258,
+      "step": 6360
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6835540049571205e-05,
+      "loss": 0.9058,
+      "step": 6361
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6834489519171385e-05,
+      "loss": 0.9365,
+      "step": 6362
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683343884721235e-05,
+      "loss": 0.9395,
+      "step": 6363
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683238803371586e-05,
+      "loss": 0.8677,
+      "step": 6364
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6831337078703683e-05,
+      "loss": 1.1191,
+      "step": 6365
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6830285982197585e-05,
+      "loss": 0.9551,
+      "step": 6366
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6829234744219337e-05,
+      "loss": 1.0347,
+      "step": 6367
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6828183364790717e-05,
+      "loss": 0.9585,
+      "step": 6368
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6827131843933496e-05,
+      "loss": 0.9189,
+      "step": 6369
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6826080181669455e-05,
+      "loss": 0.9902,
+      "step": 6370
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682502837802038e-05,
+      "loss": 1.0078,
+      "step": 6371
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6823976433008055e-05,
+      "loss": 1.1099,
+      "step": 6372
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6822924346654265e-05,
+      "loss": 0.9395,
+      "step": 6373
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6821872118980803e-05,
+      "loss": 0.9697,
+      "step": 6374
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682081975000947e-05,
+      "loss": 0.9434,
+      "step": 6375
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681976723976205e-05,
+      "loss": 0.8887,
+      "step": 6376
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681871458826035e-05,
+      "loss": 0.9602,
+      "step": 6377
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6817661795526178e-05,
+      "loss": 0.9746,
+      "step": 6378
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681660886158133e-05,
+      "loss": 1.0,
+      "step": 6379
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6815555786447624e-05,
+      "loss": 1.0269,
+      "step": 6380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6814502570146867e-05,
+      "loss": 0.9385,
+      "step": 6381
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6813449212700877e-05,
+      "loss": 0.9429,
+      "step": 6382
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6812395714131463e-05,
+      "loss": 0.8892,
+      "step": 6383
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6811342074460455e-05,
+      "loss": 0.9219,
+      "step": 6384
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6810288293709675e-05,
+      "loss": 0.8677,
+      "step": 6385
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6809234371900943e-05,
+      "loss": 0.9067,
+      "step": 6386
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6808180309056096e-05,
+      "loss": 0.9692,
+      "step": 6387
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6807126105196964e-05,
+      "loss": 0.999,
+      "step": 6388
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6806071760345382e-05,
+      "loss": 0.9673,
+      "step": 6389
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6805017274523187e-05,
+      "loss": 0.9443,
+      "step": 6390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680396264775222e-05,
+      "loss": 0.9502,
+      "step": 6391
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6802907880054324e-05,
+      "loss": 0.9312,
+      "step": 6392
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6801852971451348e-05,
+      "loss": 1.0474,
+      "step": 6393
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6800797921965143e-05,
+      "loss": 1.0454,
+      "step": 6394
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6799742731617555e-05,
+      "loss": 0.8181,
+      "step": 6395
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6798687400430452e-05,
+      "loss": 1.0112,
+      "step": 6396
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6797631928425676e-05,
+      "loss": 1.001,
+      "step": 6397
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6796576315625105e-05,
+      "loss": 0.9673,
+      "step": 6398
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6795520562050593e-05,
+      "loss": 0.8691,
+      "step": 6399
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679446466772401e-05,
+      "loss": 0.8833,
+      "step": 6400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6793408632667224e-05,
+      "loss": 1.0166,
+      "step": 6401
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6792352456902112e-05,
+      "loss": 1.0229,
+      "step": 6402
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6791296140450547e-05,
+      "loss": 0.834,
+      "step": 6403
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6790239683334412e-05,
+      "loss": 0.9937,
+      "step": 6404
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6789183085575585e-05,
+      "loss": 1.0308,
+      "step": 6405
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678812634719595e-05,
+      "loss": 0.9971,
+      "step": 6406
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.67870694682174e-05,
+      "loss": 1.0464,
+      "step": 6407
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6786012448661817e-05,
+      "loss": 0.9829,
+      "step": 6408
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6784955288551104e-05,
+      "loss": 0.9351,
+      "step": 6409
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6783897987907153e-05,
+      "loss": 0.8574,
+      "step": 6410
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6782840546751857e-05,
+      "loss": 0.9438,
+      "step": 6411
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678178296510713e-05,
+      "loss": 0.9067,
+      "step": 6412
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6780725242994872e-05,
+      "loss": 0.9438,
+      "step": 6413
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677966738043699e-05,
+      "loss": 0.9434,
+      "step": 6414
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677860937745539e-05,
+      "loss": 0.9775,
+      "step": 6415
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6777551234072e-05,
+      "loss": 0.978,
+      "step": 6416
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6776492950308726e-05,
+      "loss": 0.9409,
+      "step": 6417
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6775434526187487e-05,
+      "loss": 1.0181,
+      "step": 6418
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677437596173021e-05,
+      "loss": 0.9849,
+      "step": 6419
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677331725695882e-05,
+      "loss": 0.9473,
+      "step": 6420
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6772258411895243e-05,
+      "loss": 0.9966,
+      "step": 6421
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677119942656141e-05,
+      "loss": 0.9395,
+      "step": 6422
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6770140300979263e-05,
+      "loss": 1.0547,
+      "step": 6423
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676908103517073e-05,
+      "loss": 0.8093,
+      "step": 6424
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676802162915775e-05,
+      "loss": 1.0156,
+      "step": 6425
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6766962082962276e-05,
+      "loss": 0.8589,
+      "step": 6426
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6765902396606244e-05,
+      "loss": 0.8711,
+      "step": 6427
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676484257011161e-05,
+      "loss": 0.937,
+      "step": 6428
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676378260350032e-05,
+      "loss": 1.0454,
+      "step": 6429
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676272249679433e-05,
+      "loss": 0.9795,
+      "step": 6430
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6761662250015596e-05,
+      "loss": 0.9663,
+      "step": 6431
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6760601863186085e-05,
+      "loss": 0.959,
+      "step": 6432
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6759541336327757e-05,
+      "loss": 0.9419,
+      "step": 6433
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6758480669462574e-05,
+      "loss": 1.0181,
+      "step": 6434
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6757419862612507e-05,
+      "loss": 1.002,
+      "step": 6435
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675635891579953e-05,
+      "loss": 0.9756,
+      "step": 6436
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6755297829045617e-05,
+      "loss": 0.9219,
+      "step": 6437
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6754236602372747e-05,
+      "loss": 0.9746,
+      "step": 6438
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6753175235802896e-05,
+      "loss": 1.0181,
+      "step": 6439
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6752113729358052e-05,
+      "loss": 0.9009,
+      "step": 6440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6751052083060198e-05,
+      "loss": 0.9045,
+      "step": 6441
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674999029693133e-05,
+      "loss": 0.98,
+      "step": 6442
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6748928370993432e-05,
+      "loss": 0.9209,
+      "step": 6443
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6747866305268502e-05,
+      "loss": 0.9873,
+      "step": 6444
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674680409977854e-05,
+      "loss": 1.0063,
+      "step": 6445
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6745741754545545e-05,
+      "loss": 0.9014,
+      "step": 6446
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6744679269591524e-05,
+      "loss": 1.04,
+      "step": 6447
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674361664493848e-05,
+      "loss": 0.9082,
+      "step": 6448
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6742553880608425e-05,
+      "loss": 1.001,
+      "step": 6449
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6741490976623365e-05,
+      "loss": 0.8975,
+      "step": 6450
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6740427933005325e-05,
+      "loss": 1.0239,
+      "step": 6451
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673936474977632e-05,
+      "loss": 0.9536,
+      "step": 6452
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6738301426958366e-05,
+      "loss": 0.936,
+      "step": 6453
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673723796457349e-05,
+      "loss": 0.8467,
+      "step": 6454
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6736174362643725e-05,
+      "loss": 0.8906,
+      "step": 6455
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6735110621191093e-05,
+      "loss": 0.9673,
+      "step": 6456
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673404674023763e-05,
+      "loss": 1.0337,
+      "step": 6457
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673298271980537e-05,
+      "loss": 0.9795,
+      "step": 6458
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6731918559916355e-05,
+      "loss": 1.0796,
+      "step": 6459
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.673085426059262e-05,
+      "loss": 1.0249,
+      "step": 6460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6729789821856218e-05,
+      "loss": 0.9614,
+      "step": 6461
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672872524372919e-05,
+      "loss": 0.8789,
+      "step": 6462
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6727660526233587e-05,
+      "loss": 0.9507,
+      "step": 6463
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6726595669391463e-05,
+      "loss": 0.9473,
+      "step": 6464
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6725530673224876e-05,
+      "loss": 0.9795,
+      "step": 6465
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6724465537755875e-05,
+      "loss": 0.9849,
+      "step": 6466
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6723400263006535e-05,
+      "loss": 1.0737,
+      "step": 6467
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672233484899891e-05,
+      "loss": 0.9351,
+      "step": 6468
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6721269295755073e-05,
+      "loss": 0.8877,
+      "step": 6469
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6720203603297093e-05,
+      "loss": 1.0605,
+      "step": 6470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6719137771647044e-05,
+      "loss": 0.9966,
+      "step": 6471
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6718071800826998e-05,
+      "loss": 0.9351,
+      "step": 6472
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6717005690859038e-05,
+      "loss": 1.0757,
+      "step": 6473
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6715939441765247e-05,
+      "loss": 0.9556,
+      "step": 6474
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6714873053567703e-05,
+      "loss": 1.0312,
+      "step": 6475
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6713806526288498e-05,
+      "loss": 0.9482,
+      "step": 6476
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6712739859949725e-05,
+      "loss": 0.9541,
+      "step": 6477
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6711673054573473e-05,
+      "loss": 0.9009,
+      "step": 6478
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6710606110181844e-05,
+      "loss": 1.0562,
+      "step": 6479
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6709539026796927e-05,
+      "loss": 0.9976,
+      "step": 6480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6708471804440832e-05,
+      "loss": 0.8975,
+      "step": 6481
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670740444313566e-05,
+      "loss": 0.8865,
+      "step": 6482
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670633694290352e-05,
+      "loss": 0.9619,
+      "step": 6483
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6705269303766526e-05,
+      "loss": 1.0112,
+      "step": 6484
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6704201525746785e-05,
+      "loss": 0.9165,
+      "step": 6485
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6703133608866415e-05,
+      "loss": 1.0151,
+      "step": 6486
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670206555314754e-05,
+      "loss": 1.0503,
+      "step": 6487
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6700997358612276e-05,
+      "loss": 0.9136,
+      "step": 6488
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6699929025282753e-05,
+      "loss": 0.9282,
+      "step": 6489
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6698860553181095e-05,
+      "loss": 0.9644,
+      "step": 6490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669779194232944e-05,
+      "loss": 0.9473,
+      "step": 6491
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6696723192749908e-05,
+      "loss": 1.0059,
+      "step": 6492
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6695654304464645e-05,
+      "loss": 0.9658,
+      "step": 6493
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6694585277495793e-05,
+      "loss": 1.0386,
+      "step": 6494
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6693516111865486e-05,
+      "loss": 1.063,
+      "step": 6495
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6692446807595875e-05,
+      "loss": 0.9014,
+      "step": 6496
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6691377364709102e-05,
+      "loss": 1.0112,
+      "step": 6497
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6690307783227325e-05,
+      "loss": 1.0312,
+      "step": 6498
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6689238063172693e-05,
+      "loss": 0.9741,
+      "step": 6499
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6688168204567364e-05,
+      "loss": 1.1924,
+      "step": 6500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.66870982074335e-05,
+      "loss": 0.9414,
+      "step": 6501
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6686028071793258e-05,
+      "loss": 0.9331,
+      "step": 6502
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6684957797668805e-05,
+      "loss": 1.0063,
+      "step": 6503
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6683887385082312e-05,
+      "loss": 0.8611,
+      "step": 6504
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6682816834055948e-05,
+      "loss": 0.9751,
+      "step": 6505
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6681746144611884e-05,
+      "loss": 1.0181,
+      "step": 6506
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.66806753167723e-05,
+      "loss": 1.0312,
+      "step": 6507
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6679604350559374e-05,
+      "loss": 1.002,
+      "step": 6508
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667853324599529e-05,
+      "loss": 0.9829,
+      "step": 6509
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6677462003102235e-05,
+      "loss": 0.9463,
+      "step": 6510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6676390621902392e-05,
+      "loss": 1.064,
+      "step": 6511
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6675319102417953e-05,
+      "loss": 0.9619,
+      "step": 6512
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6674247444671114e-05,
+      "loss": 1.0342,
+      "step": 6513
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667317564868407e-05,
+      "loss": 1.0127,
+      "step": 6514
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667210371447902e-05,
+      "loss": 0.8445,
+      "step": 6515
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667103164207817e-05,
+      "loss": 1.0342,
+      "step": 6516
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6669959431503722e-05,
+      "loss": 0.9629,
+      "step": 6517
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6668887082777887e-05,
+      "loss": 1.0625,
+      "step": 6518
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6667814595922872e-05,
+      "loss": 0.9629,
+      "step": 6519
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6666741970960894e-05,
+      "loss": 0.8684,
+      "step": 6520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6665669207914165e-05,
+      "loss": 0.9756,
+      "step": 6521
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6664596306804914e-05,
+      "loss": 1.0249,
+      "step": 6522
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6663523267655354e-05,
+      "loss": 1.0029,
+      "step": 6523
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666245009048772e-05,
+      "loss": 1.0212,
+      "step": 6524
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6661376775324226e-05,
+      "loss": 1.0,
+      "step": 6525
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6660303322187113e-05,
+      "loss": 1.0112,
+      "step": 6526
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665922973109862e-05,
+      "loss": 0.9585,
+      "step": 6527
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665815600208097e-05,
+      "loss": 1.0703,
+      "step": 6528
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665708213515641e-05,
+      "loss": 0.9478,
+      "step": 6529
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6656008130347183e-05,
+      "loss": 1.0298,
+      "step": 6530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6654933987675533e-05,
+      "loss": 1.1304,
+      "step": 6531
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665385970716371e-05,
+      "loss": 1.0024,
+      "step": 6532
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6652785288833962e-05,
+      "loss": 1.0181,
+      "step": 6533
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6651710732708544e-05,
+      "loss": 1.0063,
+      "step": 6534
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6650636038809713e-05,
+      "loss": 0.8584,
+      "step": 6535
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6649561207159726e-05,
+      "loss": 0.9341,
+      "step": 6536
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664848623778085e-05,
+      "loss": 0.9888,
+      "step": 6537
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664741113069535e-05,
+      "loss": 0.9434,
+      "step": 6538
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6646335885925494e-05,
+      "loss": 0.9023,
+      "step": 6539
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6645260503493547e-05,
+      "loss": 0.9316,
+      "step": 6540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6644184983421788e-05,
+      "loss": 0.9512,
+      "step": 6541
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6643109325732494e-05,
+      "loss": 0.8308,
+      "step": 6542
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664203353044794e-05,
+      "loss": 1.0283,
+      "step": 6543
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6640957597590416e-05,
+      "loss": 0.8496,
+      "step": 6544
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6639881527182198e-05,
+      "loss": 0.9294,
+      "step": 6545
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6638805319245584e-05,
+      "loss": 0.9536,
+      "step": 6546
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6637728973802858e-05,
+      "loss": 0.9077,
+      "step": 6547
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6636652490876316e-05,
+      "loss": 1.0166,
+      "step": 6548
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6635575870488254e-05,
+      "loss": 1.0195,
+      "step": 6549
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6634499112660974e-05,
+      "loss": 0.9482,
+      "step": 6550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6633422217416772e-05,
+      "loss": 1.0073,
+      "step": 6551
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.663234518477796e-05,
+      "loss": 0.9272,
+      "step": 6552
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6631268014766844e-05,
+      "loss": 0.9873,
+      "step": 6553
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6630190707405737e-05,
+      "loss": 0.856,
+      "step": 6554
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6629113262716944e-05,
+      "loss": 0.9375,
+      "step": 6555
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6628035680722788e-05,
+      "loss": 0.9751,
+      "step": 6556
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662695796144559e-05,
+      "loss": 0.9136,
+      "step": 6557
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662588010490767e-05,
+      "loss": 0.9331,
+      "step": 6558
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6624802111131356e-05,
+      "loss": 0.8828,
+      "step": 6559
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6623723980138972e-05,
+      "loss": 0.855,
+      "step": 6560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6622645711952845e-05,
+      "loss": 1.0269,
+      "step": 6561
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662156730659532e-05,
+      "loss": 0.9258,
+      "step": 6562
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662048876408872e-05,
+      "loss": 0.9624,
+      "step": 6563
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6619410084455395e-05,
+      "loss": 1.0332,
+      "step": 6564
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6618331267717683e-05,
+      "loss": 0.915,
+      "step": 6565
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661725231389793e-05,
+      "loss": 0.8618,
+      "step": 6566
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6616173223018484e-05,
+      "loss": 0.8413,
+      "step": 6567
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6615093995101696e-05,
+      "loss": 0.9089,
+      "step": 6568
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6614014630169916e-05,
+      "loss": 0.9575,
+      "step": 6569
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66129351282455e-05,
+      "loss": 0.9644,
+      "step": 6570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6611855489350815e-05,
+      "loss": 1.0615,
+      "step": 6571
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6610775713508217e-05,
+      "loss": 0.9902,
+      "step": 6572
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660969580074007e-05,
+      "loss": 0.9702,
+      "step": 6573
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660861575106874e-05,
+      "loss": 1.0132,
+      "step": 6574
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6607535564516606e-05,
+      "loss": 1.0386,
+      "step": 6575
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6606455241106032e-05,
+      "loss": 0.9341,
+      "step": 6576
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66053747808594e-05,
+      "loss": 1.0947,
+      "step": 6577
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660429418379909e-05,
+      "loss": 1.0576,
+      "step": 6578
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660321344994748e-05,
+      "loss": 0.8889,
+      "step": 6579
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660213257932695e-05,
+      "loss": 1.0435,
+      "step": 6580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6601051571959897e-05,
+      "loss": 1.0962,
+      "step": 6581
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659997042786871e-05,
+      "loss": 0.855,
+      "step": 6582
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659888914707578e-05,
+      "loss": 0.9873,
+      "step": 6583
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6597807729603503e-05,
+      "loss": 0.9868,
+      "step": 6584
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6596726175474274e-05,
+      "loss": 0.9731,
+      "step": 6585
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.65956444847105e-05,
+      "loss": 0.8479,
+      "step": 6586
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6594562657334584e-05,
+      "loss": 1.0786,
+      "step": 6587
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6593480693368934e-05,
+      "loss": 0.9072,
+      "step": 6588
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6592398592835956e-05,
+      "loss": 0.938,
+      "step": 6589
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6591316355758068e-05,
+      "loss": 1.002,
+      "step": 6590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6590233982157687e-05,
+      "loss": 0.918,
+      "step": 6591
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6589151472057228e-05,
+      "loss": 0.957,
+      "step": 6592
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.658806882547911e-05,
+      "loss": 0.9644,
+      "step": 6593
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6586986042445764e-05,
+      "loss": 0.9136,
+      "step": 6594
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.658590312297961e-05,
+      "loss": 1.0435,
+      "step": 6595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6584820067103084e-05,
+      "loss": 0.9624,
+      "step": 6596
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6583736874838617e-05,
+      "loss": 1.0498,
+      "step": 6597
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6582653546208645e-05,
+      "loss": 0.9531,
+      "step": 6598
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65815700812356e-05,
+      "loss": 1.0366,
+      "step": 6599
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6580486479941936e-05,
+      "loss": 0.9805,
+      "step": 6600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6579402742350085e-05,
+      "loss": 0.8994,
+      "step": 6601
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6578318868482497e-05,
+      "loss": 0.9458,
+      "step": 6602
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6577234858361627e-05,
+      "loss": 1.0015,
+      "step": 6603
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657615071200992e-05,
+      "loss": 0.803,
+      "step": 6604
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6575066429449838e-05,
+      "loss": 0.9961,
+      "step": 6605
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657398201070384e-05,
+      "loss": 0.9678,
+      "step": 6606
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6572897455794375e-05,
+      "loss": 0.894,
+      "step": 6607
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657181276474392e-05,
+      "loss": 0.8574,
+      "step": 6608
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6570727937574936e-05,
+      "loss": 0.8936,
+      "step": 6609
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6569642974309893e-05,
+      "loss": 0.9751,
+      "step": 6610
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6568557874971263e-05,
+      "loss": 0.9404,
+      "step": 6611
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6567472639581524e-05,
+      "loss": 0.916,
+      "step": 6612
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6566387268163152e-05,
+      "loss": 0.9575,
+      "step": 6613
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6565301760738625e-05,
+      "loss": 0.8555,
+      "step": 6614
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6564216117330428e-05,
+      "loss": 0.936,
+      "step": 6615
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656313033796105e-05,
+      "loss": 0.9062,
+      "step": 6616
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656204442265298e-05,
+      "loss": 0.9351,
+      "step": 6617
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6560958371428706e-05,
+      "loss": 0.8779,
+      "step": 6618
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6559872184310724e-05,
+      "loss": 0.9883,
+      "step": 6619
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6558785861321536e-05,
+      "loss": 0.9937,
+      "step": 6620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655769940248364e-05,
+      "loss": 0.915,
+      "step": 6621
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6556612807819533e-05,
+      "loss": 0.981,
+      "step": 6622
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6555526077351732e-05,
+      "loss": 0.9302,
+      "step": 6623
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6554439211102737e-05,
+      "loss": 0.8857,
+      "step": 6624
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6553352209095065e-05,
+      "loss": 1.0244,
+      "step": 6625
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6552265071351228e-05,
+      "loss": 0.9185,
+      "step": 6626
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6551177797893742e-05,
+      "loss": 0.8862,
+      "step": 6627
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6550090388745136e-05,
+      "loss": 0.8589,
+      "step": 6628
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654900284392792e-05,
+      "loss": 1.0308,
+      "step": 6629
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654791516346463e-05,
+      "loss": 0.9619,
+      "step": 6630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6546827347377786e-05,
+      "loss": 1.0815,
+      "step": 6631
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6545739395689927e-05,
+      "loss": 0.9297,
+      "step": 6632
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6544651308423582e-05,
+      "loss": 0.874,
+      "step": 6633
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6543563085601286e-05,
+      "loss": 0.8857,
+      "step": 6634
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6542474727245587e-05,
+      "loss": 0.9897,
+      "step": 6635
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6541386233379023e-05,
+      "loss": 1.0039,
+      "step": 6636
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654029760402414e-05,
+      "loss": 0.9172,
+      "step": 6637
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6539208839203487e-05,
+      "loss": 0.8594,
+      "step": 6638
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6538119938939612e-05,
+      "loss": 1.0664,
+      "step": 6639
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653703090325507e-05,
+      "loss": 1.0669,
+      "step": 6640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653594173217242e-05,
+      "loss": 0.8616,
+      "step": 6641
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653485242571422e-05,
+      "loss": 0.958,
+      "step": 6642
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6533762983903033e-05,
+      "loss": 0.8984,
+      "step": 6643
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653267340676142e-05,
+      "loss": 0.9409,
+      "step": 6644
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6531583694311954e-05,
+      "loss": 1.0386,
+      "step": 6645
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6530493846577204e-05,
+      "loss": 0.98,
+      "step": 6646
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6529403863579742e-05,
+      "loss": 0.9897,
+      "step": 6647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6528313745342144e-05,
+      "loss": 1.0781,
+      "step": 6648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6527223491886994e-05,
+      "loss": 0.9644,
+      "step": 6649
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6526133103236865e-05,
+      "loss": 1.1211,
+      "step": 6650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652504257941435e-05,
+      "loss": 0.9746,
+      "step": 6651
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6523951920442032e-05,
+      "loss": 1.0586,
+      "step": 6652
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6522861126342507e-05,
+      "loss": 0.9312,
+      "step": 6653
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652177019713836e-05,
+      "loss": 1.0771,
+      "step": 6654
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652067913285219e-05,
+      "loss": 1.0347,
+      "step": 6655
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6519587933506596e-05,
+      "loss": 0.9653,
+      "step": 6656
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6518496599124178e-05,
+      "loss": 0.8809,
+      "step": 6657
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6517405129727547e-05,
+      "loss": 0.9473,
+      "step": 6658
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.65163135253393e-05,
+      "loss": 0.8867,
+      "step": 6659
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6515221785982053e-05,
+      "loss": 0.9487,
+      "step": 6660
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6514129911678418e-05,
+      "loss": 0.9268,
+      "step": 6661
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651303790245101e-05,
+      "loss": 1.0093,
+      "step": 6662
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6511945758322445e-05,
+      "loss": 0.9487,
+      "step": 6663
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6510853479315345e-05,
+      "loss": 0.8481,
+      "step": 6664
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6509761065452335e-05,
+      "loss": 0.9597,
+      "step": 6665
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6508668516756047e-05,
+      "loss": 1.019,
+      "step": 6666
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6507575833249098e-05,
+      "loss": 0.8359,
+      "step": 6667
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6506483014954128e-05,
+      "loss": 0.981,
+      "step": 6668
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6505390061893774e-05,
+      "loss": 0.9229,
+      "step": 6669
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6504296974090667e-05,
+      "loss": 0.958,
+      "step": 6670
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6503203751567455e-05,
+      "loss": 0.9829,
+      "step": 6671
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6502110394346776e-05,
+      "loss": 0.9634,
+      "step": 6672
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6501016902451278e-05,
+      "loss": 0.9629,
+      "step": 6673
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6499923275903605e-05,
+      "loss": 0.8418,
+      "step": 6674
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649882951472642e-05,
+      "loss": 0.9502,
+      "step": 6675
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649773561894237e-05,
+      "loss": 0.896,
+      "step": 6676
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6496641588574107e-05,
+      "loss": 0.9761,
+      "step": 6677
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.64955474236443e-05,
+      "loss": 0.9927,
+      "step": 6678
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649445312417561e-05,
+      "loss": 0.9536,
+      "step": 6679
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.64933586901907e-05,
+      "loss": 0.8882,
+      "step": 6680
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6492264121712244e-05,
+      "loss": 0.9702,
+      "step": 6681
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6491169418762905e-05,
+      "loss": 0.9194,
+      "step": 6682
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6490074581365365e-05,
+      "loss": 1.0195,
+      "step": 6683
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6488979609542293e-05,
+      "loss": 0.9238,
+      "step": 6684
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6487884503316377e-05,
+      "loss": 0.9111,
+      "step": 6685
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6486789262710294e-05,
+      "loss": 0.8232,
+      "step": 6686
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6485693887746732e-05,
+      "loss": 1.0229,
+      "step": 6687
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6484598378448374e-05,
+      "loss": 0.855,
+      "step": 6688
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6483502734837915e-05,
+      "loss": 0.9014,
+      "step": 6689
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6482406956938047e-05,
+      "loss": 1.0293,
+      "step": 6690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6481311044771466e-05,
+      "loss": 1.0933,
+      "step": 6691
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6480214998360872e-05,
+      "loss": 0.897,
+      "step": 6692
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6479118817728967e-05,
+      "loss": 1.0332,
+      "step": 6693
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6478022502898454e-05,
+      "loss": 0.894,
+      "step": 6694
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6476926053892044e-05,
+      "loss": 0.9453,
+      "step": 6695
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6475829470732443e-05,
+      "loss": 1.0889,
+      "step": 6696
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6474732753442367e-05,
+      "loss": 0.9341,
+      "step": 6697
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647363590204453e-05,
+      "loss": 0.9292,
+      "step": 6698
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6472538916561653e-05,
+      "loss": 1.082,
+      "step": 6699
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647144179701645e-05,
+      "loss": 0.978,
+      "step": 6700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647034454343165e-05,
+      "loss": 1.0342,
+      "step": 6701
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646924715582999e-05,
+      "loss": 0.9634,
+      "step": 6702
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646814963423418e-05,
+      "loss": 0.9722,
+      "step": 6703
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6467051978666963e-05,
+      "loss": 1.0063,
+      "step": 6704
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6465954189151075e-05,
+      "loss": 1.0508,
+      "step": 6705
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6464856265709257e-05,
+      "loss": 0.9761,
+      "step": 6706
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646375820836424e-05,
+      "loss": 0.8589,
+      "step": 6707
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6462660017138775e-05,
+      "loss": 0.9814,
+      "step": 6708
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.64615616920556e-05,
+      "loss": 0.9312,
+      "step": 6709
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6460463233137475e-05,
+      "loss": 1.0669,
+      "step": 6710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6459364640407146e-05,
+      "loss": 0.9922,
+      "step": 6711
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6458265913887366e-05,
+      "loss": 1.0059,
+      "step": 6712
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6457167053600894e-05,
+      "loss": 0.9595,
+      "step": 6713
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6456068059570495e-05,
+      "loss": 0.9858,
+      "step": 6714
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6454968931818926e-05,
+      "loss": 1.021,
+      "step": 6715
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6453869670368953e-05,
+      "loss": 1.0232,
+      "step": 6716
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6452770275243347e-05,
+      "loss": 1.0234,
+      "step": 6717
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6451670746464873e-05,
+      "loss": 0.9514,
+      "step": 6718
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6450571084056316e-05,
+      "loss": 0.9736,
+      "step": 6719
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644947128804044e-05,
+      "loss": 0.9795,
+      "step": 6720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6448371358440037e-05,
+      "loss": 1.0273,
+      "step": 6721
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6447271295277883e-05,
+      "loss": 0.9131,
+      "step": 6722
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644617109857676e-05,
+      "loss": 0.9473,
+      "step": 6723
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644507076835946e-05,
+      "loss": 0.9785,
+      "step": 6724
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644397030464877e-05,
+      "loss": 0.8762,
+      "step": 6725
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644286970746749e-05,
+      "loss": 1.0615,
+      "step": 6726
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644176897683841e-05,
+      "loss": 0.9976,
+      "step": 6727
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6440668112784333e-05,
+      "loss": 0.98,
+      "step": 6728
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643956711532806e-05,
+      "loss": 0.9873,
+      "step": 6729
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643846598449239e-05,
+      "loss": 1.0225,
+      "step": 6730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6437364720300136e-05,
+      "loss": 0.9097,
+      "step": 6731
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6436263322774104e-05,
+      "loss": 0.9893,
+      "step": 6732
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643516179193711e-05,
+      "loss": 0.8389,
+      "step": 6733
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643406012781197e-05,
+      "loss": 0.998,
+      "step": 6734
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6432958330421497e-05,
+      "loss": 0.8804,
+      "step": 6735
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6431856399788517e-05,
+      "loss": 1.0503,
+      "step": 6736
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643075433593585e-05,
+      "loss": 0.9268,
+      "step": 6737
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642965213888633e-05,
+      "loss": 0.946,
+      "step": 6738
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6428549808662776e-05,
+      "loss": 0.895,
+      "step": 6739
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6427447345288024e-05,
+      "loss": 1.0586,
+      "step": 6740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6426344748784914e-05,
+      "loss": 0.918,
+      "step": 6741
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6425242019176275e-05,
+      "loss": 1.0352,
+      "step": 6742
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6424139156484954e-05,
+      "loss": 0.9873,
+      "step": 6743
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642303616073379e-05,
+      "loss": 1.0601,
+      "step": 6744
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6421933031945627e-05,
+      "loss": 0.9697,
+      "step": 6745
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6420829770143318e-05,
+      "loss": 1.0156,
+      "step": 6746
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6419726375349717e-05,
+      "loss": 0.8123,
+      "step": 6747
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6418622847587668e-05,
+      "loss": 0.9985,
+      "step": 6748
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6417519186880037e-05,
+      "loss": 1.02,
+      "step": 6749
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6416415393249678e-05,
+      "loss": 0.9551,
+      "step": 6750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6415311466719455e-05,
+      "loss": 0.9907,
+      "step": 6751
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6414207407312237e-05,
+      "loss": 0.9536,
+      "step": 6752
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6413103215050886e-05,
+      "loss": 0.8638,
+      "step": 6753
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6411998889958276e-05,
+      "loss": 0.8813,
+      "step": 6754
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641089443205728e-05,
+      "loss": 0.949,
+      "step": 6755
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6409789841370772e-05,
+      "loss": 0.9751,
+      "step": 6756
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6408685117921633e-05,
+      "loss": 0.9429,
+      "step": 6757
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640758026173274e-05,
+      "loss": 0.8938,
+      "step": 6758
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6406475272826986e-05,
+      "loss": 0.8936,
+      "step": 6759
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6405370151227248e-05,
+      "loss": 0.9375,
+      "step": 6760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6404264896956425e-05,
+      "loss": 1.0811,
+      "step": 6761
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6403159510037406e-05,
+      "loss": 0.9736,
+      "step": 6762
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6402053990493083e-05,
+      "loss": 1.1035,
+      "step": 6763
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6400948338346357e-05,
+      "loss": 0.9668,
+      "step": 6764
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6399842553620128e-05,
+      "loss": 0.9482,
+      "step": 6765
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.63987366363373e-05,
+      "loss": 0.9375,
+      "step": 6766
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6397630586520784e-05,
+      "loss": 0.9092,
+      "step": 6767
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6396524404193477e-05,
+      "loss": 0.9619,
+      "step": 6768
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.63954180893783e-05,
+      "loss": 0.9702,
+      "step": 6769
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639431164209817e-05,
+      "loss": 1.0225,
+      "step": 6770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6393205062375995e-05,
+      "loss": 1.0122,
+      "step": 6771
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6392098350234703e-05,
+      "loss": 1.0015,
+      "step": 6772
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6390991505697214e-05,
+      "loss": 1.0483,
+      "step": 6773
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6389884528786453e-05,
+      "loss": 0.9673,
+      "step": 6774
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6388777419525344e-05,
+      "loss": 0.9055,
+      "step": 6775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6387670177936828e-05,
+      "loss": 0.938,
+      "step": 6776
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6386562804043832e-05,
+      "loss": 0.8872,
+      "step": 6777
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638545529786929e-05,
+      "loss": 0.9897,
+      "step": 6778
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638434765943615e-05,
+      "loss": 0.9883,
+      "step": 6779
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6383239888767347e-05,
+      "loss": 0.8906,
+      "step": 6780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6382131985885827e-05,
+      "loss": 1.0747,
+      "step": 6781
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638102395081454e-05,
+      "loss": 1.0132,
+      "step": 6782
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6379915783576427e-05,
+      "loss": 0.8853,
+      "step": 6783
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6378807484194454e-05,
+      "loss": 0.8628,
+      "step": 6784
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637769905269157e-05,
+      "loss": 0.9136,
+      "step": 6785
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637659048909073e-05,
+      "loss": 0.895,
+      "step": 6786
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6375481793414902e-05,
+      "loss": 1.0151,
+      "step": 6787
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6374372965687047e-05,
+      "loss": 0.9424,
+      "step": 6788
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6373264005930128e-05,
+      "loss": 1.0278,
+      "step": 6789
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637215491416712e-05,
+      "loss": 0.9697,
+      "step": 6790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6371045690420997e-05,
+      "loss": 0.8203,
+      "step": 6791
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6369936334714724e-05,
+      "loss": 0.9463,
+      "step": 6792
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6368826847071287e-05,
+      "loss": 0.9922,
+      "step": 6793
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636771722751366e-05,
+      "loss": 0.9043,
+      "step": 6794
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636660747606483e-05,
+      "loss": 0.9243,
+      "step": 6795
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6365497592747784e-05,
+      "loss": 0.9185,
+      "step": 6796
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636438757758551e-05,
+      "loss": 0.8984,
+      "step": 6797
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636327743060099e-05,
+      "loss": 0.9233,
+      "step": 6798
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636216715181723e-05,
+      "loss": 0.8638,
+      "step": 6799
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6361056741257222e-05,
+      "loss": 0.936,
+      "step": 6800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6359946198943965e-05,
+      "loss": 1.0679,
+      "step": 6801
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6358835524900462e-05,
+      "loss": 0.9043,
+      "step": 6802
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6357724719149715e-05,
+      "loss": 0.8984,
+      "step": 6803
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6356613781714738e-05,
+      "loss": 1.083,
+      "step": 6804
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6355502712618534e-05,
+      "loss": 0.9604,
+      "step": 6805
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635439151188412e-05,
+      "loss": 0.9795,
+      "step": 6806
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635328017953451e-05,
+      "loss": 1.0449,
+      "step": 6807
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6352168715592724e-05,
+      "loss": 0.981,
+      "step": 6808
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635105712008178e-05,
+      "loss": 0.9683,
+      "step": 6809
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6349945393024708e-05,
+      "loss": 0.9536,
+      "step": 6810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634883353444453e-05,
+      "loss": 0.8828,
+      "step": 6811
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6347721544364277e-05,
+      "loss": 0.9736,
+      "step": 6812
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6346609422806977e-05,
+      "loss": 0.9834,
+      "step": 6813
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634549716979567e-05,
+      "loss": 1.0566,
+      "step": 6814
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6344384785353392e-05,
+      "loss": 1.0869,
+      "step": 6815
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634327226950318e-05,
+      "loss": 1.0537,
+      "step": 6816
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6342159622268085e-05,
+      "loss": 0.9062,
+      "step": 6817
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6341046843671145e-05,
+      "loss": 1.0098,
+      "step": 6818
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.633993393373541e-05,
+      "loss": 1.0005,
+      "step": 6819
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633882089248393e-05,
+      "loss": 0.9399,
+      "step": 6820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6337707719939767e-05,
+      "loss": 0.9761,
+      "step": 6821
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6336594416125968e-05,
+      "loss": 0.9185,
+      "step": 6822
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6335480981065596e-05,
+      "loss": 1.0449,
+      "step": 6823
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633436741478171e-05,
+      "loss": 1.0142,
+      "step": 6824
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6333253717297383e-05,
+      "loss": 0.9629,
+      "step": 6825
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633213988863567e-05,
+      "loss": 0.9565,
+      "step": 6826
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6331025928819655e-05,
+      "loss": 0.915,
+      "step": 6827
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.63299118378724e-05,
+      "loss": 0.8989,
+      "step": 6828
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632879761581699e-05,
+      "loss": 0.9131,
+      "step": 6829
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6327683262676487e-05,
+      "loss": 0.8613,
+      "step": 6830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6326568778473992e-05,
+      "loss": 1.0366,
+      "step": 6831
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6325454163232575e-05,
+      "loss": 0.917,
+      "step": 6832
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632433941697533e-05,
+      "loss": 0.8945,
+      "step": 6833
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6323224539725337e-05,
+      "loss": 1.1206,
+      "step": 6834
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.63221095315057e-05,
+      "loss": 0.8745,
+      "step": 6835
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6320994392339507e-05,
+      "loss": 1.0405,
+      "step": 6836
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6319879122249855e-05,
+      "loss": 1.0688,
+      "step": 6837
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6318763721259846e-05,
+      "loss": 0.9517,
+      "step": 6838
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6317648189392578e-05,
+      "loss": 0.9888,
+      "step": 6839
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6316532526671164e-05,
+      "loss": 0.9014,
+      "step": 6840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631541673311871e-05,
+      "loss": 1.0146,
+      "step": 6841
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6314300808758317e-05,
+      "loss": 0.9839,
+      "step": 6842
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6313184753613112e-05,
+      "loss": 0.8955,
+      "step": 6843
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6312068567706206e-05,
+      "loss": 1.0024,
+      "step": 6844
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6310952251060714e-05,
+      "loss": 0.9473,
+      "step": 6845
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6309835803699765e-05,
+      "loss": 1.0137,
+      "step": 6846
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6308719225646478e-05,
+      "loss": 0.9668,
+      "step": 6847
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6307602516923986e-05,
+      "loss": 0.8613,
+      "step": 6848
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630648567755541e-05,
+      "loss": 0.9546,
+      "step": 6849
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630536870756389e-05,
+      "loss": 0.9736,
+      "step": 6850
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630425160697256e-05,
+      "loss": 1.0337,
+      "step": 6851
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6303134375804554e-05,
+      "loss": 0.9053,
+      "step": 6852
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6302017014083014e-05,
+      "loss": 0.9175,
+      "step": 6853
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6300899521831085e-05,
+      "loss": 0.895,
+      "step": 6854
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6299781899071915e-05,
+      "loss": 0.9331,
+      "step": 6855
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629866414582865e-05,
+      "loss": 0.9282,
+      "step": 6856
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629754626212444e-05,
+      "loss": 1.0098,
+      "step": 6857
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629642824798244e-05,
+      "loss": 1.0044,
+      "step": 6858
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6295310103425806e-05,
+      "loss": 0.9878,
+      "step": 6859
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6294191828477704e-05,
+      "loss": 0.8677,
+      "step": 6860
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6293073423161287e-05,
+      "loss": 0.9473,
+      "step": 6861
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629195488749973e-05,
+      "loss": 1.0405,
+      "step": 6862
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629083622151619e-05,
+      "loss": 0.9653,
+      "step": 6863
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.628971742523384e-05,
+      "loss": 0.9805,
+      "step": 6864
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628859849867586e-05,
+      "loss": 0.8804,
+      "step": 6865
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6287479441865422e-05,
+      "loss": 0.9785,
+      "step": 6866
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6286360254825703e-05,
+      "loss": 0.9805,
+      "step": 6867
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6285240937579882e-05,
+      "loss": 0.9324,
+      "step": 6868
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6284121490151146e-05,
+      "loss": 1.0308,
+      "step": 6869
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628300191256268e-05,
+      "loss": 0.8843,
+      "step": 6870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6281882204837678e-05,
+      "loss": 1.0415,
+      "step": 6871
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6280762366999326e-05,
+      "loss": 1.042,
+      "step": 6872
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6279642399070822e-05,
+      "loss": 0.9199,
+      "step": 6873
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627852230107536e-05,
+      "loss": 0.8953,
+      "step": 6874
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6277402073036148e-05,
+      "loss": 0.9517,
+      "step": 6875
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6276281714976377e-05,
+      "loss": 1.0225,
+      "step": 6876
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6275161226919255e-05,
+      "loss": 0.9204,
+      "step": 6877
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6274040608888e-05,
+      "loss": 0.8818,
+      "step": 6878
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6272919860905812e-05,
+      "loss": 0.9365,
+      "step": 6879
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627179898299591e-05,
+      "loss": 1.0864,
+      "step": 6880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6270677975181514e-05,
+      "loss": 1.0342,
+      "step": 6881
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6269556837485833e-05,
+      "loss": 1.1035,
+      "step": 6882
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626843556993209e-05,
+      "loss": 0.9814,
+      "step": 6883
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6267314172543516e-05,
+      "loss": 0.9111,
+      "step": 6884
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6266192645343335e-05,
+      "loss": 1.0757,
+      "step": 6885
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6265070988354773e-05,
+      "loss": 1.061,
+      "step": 6886
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6263949201601068e-05,
+      "loss": 1.0186,
+      "step": 6887
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626282728510545e-05,
+      "loss": 1.0137,
+      "step": 6888
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626170523889116e-05,
+      "loss": 0.9902,
+      "step": 6889
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6260583062981435e-05,
+      "loss": 0.9995,
+      "step": 6890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625946075739952e-05,
+      "loss": 0.9741,
+      "step": 6891
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6258338322168663e-05,
+      "loss": 1.0381,
+      "step": 6892
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6257215757312112e-05,
+      "loss": 1.1318,
+      "step": 6893
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6256093062853113e-05,
+      "loss": 0.9136,
+      "step": 6894
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6254970238814926e-05,
+      "loss": 0.9487,
+      "step": 6895
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6253847285220807e-05,
+      "loss": 1.0405,
+      "step": 6896
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6252724202094004e-05,
+      "loss": 0.9126,
+      "step": 6897
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6251600989457796e-05,
+      "loss": 0.9399,
+      "step": 6898
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6250477647335436e-05,
+      "loss": 0.9233,
+      "step": 6899
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6249354175750193e-05,
+      "loss": 0.9561,
+      "step": 6900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624823057472534e-05,
+      "loss": 0.8687,
+      "step": 6901
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6247106844284148e-05,
+      "loss": 0.9033,
+      "step": 6902
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6245982984449895e-05,
+      "loss": 0.8752,
+      "step": 6903
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624485899524585e-05,
+      "loss": 0.7979,
+      "step": 6904
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6243734876695304e-05,
+      "loss": 1.1538,
+      "step": 6905
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6242610628821536e-05,
+      "loss": 0.917,
+      "step": 6906
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624148625164783e-05,
+      "loss": 0.8677,
+      "step": 6907
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6240361745197474e-05,
+      "loss": 0.9966,
+      "step": 6908
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6239237109493767e-05,
+      "loss": 0.8704,
+      "step": 6909
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6238112344559995e-05,
+      "loss": 0.9248,
+      "step": 6910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6236987450419458e-05,
+      "loss": 0.9424,
+      "step": 6911
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6235862427095456e-05,
+      "loss": 0.9604,
+      "step": 6912
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623473727461129e-05,
+      "loss": 0.9409,
+      "step": 6913
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6233611992990263e-05,
+      "loss": 0.9893,
+      "step": 6914
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6232486582255682e-05,
+      "loss": 0.9414,
+      "step": 6915
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6231361042430864e-05,
+      "loss": 0.9399,
+      "step": 6916
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6230235373539112e-05,
+      "loss": 1.0044,
+      "step": 6917
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6229109575603748e-05,
+      "loss": 0.8618,
+      "step": 6918
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6227983648648088e-05,
+      "loss": 1.0532,
+      "step": 6919
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6226857592695453e-05,
+      "loss": 1.0688,
+      "step": 6920
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6225731407769167e-05,
+      "loss": 0.9707,
+      "step": 6921
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6224605093892553e-05,
+      "loss": 0.9355,
+      "step": 6922
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6223478651088946e-05,
+      "loss": 0.8262,
+      "step": 6923
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6222352079381666e-05,
+      "loss": 0.9731,
+      "step": 6924
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6221225378794064e-05,
+      "loss": 0.9214,
+      "step": 6925
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622009854934946e-05,
+      "loss": 1.0381,
+      "step": 6926
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6218971591071208e-05,
+      "loss": 0.9897,
+      "step": 6927
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6217844503982635e-05,
+      "loss": 0.8818,
+      "step": 6928
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6216717288107096e-05,
+      "loss": 0.9521,
+      "step": 6929
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621558994346794e-05,
+      "loss": 0.9111,
+      "step": 6930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621446247008851e-05,
+      "loss": 1.0269,
+      "step": 6931
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6213334867992167e-05,
+      "loss": 0.9526,
+      "step": 6932
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6212207137202254e-05,
+      "loss": 1.0215,
+      "step": 6933
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6211079277742138e-05,
+      "loss": 0.9116,
+      "step": 6934
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6209951289635183e-05,
+      "loss": 1.0103,
+      "step": 6935
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6208823172904743e-05,
+      "loss": 0.9688,
+      "step": 6936
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620769492757419e-05,
+      "loss": 1.1011,
+      "step": 6937
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6206566553666893e-05,
+      "loss": 0.8779,
+      "step": 6938
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620543805120622e-05,
+      "loss": 0.9458,
+      "step": 6939
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620430942021555e-05,
+      "loss": 0.98,
+      "step": 6940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6203180660718255e-05,
+      "loss": 0.9639,
+      "step": 6941
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6202051772737717e-05,
+      "loss": 1.0762,
+      "step": 6942
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6200922756297317e-05,
+      "loss": 0.9294,
+      "step": 6943
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6199793611420438e-05,
+      "loss": 0.9116,
+      "step": 6944
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6198664338130473e-05,
+      "loss": 0.9888,
+      "step": 6945
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6197534936450805e-05,
+      "loss": 0.9746,
+      "step": 6946
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6196405406404833e-05,
+      "loss": 0.8472,
+      "step": 6947
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6195275748015947e-05,
+      "loss": 1.0229,
+      "step": 6948
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619414596130755e-05,
+      "loss": 1.041,
+      "step": 6949
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619301604630304e-05,
+      "loss": 0.9863,
+      "step": 6950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6191886003025817e-05,
+      "loss": 0.9414,
+      "step": 6951
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6190755831499292e-05,
+      "loss": 0.9233,
+      "step": 6952
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6189625531746874e-05,
+      "loss": 0.894,
+      "step": 6953
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.618849510379197e-05,
+      "loss": 0.9849,
+      "step": 6954
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6187364547658e-05,
+      "loss": 0.9971,
+      "step": 6955
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6186233863368375e-05,
+      "loss": 0.8926,
+      "step": 6956
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6185103050946518e-05,
+      "loss": 0.9346,
+      "step": 6957
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6183972110415847e-05,
+      "loss": 0.7402,
+      "step": 6958
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618284104179979e-05,
+      "loss": 0.9082,
+      "step": 6959
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6181709845121772e-05,
+      "loss": 0.8682,
+      "step": 6960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6180578520405227e-05,
+      "loss": 0.9692,
+      "step": 6961
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617944706767358e-05,
+      "loss": 0.9399,
+      "step": 6962
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6178315486950275e-05,
+      "loss": 1.0547,
+      "step": 6963
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6177183778258742e-05,
+      "loss": 1.0107,
+      "step": 6964
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617605194162243e-05,
+      "loss": 1.0151,
+      "step": 6965
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6174919977064775e-05,
+      "loss": 0.9077,
+      "step": 6966
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617378788460922e-05,
+      "loss": 0.9995,
+      "step": 6967
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6172655664279225e-05,
+      "loss": 0.9888,
+      "step": 6968
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617152331609823e-05,
+      "loss": 1.1206,
+      "step": 6969
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.61703908400897e-05,
+      "loss": 1.0039,
+      "step": 6970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6169258236277076e-05,
+      "loss": 0.8447,
+      "step": 6971
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6168125504683833e-05,
+      "loss": 0.96,
+      "step": 6972
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616699264533342e-05,
+      "loss": 0.8623,
+      "step": 6973
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616585965824931e-05,
+      "loss": 0.9541,
+      "step": 6974
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6164726543454966e-05,
+      "loss": 1.0249,
+      "step": 6975
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6163593300973855e-05,
+      "loss": 0.9365,
+      "step": 6976
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616245993082946e-05,
+      "loss": 0.9438,
+      "step": 6977
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616132643304524e-05,
+      "loss": 0.9482,
+      "step": 6978
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6160192807644685e-05,
+      "loss": 0.8638,
+      "step": 6979
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615905905465127e-05,
+      "loss": 1.0327,
+      "step": 6980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615792517408848e-05,
+      "loss": 0.9062,
+      "step": 6981
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6156791165979802e-05,
+      "loss": 1.0029,
+      "step": 6982
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615565703034872e-05,
+      "loss": 0.9419,
+      "step": 6983
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6154522767218726e-05,
+      "loss": 0.8286,
+      "step": 6984
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615338837661331e-05,
+      "loss": 0.8965,
+      "step": 6985
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6152253858555976e-05,
+      "loss": 0.9448,
+      "step": 6986
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615111921307022e-05,
+      "loss": 0.9722,
+      "step": 6987
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614998444017954e-05,
+      "loss": 0.9761,
+      "step": 6988
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6148849539907438e-05,
+      "loss": 0.9946,
+      "step": 6989
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614771451227743e-05,
+      "loss": 0.9429,
+      "step": 6990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6146579357313014e-05,
+      "loss": 0.9087,
+      "step": 6991
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6145444075037714e-05,
+      "loss": 0.9175,
+      "step": 6992
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6144308665475032e-05,
+      "loss": 1.0093,
+      "step": 6993
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6143173128648493e-05,
+      "loss": 1.0132,
+      "step": 6994
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6142037464581617e-05,
+      "loss": 0.9702,
+      "step": 6995
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6140901673297924e-05,
+      "loss": 0.9775,
+      "step": 6996
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6139765754820938e-05,
+      "loss": 0.9453,
+      "step": 6997
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6138629709174186e-05,
+      "loss": 1.062,
+      "step": 6998
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6137493536381205e-05,
+      "loss": 0.9619,
+      "step": 6999
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613635723646552e-05,
+      "loss": 0.9182,
+      "step": 7000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6135220809450674e-05,
+      "loss": 0.9546,
+      "step": 7001
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6134084255360198e-05,
+      "loss": 1.0449,
+      "step": 7002
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6132947574217636e-05,
+      "loss": 0.9771,
+      "step": 7003
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613181076604653e-05,
+      "loss": 0.936,
+      "step": 7004
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6130673830870428e-05,
+      "loss": 1.0435,
+      "step": 7005
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612953676871288e-05,
+      "loss": 0.9136,
+      "step": 7006
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6128399579597436e-05,
+      "loss": 0.856,
+      "step": 7007
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6127262263547647e-05,
+      "loss": 0.9814,
+      "step": 7008
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6126124820587078e-05,
+      "loss": 0.9897,
+      "step": 7009
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6124987250739276e-05,
+      "loss": 0.9907,
+      "step": 7010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6123849554027812e-05,
+      "loss": 0.9629,
+      "step": 7011
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612271173047625e-05,
+      "loss": 0.9487,
+      "step": 7012
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6121573780108154e-05,
+      "loss": 0.9634,
+      "step": 7013
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6120435702947098e-05,
+      "loss": 0.9761,
+      "step": 7014
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6119297499016645e-05,
+      "loss": 0.9124,
+      "step": 7015
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611815916834038e-05,
+      "loss": 1.0117,
+      "step": 7016
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6117020710941874e-05,
+      "loss": 1.0659,
+      "step": 7017
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6115882126844712e-05,
+      "loss": 1.0283,
+      "step": 7018
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6114743416072478e-05,
+      "loss": 0.9473,
+      "step": 7019
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6113604578648754e-05,
+      "loss": 0.9688,
+      "step": 7020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6112465614597125e-05,
+      "loss": 1.0044,
+      "step": 7021
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6111326523941188e-05,
+      "loss": 0.9253,
+      "step": 7022
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6110187306704533e-05,
+      "loss": 1.0312,
+      "step": 7023
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610904796291076e-05,
+      "loss": 0.998,
+      "step": 7024
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6107908492583463e-05,
+      "loss": 0.9854,
+      "step": 7025
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6106768895746245e-05,
+      "loss": 0.9731,
+      "step": 7026
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610562917242271e-05,
+      "loss": 0.9521,
+      "step": 7027
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6104489322636464e-05,
+      "loss": 0.9438,
+      "step": 7028
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6103349346411118e-05,
+      "loss": 0.96,
+      "step": 7029
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610220924377028e-05,
+      "loss": 1.0098,
+      "step": 7030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6101069014737566e-05,
+      "loss": 1.0566,
+      "step": 7031
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6099928659336594e-05,
+      "loss": 0.8774,
+      "step": 7032
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6098788177590984e-05,
+      "loss": 1.0117,
+      "step": 7033
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6097647569524356e-05,
+      "loss": 1.0107,
+      "step": 7034
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6096506835160334e-05,
+      "loss": 1.0122,
+      "step": 7035
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6095365974522552e-05,
+      "loss": 0.9897,
+      "step": 7036
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609422498763463e-05,
+      "loss": 1.0264,
+      "step": 7037
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6093083874520214e-05,
+      "loss": 1.0229,
+      "step": 7038
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6091942635202925e-05,
+      "loss": 0.9404,
+      "step": 7039
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6090801269706408e-05,
+      "loss": 0.8745,
+      "step": 7040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6089659778054305e-05,
+      "loss": 0.9263,
+      "step": 7041
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6088518160270258e-05,
+      "loss": 1.0635,
+      "step": 7042
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6087376416377913e-05,
+      "loss": 0.9624,
+      "step": 7043
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6086234546400912e-05,
+      "loss": 0.9702,
+      "step": 7044
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6085092550362914e-05,
+      "loss": 0.9854,
+      "step": 7045
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608395042828757e-05,
+      "loss": 0.9971,
+      "step": 7046
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6082808180198534e-05,
+      "loss": 0.9956,
+      "step": 7047
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608166580611947e-05,
+      "loss": 0.9087,
+      "step": 7048
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6080523306074033e-05,
+      "loss": 0.8945,
+      "step": 7049
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607938068008589e-05,
+      "loss": 0.9258,
+      "step": 7050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6078237928178706e-05,
+      "loss": 1.0029,
+      "step": 7051
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6077095050376155e-05,
+      "loss": 0.8481,
+      "step": 7052
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6075952046701902e-05,
+      "loss": 0.9038,
+      "step": 7053
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607480891717963e-05,
+      "loss": 0.9268,
+      "step": 7054
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6073665661833008e-05,
+      "loss": 1.0474,
+      "step": 7055
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607252228068572e-05,
+      "loss": 0.8984,
+      "step": 7056
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6071378773761446e-05,
+      "loss": 0.936,
+      "step": 7057
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607023514108387e-05,
+      "loss": 0.8784,
+      "step": 7058
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606909138267669e-05,
+      "loss": 0.897,
+      "step": 7059
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606794749856358e-05,
+      "loss": 0.9688,
+      "step": 7060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6066803488768238e-05,
+      "loss": 0.8904,
+      "step": 7061
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6065659353314368e-05,
+      "loss": 0.9521,
+      "step": 7062
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6064515092225658e-05,
+      "loss": 0.9277,
+      "step": 7063
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606337070552581e-05,
+      "loss": 0.9795,
+      "step": 7064
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6062226193238534e-05,
+      "loss": 1.0059,
+      "step": 7065
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6061081555387526e-05,
+      "loss": 0.8918,
+      "step": 7066
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.60599367919965e-05,
+      "loss": 0.9023,
+      "step": 7067
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6058791903089166e-05,
+      "loss": 1.0371,
+      "step": 7068
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6057646888689234e-05,
+      "loss": 0.9141,
+      "step": 7069
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605650174882043e-05,
+      "loss": 0.9595,
+      "step": 7070
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6055356483506462e-05,
+      "loss": 1.0283,
+      "step": 7071
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6054211092771056e-05,
+      "loss": 1.0083,
+      "step": 7072
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6053065576637935e-05,
+      "loss": 0.9995,
+      "step": 7073
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6051919935130828e-05,
+      "loss": 0.8225,
+      "step": 7074
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6050774168273457e-05,
+      "loss": 0.897,
+      "step": 7075
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604962827608956e-05,
+      "loss": 1.0176,
+      "step": 7076
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604848225860287e-05,
+      "loss": 0.9521,
+      "step": 7077
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6047336115837122e-05,
+      "loss": 0.9399,
+      "step": 7078
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6046189847816062e-05,
+      "loss": 0.9399,
+      "step": 7079
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604504345456342e-05,
+      "loss": 0.9785,
+      "step": 7080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604389693610295e-05,
+      "loss": 0.9946,
+      "step": 7081
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.60427502924584e-05,
+      "loss": 0.8926,
+      "step": 7082
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6041603523653516e-05,
+      "loss": 1.0068,
+      "step": 7083
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6040456629712048e-05,
+      "loss": 0.9756,
+      "step": 7084
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6039309610657756e-05,
+      "loss": 1.0425,
+      "step": 7085
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6038162466514395e-05,
+      "loss": 0.8867,
+      "step": 7086
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6037015197305727e-05,
+      "loss": 1.0,
+      "step": 7087
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6035867803055515e-05,
+      "loss": 0.9531,
+      "step": 7088
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6034720283787522e-05,
+      "loss": 0.8054,
+      "step": 7089
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6033572639525517e-05,
+      "loss": 0.9814,
+      "step": 7090
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.603242487029327e-05,
+      "loss": 0.9741,
+      "step": 7091
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6031276976114556e-05,
+      "loss": 0.9214,
+      "step": 7092
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6030128957013146e-05,
+      "loss": 1.0098,
+      "step": 7093
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6028980813012825e-05,
+      "loss": 0.9736,
+      "step": 7094
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602783254413737e-05,
+      "loss": 1.0151,
+      "step": 7095
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6026684150410568e-05,
+      "loss": 0.9971,
+      "step": 7096
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60255356318562e-05,
+      "loss": 0.916,
+      "step": 7097
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6024386988498055e-05,
+      "loss": 0.9888,
+      "step": 7098
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6023238220359927e-05,
+      "loss": 0.9307,
+      "step": 7099
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602208932746561e-05,
+      "loss": 0.9443,
+      "step": 7100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60209403098389e-05,
+      "loss": 0.9648,
+      "step": 7101
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6019791167503592e-05,
+      "loss": 0.9321,
+      "step": 7102
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60186419004835e-05,
+      "loss": 1.0161,
+      "step": 7103
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6017492508802414e-05,
+      "loss": 0.8022,
+      "step": 7104
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6016342992484143e-05,
+      "loss": 1.082,
+      "step": 7105
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60151933515525e-05,
+      "loss": 1.0215,
+      "step": 7106
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6014043586031302e-05,
+      "loss": 0.9888,
+      "step": 7107
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601289369594435e-05,
+      "loss": 0.9092,
+      "step": 7108
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6011743681315473e-05,
+      "loss": 1.0229,
+      "step": 7109
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6010593542168486e-05,
+      "loss": 0.9365,
+      "step": 7110
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6009443278527212e-05,
+      "loss": 0.9907,
+      "step": 7111
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6008292890415476e-05,
+      "loss": 0.9194,
+      "step": 7112
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6007142377857104e-05,
+      "loss": 0.9561,
+      "step": 7113
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600599174087593e-05,
+      "loss": 0.9907,
+      "step": 7114
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6004840979495777e-05,
+      "loss": 0.9751,
+      "step": 7115
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600369009374049e-05,
+      "loss": 1.0142,
+      "step": 7116
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6002539083633903e-05,
+      "loss": 1.0391,
+      "step": 7117
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600138794919986e-05,
+      "loss": 0.9722,
+      "step": 7118
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6000236690462195e-05,
+      "loss": 1.0239,
+      "step": 7119
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599908530744476e-05,
+      "loss": 1.0112,
+      "step": 7120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.59979338001714e-05,
+      "loss": 1.043,
+      "step": 7121
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599678216866597e-05,
+      "loss": 1.0889,
+      "step": 7122
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5995630412952325e-05,
+      "loss": 0.978,
+      "step": 7123
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599447853305431e-05,
+      "loss": 0.9717,
+      "step": 7124
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599332652899579e-05,
+      "loss": 0.9419,
+      "step": 7125
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5992174400800627e-05,
+      "loss": 1.0117,
+      "step": 7126
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5991022148492682e-05,
+      "loss": 0.873,
+      "step": 7127
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598986977209582e-05,
+      "loss": 0.9185,
+      "step": 7128
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5988717271633913e-05,
+      "loss": 0.9424,
+      "step": 7129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598756464713083e-05,
+      "loss": 0.9771,
+      "step": 7130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5986411898610447e-05,
+      "loss": 0.958,
+      "step": 7131
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5985259026096637e-05,
+      "loss": 0.8887,
+      "step": 7132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5984106029613283e-05,
+      "loss": 0.9077,
+      "step": 7133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5982952909184262e-05,
+      "loss": 0.8921,
+      "step": 7134
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5981799664833457e-05,
+      "loss": 0.8162,
+      "step": 7135
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.598064629658476e-05,
+      "loss": 1.0303,
+      "step": 7136
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5979492804462057e-05,
+      "loss": 1.0234,
+      "step": 7137
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5978339188489238e-05,
+      "loss": 0.9351,
+      "step": 7138
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5977185448690204e-05,
+      "loss": 1.0381,
+      "step": 7139
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597603158508884e-05,
+      "loss": 0.9888,
+      "step": 7140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597487759770906e-05,
+      "loss": 0.8462,
+      "step": 7141
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597372348657475e-05,
+      "loss": 1.0664,
+      "step": 7142
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597256925170983e-05,
+      "loss": 0.9326,
+      "step": 7143
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5971414893138197e-05,
+      "loss": 0.9722,
+      "step": 7144
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5970260410883766e-05,
+      "loss": 0.9736,
+      "step": 7145
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5969105804970443e-05,
+      "loss": 0.8921,
+      "step": 7146
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596795107542215e-05,
+      "loss": 0.8979,
+      "step": 7147
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59667962222628e-05,
+      "loss": 0.938,
+      "step": 7148
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5965641245516315e-05,
+      "loss": 0.9741,
+      "step": 7149
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5964486145206612e-05,
+      "loss": 0.9263,
+      "step": 7150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5963330921357624e-05,
+      "loss": 0.895,
+      "step": 7151
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5962175573993272e-05,
+      "loss": 1.0273,
+      "step": 7152
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5961020103137493e-05,
+      "loss": 0.9707,
+      "step": 7153
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595986450881421e-05,
+      "loss": 0.9824,
+      "step": 7154
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5958708791047368e-05,
+      "loss": 0.9014,
+      "step": 7155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59575529498609e-05,
+      "loss": 0.8359,
+      "step": 7156
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595639698527874e-05,
+      "loss": 0.824,
+      "step": 7157
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5955240897324846e-05,
+      "loss": 0.8467,
+      "step": 7158
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595408468602315e-05,
+      "loss": 0.8613,
+      "step": 7159
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595292835139761e-05,
+      "loss": 1.001,
+      "step": 7160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5951771893472168e-05,
+      "loss": 1.0576,
+      "step": 7161
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5950615312270782e-05,
+      "loss": 1.0786,
+      "step": 7162
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5949458607817408e-05,
+      "loss": 0.9316,
+      "step": 7163
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5948301780136003e-05,
+      "loss": 1.0176,
+      "step": 7164
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5947144829250526e-05,
+      "loss": 0.9873,
+      "step": 7165
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594598775518494e-05,
+      "loss": 1.0039,
+      "step": 7166
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5944830557963214e-05,
+      "loss": 0.9873,
+      "step": 7167
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5943673237609317e-05,
+      "loss": 0.9619,
+      "step": 7168
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5942515794147215e-05,
+      "loss": 1.1162,
+      "step": 7169
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5941358227600885e-05,
+      "loss": 0.9248,
+      "step": 7170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5940200537994305e-05,
+      "loss": 0.9805,
+      "step": 7171
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593904272535145e-05,
+      "loss": 0.9907,
+      "step": 7172
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59378847896963e-05,
+      "loss": 0.9673,
+      "step": 7173
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5936726731052843e-05,
+      "loss": 1.0352,
+      "step": 7174
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5935568549445065e-05,
+      "loss": 0.9937,
+      "step": 7175
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5934410244896953e-05,
+      "loss": 0.9473,
+      "step": 7176
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5933251817432495e-05,
+      "loss": 0.9619,
+      "step": 7177
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5932093267075692e-05,
+      "loss": 0.9644,
+      "step": 7178
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5930934593850537e-05,
+      "loss": 1.0171,
+      "step": 7179
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592977579778103e-05,
+      "loss": 0.9902,
+      "step": 7180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5928616878891167e-05,
+      "loss": 0.8672,
+      "step": 7181
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5927457837204955e-05,
+      "loss": 0.9766,
+      "step": 7182
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5926298672746406e-05,
+      "loss": 1.02,
+      "step": 7183
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592513938553953e-05,
+      "loss": 0.9907,
+      "step": 7184
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5923979975608328e-05,
+      "loss": 1.0513,
+      "step": 7185
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592282044297682e-05,
+      "loss": 1.0308,
+      "step": 7186
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5921660787669027e-05,
+      "loss": 0.9834,
+      "step": 7187
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5920501009708963e-05,
+      "loss": 0.8289,
+      "step": 7188
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5919341109120646e-05,
+      "loss": 1.0264,
+      "step": 7189
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5918181085928112e-05,
+      "loss": 0.8657,
+      "step": 7190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591702094015538e-05,
+      "loss": 1.0034,
+      "step": 7191
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591586067182648e-05,
+      "loss": 0.9961,
+      "step": 7192
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591470028096545e-05,
+      "loss": 0.8394,
+      "step": 7193
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5913539767596313e-05,
+      "loss": 0.9438,
+      "step": 7194
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5912379131743117e-05,
+      "loss": 0.9756,
+      "step": 7195
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5911218373429894e-05,
+      "loss": 0.9839,
+      "step": 7196
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5910057492680696e-05,
+      "loss": 0.9678,
+      "step": 7197
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5908896489519556e-05,
+      "loss": 0.9375,
+      "step": 7198
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5907735363970528e-05,
+      "loss": 0.9409,
+      "step": 7199
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.590657411605766e-05,
+      "loss": 0.8887,
+      "step": 7200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5905412745805007e-05,
+      "loss": 0.979,
+      "step": 7201
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.590425125323662e-05,
+      "loss": 0.9961,
+      "step": 7202
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5903089638376555e-05,
+      "loss": 0.9009,
+      "step": 7203
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5901927901248878e-05,
+      "loss": 0.9541,
+      "step": 7204
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.590076604187765e-05,
+      "loss": 0.9023,
+      "step": 7205
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5899604060286933e-05,
+      "loss": 0.8948,
+      "step": 7206
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5898441956500797e-05,
+      "loss": 1.1406,
+      "step": 7207
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5897279730543308e-05,
+      "loss": 1.0649,
+      "step": 7208
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5896117382438545e-05,
+      "loss": 0.8999,
+      "step": 7209
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5894954912210578e-05,
+      "loss": 0.8965,
+      "step": 7210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589379231988349e-05,
+      "loss": 0.9307,
+      "step": 7211
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589262960548135e-05,
+      "loss": 1.0156,
+      "step": 7212
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5891466769028252e-05,
+      "loss": 1.0029,
+      "step": 7213
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5890303810548274e-05,
+      "loss": 1.0488,
+      "step": 7214
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588914073006551e-05,
+      "loss": 0.9971,
+      "step": 7215
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588797752760405e-05,
+      "loss": 1.0239,
+      "step": 7216
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5886814203187985e-05,
+      "loss": 1.0659,
+      "step": 7217
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5885650756841403e-05,
+      "loss": 1.0464,
+      "step": 7218
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5884487188588414e-05,
+      "loss": 0.8911,
+      "step": 7219
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588332349845311e-05,
+      "loss": 0.9902,
+      "step": 7220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5882159686459597e-05,
+      "loss": 0.9204,
+      "step": 7221
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588099575263198e-05,
+      "loss": 0.9941,
+      "step": 7222
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5879831696994367e-05,
+      "loss": 0.9253,
+      "step": 7223
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.587866751957087e-05,
+      "loss": 0.8999,
+      "step": 7224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5877503220385603e-05,
+      "loss": 0.9751,
+      "step": 7225
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5876338799462675e-05,
+      "loss": 0.9299,
+      "step": 7226
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5875174256826206e-05,
+      "loss": 0.9561,
+      "step": 7227
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.587400959250032e-05,
+      "loss": 0.9326,
+      "step": 7228
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.587284480650914e-05,
+      "loss": 1.0239,
+      "step": 7229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5871679898876797e-05,
+      "loss": 0.9106,
+      "step": 7230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5870514869627404e-05,
+      "loss": 1.0029,
+      "step": 7231
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.58693497187851e-05,
+      "loss": 0.9053,
+      "step": 7232
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586818444637402e-05,
+      "loss": 0.9526,
+      "step": 7233
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.58670190524183e-05,
+      "loss": 1.0337,
+      "step": 7234
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5865853536942074e-05,
+      "loss": 1.0415,
+      "step": 7235
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586468789996949e-05,
+      "loss": 0.9541,
+      "step": 7236
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5863522141524682e-05,
+      "loss": 0.9541,
+      "step": 7237
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.58623562616318e-05,
+      "loss": 0.999,
+      "step": 7238
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586119026031499e-05,
+      "loss": 0.9814,
+      "step": 7239
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5860024137598408e-05,
+      "loss": 0.9722,
+      "step": 7240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.58588578935062e-05,
+      "loss": 1.0635,
+      "step": 7241
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5857691528062533e-05,
+      "loss": 0.9771,
+      "step": 7242
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585652504129155e-05,
+      "loss": 0.8481,
+      "step": 7243
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5855358433217426e-05,
+      "loss": 0.8677,
+      "step": 7244
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5854191703864318e-05,
+      "loss": 0.9922,
+      "step": 7245
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5853024853256387e-05,
+      "loss": 0.918,
+      "step": 7246
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5851857881417815e-05,
+      "loss": 1.0176,
+      "step": 7247
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5850690788372758e-05,
+      "loss": 0.9116,
+      "step": 7248
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.58495235741454e-05,
+      "loss": 1.0361,
+      "step": 7249
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584835623875991e-05,
+      "loss": 0.9727,
+      "step": 7250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5847188782240473e-05,
+      "loss": 0.8896,
+      "step": 7251
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5846021204611265e-05,
+      "loss": 1.0396,
+      "step": 7252
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5844853505896468e-05,
+      "loss": 1.022,
+      "step": 7253
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5843685686120273e-05,
+      "loss": 1.0415,
+      "step": 7254
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5842517745306865e-05,
+      "loss": 0.9722,
+      "step": 7255
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5841349683480437e-05,
+      "loss": 1.0596,
+      "step": 7256
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584018150066518e-05,
+      "loss": 0.9707,
+      "step": 7257
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583901319688529e-05,
+      "loss": 1.0327,
+      "step": 7258
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583784477216497e-05,
+      "loss": 0.8501,
+      "step": 7259
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5836676226528416e-05,
+      "loss": 1.0518,
+      "step": 7260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5835507559999835e-05,
+      "loss": 1.0396,
+      "step": 7261
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583433877260343e-05,
+      "loss": 1.0469,
+      "step": 7262
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5833169864363412e-05,
+      "loss": 0.8726,
+      "step": 7263
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5832000835303987e-05,
+      "loss": 0.9221,
+      "step": 7264
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5830831685449377e-05,
+      "loss": 0.9868,
+      "step": 7265
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.582966241482379e-05,
+      "loss": 0.9336,
+      "step": 7266
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.582849302345145e-05,
+      "loss": 1.0059,
+      "step": 7267
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5827323511356575e-05,
+      "loss": 0.9658,
+      "step": 7268
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.582615387856339e-05,
+      "loss": 0.9634,
+      "step": 7269
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5824984125096115e-05,
+      "loss": 0.9629,
+      "step": 7270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.582381425097899e-05,
+      "loss": 1.0059,
+      "step": 7271
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5822644256236235e-05,
+      "loss": 0.9121,
+      "step": 7272
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5821474140892092e-05,
+      "loss": 1.0044,
+      "step": 7273
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5820303904970794e-05,
+      "loss": 0.8818,
+      "step": 7274
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5819133548496574e-05,
+      "loss": 1.062,
+      "step": 7275
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581796307149368e-05,
+      "loss": 0.918,
+      "step": 7276
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5816792473986355e-05,
+      "loss": 0.9111,
+      "step": 7277
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581562175599884e-05,
+      "loss": 0.9351,
+      "step": 7278
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5814450917555388e-05,
+      "loss": 0.9678,
+      "step": 7279
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5813279958680247e-05,
+      "loss": 1.0786,
+      "step": 7280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5812108879397676e-05,
+      "loss": 0.856,
+      "step": 7281
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5810937679731925e-05,
+      "loss": 1.0405,
+      "step": 7282
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580976635970725e-05,
+      "loss": 0.9326,
+      "step": 7283
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580859491934792e-05,
+      "loss": 0.8916,
+      "step": 7284
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5807423358678195e-05,
+      "loss": 0.9253,
+      "step": 7285
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580625167772234e-05,
+      "loss": 1.0029,
+      "step": 7286
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5805079876504627e-05,
+      "loss": 0.9634,
+      "step": 7287
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580390795504932e-05,
+      "loss": 0.9604,
+      "step": 7288
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.58027359133807e-05,
+      "loss": 0.9028,
+      "step": 7289
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5801563751523033e-05,
+      "loss": 0.9868,
+      "step": 7290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580039146950061e-05,
+      "loss": 1.1431,
+      "step": 7291
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.57992190673377e-05,
+      "loss": 0.9326,
+      "step": 7292
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5798046545058596e-05,
+      "loss": 0.9204,
+      "step": 7293
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5796873902687583e-05,
+      "loss": 1.1196,
+      "step": 7294
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5795701140248942e-05,
+      "loss": 1.0415,
+      "step": 7295
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5794528257766965e-05,
+      "loss": 0.938,
+      "step": 7296
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5793355255265953e-05,
+      "loss": 0.959,
+      "step": 7297
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5792182132770194e-05,
+      "loss": 0.9033,
+      "step": 7298
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.579100889030399e-05,
+      "loss": 0.8438,
+      "step": 7299
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5789835527891643e-05,
+      "loss": 0.9868,
+      "step": 7300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578866204555745e-05,
+      "loss": 0.9043,
+      "step": 7301
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5787488443325726e-05,
+      "loss": 0.9922,
+      "step": 7302
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5786314721220772e-05,
+      "loss": 0.9062,
+      "step": 7303
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.57851408792669e-05,
+      "loss": 0.9448,
+      "step": 7304
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578396691748842e-05,
+      "loss": 1.0063,
+      "step": 7305
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578279283590966e-05,
+      "loss": 0.9062,
+      "step": 7306
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5781618634554923e-05,
+      "loss": 0.9033,
+      "step": 7307
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578044431344854e-05,
+      "loss": 1.0293,
+      "step": 7308
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577926987261483e-05,
+      "loss": 0.8896,
+      "step": 7309
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577809531207812e-05,
+      "loss": 0.8789,
+      "step": 7310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577692063186274e-05,
+      "loss": 0.9058,
+      "step": 7311
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5775745831993012e-05,
+      "loss": 0.8921,
+      "step": 7312
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5774570912493277e-05,
+      "loss": 1.0103,
+      "step": 7313
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577339587338787e-05,
+      "loss": 0.9624,
+      "step": 7314
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5772220714701125e-05,
+      "loss": 0.9351,
+      "step": 7315
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.577104543645738e-05,
+      "loss": 1.0562,
+      "step": 7316
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576987003868099e-05,
+      "loss": 0.9834,
+      "step": 7317
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576869452139629e-05,
+      "loss": 0.9951,
+      "step": 7318
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5767518884627634e-05,
+      "loss": 0.9922,
+      "step": 7319
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5766343128399366e-05,
+      "loss": 0.9697,
+      "step": 7320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5765167252735843e-05,
+      "loss": 0.8979,
+      "step": 7321
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5763991257661418e-05,
+      "loss": 0.8171,
+      "step": 7322
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5762815143200453e-05,
+      "loss": 0.9766,
+      "step": 7323
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5761638909377302e-05,
+      "loss": 1.0845,
+      "step": 7324
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576046255621633e-05,
+      "loss": 0.9326,
+      "step": 7325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575928608374191e-05,
+      "loss": 0.894,
+      "step": 7326
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5758109491978398e-05,
+      "loss": 0.874,
+      "step": 7327
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575693278095017e-05,
+      "loss": 1.0649,
+      "step": 7328
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5755755950681597e-05,
+      "loss": 0.8784,
+      "step": 7329
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5754579001197057e-05,
+      "loss": 0.8535,
+      "step": 7330
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5753401932520923e-05,
+      "loss": 0.9111,
+      "step": 7331
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5752224744677578e-05,
+      "loss": 0.9243,
+      "step": 7332
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57510474376914e-05,
+      "loss": 0.9922,
+      "step": 7333
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574987001158678e-05,
+      "loss": 0.8765,
+      "step": 7334
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574869246638811e-05,
+      "loss": 0.9731,
+      "step": 7335
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5747514802119767e-05,
+      "loss": 0.9614,
+      "step": 7336
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574633701880615e-05,
+      "loss": 0.8977,
+      "step": 7337
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5745159116471652e-05,
+      "loss": 0.9653,
+      "step": 7338
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5743981095140674e-05,
+      "loss": 0.8921,
+      "step": 7339
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5742802954837612e-05,
+      "loss": 0.8618,
+      "step": 7340
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574162469558687e-05,
+      "loss": 0.9922,
+      "step": 7341
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574044631741285e-05,
+      "loss": 0.9932,
+      "step": 7342
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573926782033996e-05,
+      "loss": 0.9692,
+      "step": 7343
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5738089204392615e-05,
+      "loss": 0.9331,
+      "step": 7344
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5736910469595217e-05,
+      "loss": 0.9507,
+      "step": 7345
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5735731615972196e-05,
+      "loss": 0.9419,
+      "step": 7346
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573455264354795e-05,
+      "loss": 1.0576,
+      "step": 7347
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5733373552346912e-05,
+      "loss": 0.9331,
+      "step": 7348
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5732194342393498e-05,
+      "loss": 0.9585,
+      "step": 7349
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5731015013712135e-05,
+      "loss": 0.7671,
+      "step": 7350
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572983556632725e-05,
+      "loss": 0.96,
+      "step": 7351
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5728656000263272e-05,
+      "loss": 0.9258,
+      "step": 7352
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5727476315544627e-05,
+      "loss": 0.9277,
+      "step": 7353
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572629651219576e-05,
+      "loss": 1.0522,
+      "step": 7354
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5725116590241096e-05,
+      "loss": 0.9053,
+      "step": 7355
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5723936549705086e-05,
+      "loss": 0.9951,
+      "step": 7356
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572275639061216e-05,
+      "loss": 0.9292,
+      "step": 7357
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5721576112986768e-05,
+      "loss": 0.9473,
+      "step": 7358
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5720395716853357e-05,
+      "loss": 0.9351,
+      "step": 7359
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5719215202236378e-05,
+      "loss": 0.915,
+      "step": 7360
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5718034569160273e-05,
+      "loss": 1.186,
+      "step": 7361
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5716853817649506e-05,
+      "loss": 0.9985,
+      "step": 7362
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571567294772853e-05,
+      "loss": 1.021,
+      "step": 7363
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.57144919594218e-05,
+      "loss": 0.9961,
+      "step": 7364
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5713310852753782e-05,
+      "loss": 1.0044,
+      "step": 7365
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571212962774894e-05,
+      "loss": 1.0093,
+      "step": 7366
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5710948284431733e-05,
+      "loss": 0.895,
+      "step": 7367
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5709766822826636e-05,
+      "loss": 0.9521,
+      "step": 7368
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5708585242958122e-05,
+      "loss": 1.0293,
+      "step": 7369
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5707403544850656e-05,
+      "loss": 0.9316,
+      "step": 7370
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5706221728528724e-05,
+      "loss": 1.0649,
+      "step": 7371
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.57050397940168e-05,
+      "loss": 1.0386,
+      "step": 7372
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570385774133936e-05,
+      "loss": 0.9346,
+      "step": 7373
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570267557052089e-05,
+      "loss": 0.8877,
+      "step": 7374
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570149328158588e-05,
+      "loss": 0.9429,
+      "step": 7375
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5700310874558814e-05,
+      "loss": 0.9365,
+      "step": 7376
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5699128349464185e-05,
+      "loss": 0.9282,
+      "step": 7377
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5697945706326484e-05,
+      "loss": 1.0059,
+      "step": 7378
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5696762945170206e-05,
+      "loss": 0.957,
+      "step": 7379
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569558006601985e-05,
+      "loss": 0.979,
+      "step": 7380
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5694397068899915e-05,
+      "loss": 1.0161,
+      "step": 7381
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5693213953834908e-05,
+      "loss": 1.0156,
+      "step": 7382
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569203072084933e-05,
+      "loss": 0.9858,
+      "step": 7383
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5690847369967695e-05,
+      "loss": 1.0093,
+      "step": 7384
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5689663901214506e-05,
+      "loss": 1.0146,
+      "step": 7385
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5688480314614274e-05,
+      "loss": 1.0015,
+      "step": 7386
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568729661019152e-05,
+      "loss": 0.9614,
+      "step": 7387
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568611278797076e-05,
+      "loss": 0.9946,
+      "step": 7388
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5684928847976514e-05,
+      "loss": 0.9653,
+      "step": 7389
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.56837447902333e-05,
+      "loss": 0.9497,
+      "step": 7390
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5682560614765653e-05,
+      "loss": 0.9556,
+      "step": 7391
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5681376321598088e-05,
+      "loss": 1.0381,
+      "step": 7392
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568019191075514e-05,
+      "loss": 0.9727,
+      "step": 7393
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5679007382261344e-05,
+      "loss": 0.8564,
+      "step": 7394
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5677822736141228e-05,
+      "loss": 0.8765,
+      "step": 7395
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5676637972419336e-05,
+      "loss": 0.96,
+      "step": 7396
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5675453091120204e-05,
+      "loss": 0.9551,
+      "step": 7397
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5674268092268372e-05,
+      "loss": 0.9065,
+      "step": 7398
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5673082975888388e-05,
+      "loss": 0.9072,
+      "step": 7399
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5671897742004795e-05,
+      "loss": 0.9272,
+      "step": 7400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5670712390642148e-05,
+      "loss": 0.9536,
+      "step": 7401
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566952692182499e-05,
+      "loss": 0.979,
+      "step": 7402
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566834133557788e-05,
+      "loss": 0.9185,
+      "step": 7403
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5667155631925378e-05,
+      "loss": 0.9834,
+      "step": 7404
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5665969810892037e-05,
+      "loss": 0.9932,
+      "step": 7405
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5664783872502418e-05,
+      "loss": 0.9937,
+      "step": 7406
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.566359781678109e-05,
+      "loss": 0.9766,
+      "step": 7407
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5662411643752613e-05,
+      "loss": 1.0298,
+      "step": 7408
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5661225353441563e-05,
+      "loss": 0.9707,
+      "step": 7409
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5660038945872503e-05,
+      "loss": 0.9785,
+      "step": 7410
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565885242107001e-05,
+      "loss": 0.9229,
+      "step": 7411
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5657665779058663e-05,
+      "loss": 1.0503,
+      "step": 7412
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5656479019863032e-05,
+      "loss": 0.959,
+      "step": 7413
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5655292143507707e-05,
+      "loss": 0.9341,
+      "step": 7414
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565410515001727e-05,
+      "loss": 0.9541,
+      "step": 7415
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5652918039416302e-05,
+      "loss": 0.9683,
+      "step": 7416
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565173081172939e-05,
+      "loss": 1.0439,
+      "step": 7417
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5650543466981125e-05,
+      "loss": 0.9648,
+      "step": 7418
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564935600519611e-05,
+      "loss": 0.9771,
+      "step": 7419
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5648168426398923e-05,
+      "loss": 1.0435,
+      "step": 7420
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5646980730614177e-05,
+      "loss": 0.9351,
+      "step": 7421
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5645792917866464e-05,
+      "loss": 0.9697,
+      "step": 7422
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564460498818039e-05,
+      "loss": 0.8579,
+      "step": 7423
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5643416941580558e-05,
+      "loss": 0.9692,
+      "step": 7424
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5642228778091574e-05,
+      "loss": 0.8518,
+      "step": 7425
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564104049773805e-05,
+      "loss": 0.9487,
+      "step": 7426
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5639852100544604e-05,
+      "loss": 0.9785,
+      "step": 7427
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5638663586535837e-05,
+      "loss": 0.9058,
+      "step": 7428
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563747495573638e-05,
+      "loss": 0.9814,
+      "step": 7429
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5636286208170842e-05,
+      "loss": 1.0142,
+      "step": 7430
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563509734386385e-05,
+      "loss": 0.877,
+      "step": 7431
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5633908362840032e-05,
+      "loss": 1.0034,
+      "step": 7432
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5632719265124005e-05,
+      "loss": 0.8862,
+      "step": 7433
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5631530050740404e-05,
+      "loss": 0.9722,
+      "step": 7434
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5630340719713857e-05,
+      "loss": 0.98,
+      "step": 7435
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562915127206901e-05,
+      "loss": 0.8989,
+      "step": 7436
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5627961707830482e-05,
+      "loss": 0.9731,
+      "step": 7437
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5626772027022924e-05,
+      "loss": 0.9355,
+      "step": 7438
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5625582229670974e-05,
+      "loss": 0.9487,
+      "step": 7439
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5624392315799274e-05,
+      "loss": 0.8074,
+      "step": 7440
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562320228543247e-05,
+      "loss": 1.0288,
+      "step": 7441
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5622012138595214e-05,
+      "loss": 0.9214,
+      "step": 7442
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562082187531215e-05,
+      "loss": 0.978,
+      "step": 7443
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5619631495607943e-05,
+      "loss": 1.0601,
+      "step": 7444
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5618440999507235e-05,
+      "loss": 0.9434,
+      "step": 7445
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5617250387034694e-05,
+      "loss": 0.7642,
+      "step": 7446
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5616059658214977e-05,
+      "loss": 0.8511,
+      "step": 7447
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5614868813072745e-05,
+      "loss": 1.063,
+      "step": 7448
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5613677851632668e-05,
+      "loss": 0.8794,
+      "step": 7449
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.561248677391941e-05,
+      "loss": 0.895,
+      "step": 7450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5611295579957642e-05,
+      "loss": 1.0083,
+      "step": 7451
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5610104269772034e-05,
+      "loss": 0.939,
+      "step": 7452
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5608912843387272e-05,
+      "loss": 0.9473,
+      "step": 7453
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560772130082802e-05,
+      "loss": 0.9365,
+      "step": 7454
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5606529642118963e-05,
+      "loss": 0.9697,
+      "step": 7455
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5605337867284786e-05,
+      "loss": 1.0269,
+      "step": 7456
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560414597635017e-05,
+      "loss": 0.9375,
+      "step": 7457
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.56029539693398e-05,
+      "loss": 0.8359,
+      "step": 7458
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560176184627837e-05,
+      "loss": 1.0107,
+      "step": 7459
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5600569607190574e-05,
+      "loss": 0.9785,
+      "step": 7460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.55993772521011e-05,
+      "loss": 0.8369,
+      "step": 7461
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5598184781034646e-05,
+      "loss": 0.917,
+      "step": 7462
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5596992194015913e-05,
+      "loss": 0.8052,
+      "step": 7463
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5595799491069602e-05,
+      "loss": 0.959,
+      "step": 7464
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5594606672220415e-05,
+      "loss": 1.0732,
+      "step": 7465
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5593413737493062e-05,
+      "loss": 0.9053,
+      "step": 7466
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.559222068691225e-05,
+      "loss": 0.918,
+      "step": 7467
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5591027520502685e-05,
+      "loss": 0.8916,
+      "step": 7468
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5589834238289085e-05,
+      "loss": 0.9175,
+      "step": 7469
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558864084029617e-05,
+      "loss": 1.0596,
+      "step": 7470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558744732654865e-05,
+      "loss": 0.959,
+      "step": 7471
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558625369707125e-05,
+      "loss": 1.1309,
+      "step": 7472
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5585059951888696e-05,
+      "loss": 1.0708,
+      "step": 7473
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5583866091025707e-05,
+      "loss": 0.8618,
+      "step": 7474
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5582672114507012e-05,
+      "loss": 0.8989,
+      "step": 7475
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5581478022357346e-05,
+      "loss": 1.0391,
+      "step": 7476
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558028381460144e-05,
+      "loss": 1.0317,
+      "step": 7477
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5579089491264023e-05,
+      "loss": 0.9365,
+      "step": 7478
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5577895052369836e-05,
+      "loss": 0.8804,
+      "step": 7479
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5576700497943625e-05,
+      "loss": 0.9302,
+      "step": 7480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5575505828010126e-05,
+      "loss": 0.9419,
+      "step": 7481
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557431104259408e-05,
+      "loss": 0.9033,
+      "step": 7482
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557311614172024e-05,
+      "loss": 0.9404,
+      "step": 7483
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5571921125413355e-05,
+      "loss": 0.8613,
+      "step": 7484
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5570725993698178e-05,
+      "loss": 0.8728,
+      "step": 7485
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5569530746599454e-05,
+      "loss": 0.9463,
+      "step": 7486
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556833538414195e-05,
+      "loss": 0.9731,
+      "step": 7487
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5567139906350423e-05,
+      "loss": 1.0459,
+      "step": 7488
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5565944313249633e-05,
+      "loss": 0.8335,
+      "step": 7489
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5564748604864342e-05,
+      "loss": 0.9634,
+      "step": 7490
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5563552781219314e-05,
+      "loss": 0.9639,
+      "step": 7491
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5562356842339322e-05,
+      "loss": 0.9019,
+      "step": 7492
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556116078824914e-05,
+      "loss": 0.9355,
+      "step": 7493
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5559964618973533e-05,
+      "loss": 1.0488,
+      "step": 7494
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555876833453728e-05,
+      "loss": 1.0986,
+      "step": 7495
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5557571934965158e-05,
+      "loss": 0.9185,
+      "step": 7496
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5556375420281954e-05,
+      "loss": 0.9531,
+      "step": 7497
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555517879051244e-05,
+      "loss": 0.9741,
+      "step": 7498
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555398204568141e-05,
+      "loss": 0.9424,
+      "step": 7499
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555278518581365e-05,
+      "loss": 0.8259,
+      "step": 7500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5551588210933944e-05,
+      "loss": 1.0,
+      "step": 7501
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555039112106709e-05,
+      "loss": 0.9478,
+      "step": 7502
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554919391623788e-05,
+      "loss": 0.998,
+      "step": 7503
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5547996596471115e-05,
+      "loss": 0.9951,
+      "step": 7504
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5546799161791596e-05,
+      "loss": 0.9097,
+      "step": 7505
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5545601612224113e-05,
+      "loss": 0.8806,
+      "step": 7506
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554440394779348e-05,
+      "loss": 0.9517,
+      "step": 7507
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5543206168524503e-05,
+      "loss": 1.0439,
+      "step": 7508
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554200827444199e-05,
+      "loss": 0.9326,
+      "step": 7509
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5540810265570748e-05,
+      "loss": 0.8491,
+      "step": 7510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5539612141935595e-05,
+      "loss": 1.0381,
+      "step": 7511
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553841390356135e-05,
+      "loss": 0.9565,
+      "step": 7512
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5537215550472826e-05,
+      "loss": 1.0259,
+      "step": 7513
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5536017082694846e-05,
+      "loss": 0.8965,
+      "step": 7514
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5534818500252233e-05,
+      "loss": 0.9209,
+      "step": 7515
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5533619803169815e-05,
+      "loss": 0.8555,
+      "step": 7516
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5532420991472416e-05,
+      "loss": 1.0059,
+      "step": 7517
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553122206518487e-05,
+      "loss": 0.8745,
+      "step": 7518
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5530023024332006e-05,
+      "loss": 0.9575,
+      "step": 7519
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552882386893866e-05,
+      "loss": 1.0635,
+      "step": 7520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5527624599029675e-05,
+      "loss": 0.959,
+      "step": 7521
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5526425214629885e-05,
+      "loss": 0.9263,
+      "step": 7522
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552522571576413e-05,
+      "loss": 0.9365,
+      "step": 7523
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5524026102457258e-05,
+      "loss": 0.9761,
+      "step": 7524
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5522826374734123e-05,
+      "loss": 1.0566,
+      "step": 7525
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5521626532619564e-05,
+      "loss": 0.8657,
+      "step": 7526
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5520426576138433e-05,
+      "loss": 0.9766,
+      "step": 7527
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5519226505315587e-05,
+      "loss": 0.9329,
+      "step": 7528
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5518026320175886e-05,
+      "loss": 1.0059,
+      "step": 7529
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5516826020744184e-05,
+      "loss": 0.9741,
+      "step": 7530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5515625607045345e-05,
+      "loss": 1.0234,
+      "step": 7531
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551442507910423e-05,
+      "loss": 0.8926,
+      "step": 7532
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5513224436945708e-05,
+      "loss": 0.9033,
+      "step": 7533
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5512023680594642e-05,
+      "loss": 0.9409,
+      "step": 7534
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551082281007591e-05,
+      "loss": 1.0615,
+      "step": 7535
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5509621825414378e-05,
+      "loss": 0.9341,
+      "step": 7536
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.550842072663493e-05,
+      "loss": 0.9927,
+      "step": 7537
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.550721951376243e-05,
+      "loss": 1.0742,
+      "step": 7538
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5506018186821773e-05,
+      "loss": 0.9863,
+      "step": 7539
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.550481674583783e-05,
+      "loss": 1.0356,
+      "step": 7540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5503615190835492e-05,
+      "loss": 1.0356,
+      "step": 7541
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5502413521839646e-05,
+      "loss": 0.9487,
+      "step": 7542
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5501211738875178e-05,
+      "loss": 0.9736,
+      "step": 7543
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5500009841966984e-05,
+      "loss": 1.0088,
+      "step": 7544
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5498807831139953e-05,
+      "loss": 0.8943,
+      "step": 7545
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.549760570641899e-05,
+      "loss": 0.9438,
+      "step": 7546
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5496403467828986e-05,
+      "loss": 0.9175,
+      "step": 7547
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5495201115394845e-05,
+      "loss": 1.0454,
+      "step": 7548
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5493998649141476e-05,
+      "loss": 1.0269,
+      "step": 7549
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5492796069093776e-05,
+      "loss": 0.9131,
+      "step": 7550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5491593375276656e-05,
+      "loss": 1.0132,
+      "step": 7551
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5490390567715033e-05,
+      "loss": 0.9541,
+      "step": 7552
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5489187646433812e-05,
+      "loss": 0.9395,
+      "step": 7553
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548798461145791e-05,
+      "loss": 0.8789,
+      "step": 7554
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5486781462812253e-05,
+      "loss": 0.9404,
+      "step": 7555
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5485578200521748e-05,
+      "loss": 0.8984,
+      "step": 7556
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5484374824611327e-05,
+      "loss": 0.9683,
+      "step": 7557
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5483171335105914e-05,
+      "loss": 0.9316,
+      "step": 7558
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5481967732030435e-05,
+      "loss": 1.0127,
+      "step": 7559
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5480764015409817e-05,
+      "loss": 0.9282,
+      "step": 7560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5479560185268996e-05,
+      "loss": 1.04,
+      "step": 7561
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5478356241632903e-05,
+      "loss": 1.0918,
+      "step": 7562
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5477152184526477e-05,
+      "loss": 0.8535,
+      "step": 7563
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5475948013974658e-05,
+      "loss": 0.9873,
+      "step": 7564
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5474743730002383e-05,
+      "loss": 0.8843,
+      "step": 7565
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.54735393326346e-05,
+      "loss": 0.8799,
+      "step": 7566
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5472334821896252e-05,
+      "loss": 0.9353,
+      "step": 7567
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5471130197812287e-05,
+      "loss": 0.9478,
+      "step": 7568
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546992546040766e-05,
+      "loss": 1.0137,
+      "step": 7569
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546872060970732e-05,
+      "loss": 1.0083,
+      "step": 7570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5467515645736223e-05,
+      "loss": 0.9917,
+      "step": 7571
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546631056851933e-05,
+      "loss": 0.9712,
+      "step": 7572
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5465105378081598e-05,
+      "loss": 0.9233,
+      "step": 7573
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5463900074447992e-05,
+      "loss": 1.0186,
+      "step": 7574
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5462694657643472e-05,
+      "loss": 1.001,
+      "step": 7575
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5461489127693013e-05,
+      "loss": 1.0059,
+      "step": 7576
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5460283484621575e-05,
+      "loss": 0.8271,
+      "step": 7577
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5459077728454138e-05,
+      "loss": 0.9868,
+      "step": 7578
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545787185921567e-05,
+      "loss": 0.9282,
+      "step": 7579
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5456665876931155e-05,
+      "loss": 0.9551,
+      "step": 7580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545545978162556e-05,
+      "loss": 0.9326,
+      "step": 7581
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545425357332388e-05,
+      "loss": 0.9697,
+      "step": 7582
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545304725205109e-05,
+      "loss": 0.9717,
+      "step": 7583
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545184081783218e-05,
+      "loss": 1.0117,
+      "step": 7584
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5450634270692133e-05,
+      "loss": 0.9639,
+      "step": 7585
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5449427610655944e-05,
+      "loss": 0.9658,
+      "step": 7586
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5448220837748606e-05,
+      "loss": 0.9863,
+      "step": 7587
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544701395199511e-05,
+      "loss": 0.9902,
+      "step": 7588
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5445806953420462e-05,
+      "loss": 0.9717,
+      "step": 7589
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544459984204965e-05,
+      "loss": 1.0972,
+      "step": 7590
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5443392617907686e-05,
+      "loss": 0.8496,
+      "step": 7591
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5442185281019574e-05,
+      "loss": 1.001,
+      "step": 7592
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5440977831410316e-05,
+      "loss": 0.8965,
+      "step": 7593
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5439770269104924e-05,
+      "loss": 0.9043,
+      "step": 7594
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543856259412841e-05,
+      "loss": 0.9741,
+      "step": 7595
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5437354806505785e-05,
+      "loss": 0.9604,
+      "step": 7596
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5436146906262067e-05,
+      "loss": 0.8789,
+      "step": 7597
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5434938893422276e-05,
+      "loss": 0.9507,
+      "step": 7598
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5433730768011434e-05,
+      "loss": 1.0742,
+      "step": 7599
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5432522530054562e-05,
+      "loss": 0.9277,
+      "step": 7600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5431314179576684e-05,
+      "loss": 0.9155,
+      "step": 7601
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543010571660283e-05,
+      "loss": 0.9277,
+      "step": 7602
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5428897141158033e-05,
+      "loss": 1.0117,
+      "step": 7603
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542768845326732e-05,
+      "loss": 0.9463,
+      "step": 7604
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542647965295573e-05,
+      "loss": 0.9717,
+      "step": 7605
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5425270740248297e-05,
+      "loss": 0.9468,
+      "step": 7606
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5424061715170063e-05,
+      "loss": 1.0225,
+      "step": 7607
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542285257774607e-05,
+      "loss": 0.9175,
+      "step": 7608
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542164332800136e-05,
+      "loss": 0.9834,
+      "step": 7609
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5420433965960984e-05,
+      "loss": 0.8613,
+      "step": 7610
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5419224491649987e-05,
+      "loss": 0.9258,
+      "step": 7611
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5418014905093422e-05,
+      "loss": 0.9961,
+      "step": 7612
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541680520631634e-05,
+      "loss": 0.9834,
+      "step": 7613
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54155953953438e-05,
+      "loss": 1.0195,
+      "step": 7614
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5414385472200856e-05,
+      "loss": 0.8809,
+      "step": 7615
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5413175436912572e-05,
+      "loss": 1.0044,
+      "step": 7616
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541196528950401e-05,
+      "loss": 0.7793,
+      "step": 7617
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541075503000024e-05,
+      "loss": 0.9126,
+      "step": 7618
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540954465842632e-05,
+      "loss": 0.9873,
+      "step": 7619
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5408334174807325e-05,
+      "loss": 0.8296,
+      "step": 7620
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540712357916833e-05,
+      "loss": 0.9639,
+      "step": 7621
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5405912871534402e-05,
+      "loss": 0.9336,
+      "step": 7622
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540470205193062e-05,
+      "loss": 0.9639,
+      "step": 7623
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5403491120382072e-05,
+      "loss": 0.9126,
+      "step": 7624
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540228007691383e-05,
+      "loss": 0.9375,
+      "step": 7625
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5401068921550975e-05,
+      "loss": 0.9741,
+      "step": 7626
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5399857654318603e-05,
+      "loss": 1.0977,
+      "step": 7627
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.53986462752418e-05,
+      "loss": 0.939,
+      "step": 7628
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5397434784345647e-05,
+      "loss": 0.9409,
+      "step": 7629
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5396223181655248e-05,
+      "loss": 0.9351,
+      "step": 7630
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5395011467195694e-05,
+      "loss": 1.022,
+      "step": 7631
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5393799640992083e-05,
+      "loss": 0.9941,
+      "step": 7632
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5392587703069513e-05,
+      "loss": 1.0674,
+      "step": 7633
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5391375653453093e-05,
+      "loss": 0.9287,
+      "step": 7634
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5390163492167917e-05,
+      "loss": 0.9556,
+      "step": 7635
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53889512192391e-05,
+      "loss": 0.9131,
+      "step": 7636
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5387738834691747e-05,
+      "loss": 0.9307,
+      "step": 7637
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538652633855098e-05,
+      "loss": 0.915,
+      "step": 7638
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5385313730841893e-05,
+      "loss": 0.96,
+      "step": 7639
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5384101011589616e-05,
+      "loss": 0.9214,
+      "step": 7640
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538288818081927e-05,
+      "loss": 0.9878,
+      "step": 7641
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5381675238555964e-05,
+      "loss": 0.9956,
+      "step": 7642
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5380462184824828e-05,
+      "loss": 0.8765,
+      "step": 7643
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5379249019650986e-05,
+      "loss": 0.9331,
+      "step": 7644
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5378035743059568e-05,
+      "loss": 0.9824,
+      "step": 7645
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53768223550757e-05,
+      "loss": 0.8965,
+      "step": 7646
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5375608855724523e-05,
+      "loss": 1.0015,
+      "step": 7647
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537439524503116e-05,
+      "loss": 0.9136,
+      "step": 7648
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5373181523020756e-05,
+      "loss": 0.9966,
+      "step": 7649
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5371967689718442e-05,
+      "loss": 0.9443,
+      "step": 7650
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5370753745149368e-05,
+      "loss": 0.9678,
+      "step": 7651
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5369539689338675e-05,
+      "loss": 0.9324,
+      "step": 7652
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536832552231151e-05,
+      "loss": 0.9995,
+      "step": 7653
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536711124409302e-05,
+      "loss": 0.9556,
+      "step": 7654
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5365896854708354e-05,
+      "loss": 0.9707,
+      "step": 7655
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5364682354182666e-05,
+      "loss": 0.9839,
+      "step": 7656
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5363467742541112e-05,
+      "loss": 0.9204,
+      "step": 7657
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536225301980885e-05,
+      "loss": 0.8408,
+      "step": 7658
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536103818601104e-05,
+      "loss": 1.0293,
+      "step": 7659
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5359823241172843e-05,
+      "loss": 0.8965,
+      "step": 7660
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5358608185319422e-05,
+      "loss": 1.0264,
+      "step": 7661
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535739301847595e-05,
+      "loss": 0.8257,
+      "step": 7662
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5356177740667596e-05,
+      "loss": 1.0156,
+      "step": 7663
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535496235191952e-05,
+      "loss": 0.9587,
+      "step": 7664
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5353746852256905e-05,
+      "loss": 1.063,
+      "step": 7665
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5352531241704928e-05,
+      "loss": 0.9512,
+      "step": 7666
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5351315520288762e-05,
+      "loss": 0.957,
+      "step": 7667
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5350099688033595e-05,
+      "loss": 0.9399,
+      "step": 7668
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53488837449646e-05,
+      "loss": 0.9507,
+      "step": 7669
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534766769110697e-05,
+      "loss": 0.9492,
+      "step": 7670
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534645152648589e-05,
+      "loss": 0.9937,
+      "step": 7671
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534523525112655e-05,
+      "loss": 0.9219,
+      "step": 7672
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534401886505414e-05,
+      "loss": 0.9238,
+      "step": 7673
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534280236829386e-05,
+      "loss": 0.9648,
+      "step": 7674
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5341585760870897e-05,
+      "loss": 0.8862,
+      "step": 7675
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.534036904281046e-05,
+      "loss": 0.8901,
+      "step": 7676
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5339152214137745e-05,
+      "loss": 0.9399,
+      "step": 7677
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5337935274877956e-05,
+      "loss": 0.938,
+      "step": 7678
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5336718225056296e-05,
+      "loss": 0.9033,
+      "step": 7679
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533550106469798e-05,
+      "loss": 0.8442,
+      "step": 7680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5334283793828216e-05,
+      "loss": 0.8755,
+      "step": 7681
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5333066412472212e-05,
+      "loss": 1.0225,
+      "step": 7682
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5331848920655193e-05,
+      "loss": 0.9648,
+      "step": 7683
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5330631318402365e-05,
+      "loss": 0.9316,
+      "step": 7684
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532941360573895e-05,
+      "loss": 0.999,
+      "step": 7685
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5328195782690176e-05,
+      "loss": 0.9204,
+      "step": 7686
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5326977849281265e-05,
+      "loss": 1.0259,
+      "step": 7687
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5325759805537436e-05,
+      "loss": 0.9009,
+      "step": 7688
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5324541651483927e-05,
+      "loss": 1.0063,
+      "step": 7689
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532332338714597e-05,
+      "loss": 0.9619,
+      "step": 7690
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5322105012548787e-05,
+      "loss": 0.9048,
+      "step": 7691
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5320886527717622e-05,
+      "loss": 1.0088,
+      "step": 7692
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5319667932677714e-05,
+      "loss": 0.937,
+      "step": 7693
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5318449227454296e-05,
+      "loss": 1.0693,
+      "step": 7694
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5317230412072616e-05,
+      "loss": 0.9429,
+      "step": 7695
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5316011486557918e-05,
+      "loss": 0.9302,
+      "step": 7696
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5314792450935448e-05,
+      "loss": 0.9604,
+      "step": 7697
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5313573305230455e-05,
+      "loss": 0.894,
+      "step": 7698
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5312354049468195e-05,
+      "loss": 0.9507,
+      "step": 7699
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.531113468367391e-05,
+      "loss": 0.9932,
+      "step": 7700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530991520787287e-05,
+      "loss": 1.0005,
+      "step": 7701
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530869562209033e-05,
+      "loss": 0.9795,
+      "step": 7702
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5307475926351542e-05,
+      "loss": 0.9316,
+      "step": 7703
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5306256120681777e-05,
+      "loss": 0.958,
+      "step": 7704
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.53050362051063e-05,
+      "loss": 0.8232,
+      "step": 7705
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5303816179650373e-05,
+      "loss": 1.002,
+      "step": 7706
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5302596044339272e-05,
+      "loss": 1.0552,
+      "step": 7707
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5301375799198266e-05,
+      "loss": 0.9478,
+      "step": 7708
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530015544425263e-05,
+      "loss": 0.9917,
+      "step": 7709
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5298934979527636e-05,
+      "loss": 0.8726,
+      "step": 7710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5297714405048573e-05,
+      "loss": 0.8809,
+      "step": 7711
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5296493720840714e-05,
+      "loss": 0.8989,
+      "step": 7712
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5295272926929346e-05,
+      "loss": 0.9961,
+      "step": 7713
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529405202333975e-05,
+      "loss": 1.0337,
+      "step": 7714
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529283101009722e-05,
+      "loss": 1.0059,
+      "step": 7715
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5291609887227045e-05,
+      "loss": 0.9951,
+      "step": 7716
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5290388654754512e-05,
+      "loss": 0.9609,
+      "step": 7717
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.528916731270492e-05,
+      "loss": 0.8105,
+      "step": 7718
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5287945861103567e-05,
+      "loss": 0.8652,
+      "step": 7719
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5286724299975752e-05,
+      "loss": 0.9951,
+      "step": 7720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5285502629346774e-05,
+      "loss": 0.9297,
+      "step": 7721
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.528428084924194e-05,
+      "loss": 0.98,
+      "step": 7722
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5283058959686548e-05,
+      "loss": 1.0513,
+      "step": 7723
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5281836960705923e-05,
+      "loss": 0.9761,
+      "step": 7724
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5280614852325357e-05,
+      "loss": 0.8545,
+      "step": 7725
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5279392634570175e-05,
+      "loss": 0.9658,
+      "step": 7726
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5278170307465692e-05,
+      "loss": 0.9707,
+      "step": 7727
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.527694787103722e-05,
+      "loss": 0.9136,
+      "step": 7728
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5275725325310076e-05,
+      "loss": 1.0151,
+      "step": 7729
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5274502670309592e-05,
+      "loss": 0.9541,
+      "step": 7730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5273279906061082e-05,
+      "loss": 0.9512,
+      "step": 7731
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.527205703258988e-05,
+      "loss": 0.8892,
+      "step": 7732
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5270834049921312e-05,
+      "loss": 0.9346,
+      "step": 7733
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5269610958080706e-05,
+      "loss": 1.0645,
+      "step": 7734
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5268387757093402e-05,
+      "loss": 1.1216,
+      "step": 7735
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526716444698473e-05,
+      "loss": 1.0322,
+      "step": 7736
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526594102778003e-05,
+      "loss": 1.0229,
+      "step": 7737
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5264717499504643e-05,
+      "loss": 0.9697,
+      "step": 7738
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5263493862183908e-05,
+      "loss": 0.9058,
+      "step": 7739
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5262270115843173e-05,
+      "loss": 0.9888,
+      "step": 7740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5261046260507782e-05,
+      "loss": 0.8975,
+      "step": 7741
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5259822296203088e-05,
+      "loss": 0.8167,
+      "step": 7742
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5258598222954436e-05,
+      "loss": 0.9487,
+      "step": 7743
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5257374040787185e-05,
+      "loss": 0.9561,
+      "step": 7744
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525614974972669e-05,
+      "loss": 0.9473,
+      "step": 7745
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5254925349798305e-05,
+      "loss": 0.9766,
+      "step": 7746
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5253700841027397e-05,
+      "loss": 0.9355,
+      "step": 7747
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5252476223439324e-05,
+      "loss": 0.9902,
+      "step": 7748
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5251251497059452e-05,
+      "loss": 0.999,
+      "step": 7749
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5250026661913149e-05,
+      "loss": 0.9517,
+      "step": 7750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5248801718025784e-05,
+      "loss": 0.9429,
+      "step": 7751
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5247576665422728e-05,
+      "loss": 1.0269,
+      "step": 7752
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5246351504129351e-05,
+      "loss": 0.9395,
+      "step": 7753
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5245126234171039e-05,
+      "loss": 1.043,
+      "step": 7754
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.524390085557316e-05,
+      "loss": 0.918,
+      "step": 7755
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5242675368361102e-05,
+      "loss": 0.8838,
+      "step": 7756
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5241449772560245e-05,
+      "loss": 0.9736,
+      "step": 7757
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5240224068195972e-05,
+      "loss": 0.9512,
+      "step": 7758
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5238998255293676e-05,
+      "loss": 0.9634,
+      "step": 7759
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523777233387874e-05,
+      "loss": 0.9888,
+      "step": 7760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523654630397656e-05,
+      "loss": 0.918,
+      "step": 7761
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523532016561253e-05,
+      "loss": 0.8857,
+      "step": 7762
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5234093918812045e-05,
+      "loss": 0.9272,
+      "step": 7763
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5232867563600504e-05,
+      "loss": 1.0352,
+      "step": 7764
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5231641100003309e-05,
+      "loss": 0.9771,
+      "step": 7765
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5230414528045861e-05,
+      "loss": 1.0659,
+      "step": 7766
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5229187847753567e-05,
+      "loss": 1.0537,
+      "step": 7767
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5227961059151835e-05,
+      "loss": 0.8833,
+      "step": 7768
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5226734162266072e-05,
+      "loss": 0.9038,
+      "step": 7769
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5225507157121692e-05,
+      "loss": 1.021,
+      "step": 7770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5224280043744111e-05,
+      "loss": 0.9438,
+      "step": 7771
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5223052822158744e-05,
+      "loss": 0.9092,
+      "step": 7772
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5221825492391008e-05,
+      "loss": 0.8945,
+      "step": 7773
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5220598054466327e-05,
+      "loss": 0.9204,
+      "step": 7774
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.521937050841012e-05,
+      "loss": 0.8826,
+      "step": 7775
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5218142854247819e-05,
+      "loss": 1.0835,
+      "step": 7776
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5216915092004847e-05,
+      "loss": 0.8511,
+      "step": 7777
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5215687221706637e-05,
+      "loss": 1.0593,
+      "step": 7778
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5214459243378615e-05,
+      "loss": 0.9971,
+      "step": 7779
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5213231157046222e-05,
+      "loss": 0.8867,
+      "step": 7780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5212002962734893e-05,
+      "loss": 0.9922,
+      "step": 7781
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5210774660470069e-05,
+      "loss": 0.9141,
+      "step": 7782
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5209546250277187e-05,
+      "loss": 0.9614,
+      "step": 7783
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.520831773218169e-05,
+      "loss": 0.9897,
+      "step": 7784
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5207089106209028e-05,
+      "loss": 0.9082,
+      "step": 7785
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5205860372384647e-05,
+      "loss": 0.8594,
+      "step": 7786
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5204631530733997e-05,
+      "loss": 1.0107,
+      "step": 7787
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.520340258128253e-05,
+      "loss": 1.0488,
+      "step": 7788
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.52021735240557e-05,
+      "loss": 0.9512,
+      "step": 7789
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5200944359078965e-05,
+      "loss": 0.9585,
+      "step": 7790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5199715086377783e-05,
+      "loss": 0.9785,
+      "step": 7791
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5198485705977617e-05,
+      "loss": 0.958,
+      "step": 7792
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5197256217903929e-05,
+      "loss": 0.9395,
+      "step": 7793
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5196026622182186e-05,
+      "loss": 1.0054,
+      "step": 7794
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5194796918837856e-05,
+      "loss": 0.8848,
+      "step": 7795
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5193567107896405e-05,
+      "loss": 0.9668,
+      "step": 7796
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.519233718938331e-05,
+      "loss": 1.104,
+      "step": 7797
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5191107163324046e-05,
+      "loss": 0.9678,
+      "step": 7798
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5189877029744088e-05,
+      "loss": 0.9736,
+      "step": 7799
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5188646788668914e-05,
+      "loss": 0.9077,
+      "step": 7800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5187416440124008e-05,
+      "loss": 0.9629,
+      "step": 7801
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5186185984134856e-05,
+      "loss": 0.8745,
+      "step": 7802
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5184955420726936e-05,
+      "loss": 0.9106,
+      "step": 7803
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5183724749925742e-05,
+      "loss": 0.854,
+      "step": 7804
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.518249397175676e-05,
+      "loss": 0.9263,
+      "step": 7805
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5181263086245488e-05,
+      "loss": 0.8877,
+      "step": 7806
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5180032093417417e-05,
+      "loss": 0.9302,
+      "step": 7807
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5178800993298044e-05,
+      "loss": 0.9209,
+      "step": 7808
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5177569785912866e-05,
+      "loss": 0.9155,
+      "step": 7809
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.517633847128739e-05,
+      "loss": 0.8779,
+      "step": 7810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5175107049447113e-05,
+      "loss": 0.8767,
+      "step": 7811
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.517387552041755e-05,
+      "loss": 0.9829,
+      "step": 7812
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5172643884224201e-05,
+      "loss": 0.9368,
+      "step": 7813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5171412140892577e-05,
+      "loss": 1.0088,
+      "step": 7814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5170180290448193e-05,
+      "loss": 0.8708,
+      "step": 7815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5168948332916563e-05,
+      "loss": 0.8501,
+      "step": 7816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5167716268323206e-05,
+      "loss": 0.9858,
+      "step": 7817
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5166484096693638e-05,
+      "loss": 0.9688,
+      "step": 7818
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5165251818053382e-05,
+      "loss": 0.9365,
+      "step": 7819
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5164019432427956e-05,
+      "loss": 1.0254,
+      "step": 7820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5162786939842894e-05,
+      "loss": 0.9189,
+      "step": 7821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5161554340323718e-05,
+      "loss": 0.7778,
+      "step": 7822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5160321633895963e-05,
+      "loss": 0.8403,
+      "step": 7823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5159088820585157e-05,
+      "loss": 0.9463,
+      "step": 7824
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5157855900416837e-05,
+      "loss": 0.9399,
+      "step": 7825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.515662287341654e-05,
+      "loss": 1.0,
+      "step": 7826
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5155389739609807e-05,
+      "loss": 1.0405,
+      "step": 7827
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5154156499022174e-05,
+      "loss": 1.0,
+      "step": 7828
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5152923151679186e-05,
+      "loss": 0.9395,
+      "step": 7829
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.515168969760639e-05,
+      "loss": 0.9297,
+      "step": 7830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5150456136829341e-05,
+      "loss": 0.9326,
+      "step": 7831
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5149222469373574e-05,
+      "loss": 1.0532,
+      "step": 7832
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5147988695264652e-05,
+      "loss": 0.77,
+      "step": 7833
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5146754814528129e-05,
+      "loss": 0.9956,
+      "step": 7834
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5145520827189553e-05,
+      "loss": 0.9292,
+      "step": 7835
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5144286733274491e-05,
+      "loss": 1.1465,
+      "step": 7836
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5143052532808506e-05,
+      "loss": 1.0156,
+      "step": 7837
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5141818225817155e-05,
+      "loss": 0.9565,
+      "step": 7838
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.514058381232601e-05,
+      "loss": 0.9229,
+      "step": 7839
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5139349292360632e-05,
+      "loss": 1.0708,
+      "step": 7840
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5138114665946595e-05,
+      "loss": 0.8755,
+      "step": 7841
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5136879933109472e-05,
+      "loss": 1.0259,
+      "step": 7842
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5135645093874836e-05,
+      "loss": 1.1025,
+      "step": 7843
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.513441014826826e-05,
+      "loss": 0.9521,
+      "step": 7844
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.513317509631533e-05,
+      "loss": 1.043,
+      "step": 7845
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.513193993804162e-05,
+      "loss": 1.1094,
+      "step": 7846
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5130704673472717e-05,
+      "loss": 0.9517,
+      "step": 7847
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5129469302634206e-05,
+      "loss": 0.9214,
+      "step": 7848
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5128233825551672e-05,
+      "loss": 0.9966,
+      "step": 7849
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.512699824225071e-05,
+      "loss": 0.9209,
+      "step": 7850
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5125762552756906e-05,
+      "loss": 1.0474,
+      "step": 7851
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5124526757095858e-05,
+      "loss": 0.958,
+      "step": 7852
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5123290855293164e-05,
+      "loss": 0.9036,
+      "step": 7853
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5122054847374421e-05,
+      "loss": 1.0288,
+      "step": 7854
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5120818733365226e-05,
+      "loss": 0.9492,
+      "step": 7855
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5119582513291186e-05,
+      "loss": 0.9595,
+      "step": 7856
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5118346187177903e-05,
+      "loss": 1.0039,
+      "step": 7857
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.511710975505099e-05,
+      "loss": 0.8225,
+      "step": 7858
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5115873216936053e-05,
+      "loss": 0.8579,
+      "step": 7859
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5114636572858704e-05,
+      "loss": 0.9473,
+      "step": 7860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5113399822844552e-05,
+      "loss": 0.9839,
+      "step": 7861
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5112162966919224e-05,
+      "loss": 0.9971,
+      "step": 7862
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5110926005108328e-05,
+      "loss": 1.0039,
+      "step": 7863
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510968893743749e-05,
+      "loss": 0.9702,
+      "step": 7864
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5108451763932332e-05,
+      "loss": 1.0337,
+      "step": 7865
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510721448461848e-05,
+      "loss": 0.937,
+      "step": 7866
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510597709952156e-05,
+      "loss": 0.939,
+      "step": 7867
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5104739608667198e-05,
+      "loss": 0.9658,
+      "step": 7868
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510350201208103e-05,
+      "loss": 0.8828,
+      "step": 7869
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510226430978869e-05,
+      "loss": 1.0732,
+      "step": 7870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5101026501815809e-05,
+      "loss": 1.0083,
+      "step": 7871
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509978858818803e-05,
+      "loss": 1.0122,
+      "step": 7872
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509855056893099e-05,
+      "loss": 1.0171,
+      "step": 7873
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5097312444070333e-05,
+      "loss": 1.0273,
+      "step": 7874
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5096074213631702e-05,
+      "loss": 0.9351,
+      "step": 7875
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509483587764075e-05,
+      "loss": 1.022,
+      "step": 7876
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5093597436123118e-05,
+      "loss": 0.9688,
+      "step": 7877
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5092358889104462e-05,
+      "loss": 0.8955,
+      "step": 7878
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5091120236610432e-05,
+      "loss": 0.9189,
+      "step": 7879
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5089881478666688e-05,
+      "loss": 0.8564,
+      "step": 7880
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5088642615298883e-05,
+      "loss": 0.8584,
+      "step": 7881
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5087403646532678e-05,
+      "loss": 0.8867,
+      "step": 7882
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.508616457239374e-05,
+      "loss": 0.9375,
+      "step": 7883
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5084925392907725e-05,
+      "loss": 0.9624,
+      "step": 7884
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5083686108100309e-05,
+      "loss": 0.8779,
+      "step": 7885
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5082446717997151e-05,
+      "loss": 1.0317,
+      "step": 7886
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5081207222623931e-05,
+      "loss": 1.0254,
+      "step": 7887
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5079967622006311e-05,
+      "loss": 0.9609,
+      "step": 7888
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5078727916169978e-05,
+      "loss": 0.9336,
+      "step": 7889
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5077488105140604e-05,
+      "loss": 0.9082,
+      "step": 7890
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.507624818894387e-05,
+      "loss": 0.9336,
+      "step": 7891
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5075008167605454e-05,
+      "loss": 0.9307,
+      "step": 7892
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.507376804115104e-05,
+      "loss": 0.9395,
+      "step": 7893
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.507252780960632e-05,
+      "loss": 0.9111,
+      "step": 7894
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5071287472996976e-05,
+      "loss": 0.9028,
+      "step": 7895
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5070047031348703e-05,
+      "loss": 0.9995,
+      "step": 7896
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5068806484687191e-05,
+      "loss": 0.8838,
+      "step": 7897
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5067565833038136e-05,
+      "loss": 0.8882,
+      "step": 7898
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5066325076427232e-05,
+      "loss": 0.9211,
+      "step": 7899
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5065084214880184e-05,
+      "loss": 0.9795,
+      "step": 7900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5063843248422689e-05,
+      "loss": 1.0977,
+      "step": 7901
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.506260217708045e-05,
+      "loss": 0.9238,
+      "step": 7902
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5061361000879176e-05,
+      "loss": 0.8452,
+      "step": 7903
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.506011971984457e-05,
+      "loss": 1.0107,
+      "step": 7904
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5058878334002349e-05,
+      "loss": 0.9751,
+      "step": 7905
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.505763684337822e-05,
+      "loss": 0.9248,
+      "step": 7906
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.50563952479979e-05,
+      "loss": 0.8945,
+      "step": 7907
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.50551535478871e-05,
+      "loss": 0.9634,
+      "step": 7908
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5053911743071545e-05,
+      "loss": 0.9141,
+      "step": 7909
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5052669833576954e-05,
+      "loss": 0.9507,
+      "step": 7910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.505142781942905e-05,
+      "loss": 0.9531,
+      "step": 7911
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5050185700653553e-05,
+      "loss": 0.9844,
+      "step": 7912
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5048943477276198e-05,
+      "loss": 0.9883,
+      "step": 7913
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5047701149322713e-05,
+      "loss": 1.02,
+      "step": 7914
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5046458716818829e-05,
+      "loss": 0.9844,
+      "step": 7915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5045216179790277e-05,
+      "loss": 1.0112,
+      "step": 7916
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5043973538262795e-05,
+      "loss": 0.8848,
+      "step": 7917
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5042730792262123e-05,
+      "loss": 0.8845,
+      "step": 7918
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5041487941813997e-05,
+      "loss": 0.8394,
+      "step": 7919
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5040244986944166e-05,
+      "loss": 1.0059,
+      "step": 7920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5039001927678372e-05,
+      "loss": 0.863,
+      "step": 7921
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5037758764042357e-05,
+      "loss": 1.0161,
+      "step": 7922
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5036515496061874e-05,
+      "loss": 0.9575,
+      "step": 7923
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5035272123762674e-05,
+      "loss": 0.7959,
+      "step": 7924
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503402864717051e-05,
+      "loss": 0.9253,
+      "step": 7925
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503278506631114e-05,
+      "loss": 1.0303,
+      "step": 7926
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5031541381210318e-05,
+      "loss": 1.0293,
+      "step": 7927
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5030297591893805e-05,
+      "loss": 0.9761,
+      "step": 7928
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5029053698387364e-05,
+      "loss": 0.981,
+      "step": 7929
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5027809700716755e-05,
+      "loss": 0.9194,
+      "step": 7930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5026565598907753e-05,
+      "loss": 0.7739,
+      "step": 7931
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5025321392986117e-05,
+      "loss": 0.916,
+      "step": 7932
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5024077082977622e-05,
+      "loss": 0.9023,
+      "step": 7933
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5022832668908041e-05,
+      "loss": 1.0,
+      "step": 7934
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.502158815080315e-05,
+      "loss": 0.9512,
+      "step": 7935
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5020343528688721e-05,
+      "loss": 0.9731,
+      "step": 7936
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5019098802590537e-05,
+      "loss": 0.8872,
+      "step": 7937
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.501785397253438e-05,
+      "loss": 0.9355,
+      "step": 7938
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.501660903854603e-05,
+      "loss": 0.9858,
+      "step": 7939
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5015364000651279e-05,
+      "loss": 0.9463,
+      "step": 7940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5014118858875908e-05,
+      "loss": 1.0664,
+      "step": 7941
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.501287361324571e-05,
+      "loss": 0.814,
+      "step": 7942
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5011628263786477e-05,
+      "loss": 0.9917,
+      "step": 7943
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5010382810524005e-05,
+      "loss": 0.874,
+      "step": 7944
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5009137253484087e-05,
+      "loss": 0.9204,
+      "step": 7945
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5007891592692522e-05,
+      "loss": 1.0396,
+      "step": 7946
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5006645828175113e-05,
+      "loss": 1.0562,
+      "step": 7947
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5005399959957663e-05,
+      "loss": 0.9014,
+      "step": 7948
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5004153988065975e-05,
+      "loss": 1.0825,
+      "step": 7949
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5002907912525857e-05,
+      "loss": 0.96,
+      "step": 7950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5001661733363117e-05,
+      "loss": 0.9385,
+      "step": 7951
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.500041545060357e-05,
+      "loss": 0.8951,
+      "step": 7952
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4999169064273025e-05,
+      "loss": 1.0439,
+      "step": 7953
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4997922574397302e-05,
+      "loss": 0.9912,
+      "step": 7954
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4996675981002216e-05,
+      "loss": 0.9653,
+      "step": 7955
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4995429284113587e-05,
+      "loss": 0.8877,
+      "step": 7956
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.499418248375724e-05,
+      "loss": 1.0073,
+      "step": 7957
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4992935579958996e-05,
+      "loss": 0.8779,
+      "step": 7958
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4991688572744685e-05,
+      "loss": 1.0229,
+      "step": 7959
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4990441462140133e-05,
+      "loss": 0.9072,
+      "step": 7960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.498919424817117e-05,
+      "loss": 0.957,
+      "step": 7961
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.498794693086363e-05,
+      "loss": 0.9829,
+      "step": 7962
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.498669951024335e-05,
+      "loss": 1.0381,
+      "step": 7963
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4985451986336164e-05,
+      "loss": 0.9746,
+      "step": 7964
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4984204359167915e-05,
+      "loss": 0.9824,
+      "step": 7965
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.498295662876444e-05,
+      "loss": 0.9546,
+      "step": 7966
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4981708795151586e-05,
+      "loss": 0.9199,
+      "step": 7967
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4980460858355197e-05,
+      "loss": 1.0273,
+      "step": 7968
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4979212818401121e-05,
+      "loss": 0.9463,
+      "step": 7969
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.497796467531521e-05,
+      "loss": 0.8618,
+      "step": 7970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.497671642912331e-05,
+      "loss": 0.9707,
+      "step": 7971
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4975468079851285e-05,
+      "loss": 1.0303,
+      "step": 7972
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4974219627524984e-05,
+      "loss": 0.8979,
+      "step": 7973
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4972971072170268e-05,
+      "loss": 0.9629,
+      "step": 7974
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4971722413812998e-05,
+      "loss": 0.7922,
+      "step": 7975
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4970473652479033e-05,
+      "loss": 1.0508,
+      "step": 7976
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4969224788194244e-05,
+      "loss": 0.874,
+      "step": 7977
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4967975820984495e-05,
+      "loss": 0.9448,
+      "step": 7978
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4966726750875656e-05,
+      "loss": 0.8955,
+      "step": 7979
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.49654775778936e-05,
+      "loss": 1.0005,
+      "step": 7980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4964228302064195e-05,
+      "loss": 0.9048,
+      "step": 7981
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4962978923413319e-05,
+      "loss": 0.9038,
+      "step": 7982
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4961729441966852e-05,
+      "loss": 0.9336,
+      "step": 7983
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4960479857750673e-05,
+      "loss": 0.9673,
+      "step": 7984
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4959230170790661e-05,
+      "loss": 0.9858,
+      "step": 7985
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4957980381112707e-05,
+      "loss": 1.0186,
+      "step": 7986
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4956730488742689e-05,
+      "loss": 0.8735,
+      "step": 7987
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.49554804937065e-05,
+      "loss": 0.939,
+      "step": 7988
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.495423039603003e-05,
+      "loss": 0.9219,
+      "step": 7989
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.495298019573917e-05,
+      "loss": 0.8608,
+      "step": 7990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4951729892859817e-05,
+      "loss": 0.9897,
+      "step": 7991
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4950479487417867e-05,
+      "loss": 0.9224,
+      "step": 7992
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4949228979439219e-05,
+      "loss": 0.9517,
+      "step": 7993
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4947978368949774e-05,
+      "loss": 1.0122,
+      "step": 7994
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4946727655975438e-05,
+      "loss": 0.9504,
+      "step": 7995
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494547684054211e-05,
+      "loss": 0.9492,
+      "step": 7996
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4944225922675702e-05,
+      "loss": 0.9297,
+      "step": 7997
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494297490240212e-05,
+      "loss": 1.062,
+      "step": 7998
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4941723779747283e-05,
+      "loss": 0.9072,
+      "step": 7999
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4940472554737097e-05,
+      "loss": 1.0059,
+      "step": 8000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4939221227397482e-05,
+      "loss": 0.9819,
+      "step": 8001
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4937969797754354e-05,
+      "loss": 0.8711,
+      "step": 8002
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4936718265833636e-05,
+      "loss": 0.981,
+      "step": 8003
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.493546663166125e-05,
+      "loss": 0.979,
+      "step": 8004
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.493421489526312e-05,
+      "loss": 0.9375,
+      "step": 8005
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.493296305666517e-05,
+      "loss": 0.8667,
+      "step": 8006
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.493171111589333e-05,
+      "loss": 0.9668,
+      "step": 8007
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4930459072973533e-05,
+      "loss": 0.9832,
+      "step": 8008
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.492920692793171e-05,
+      "loss": 0.9849,
+      "step": 8009
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.49279546807938e-05,
+      "loss": 0.8828,
+      "step": 8010
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4926702331585729e-05,
+      "loss": 0.981,
+      "step": 8011
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4925449880333446e-05,
+      "loss": 1.0908,
+      "step": 8012
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4924197327062893e-05,
+      "loss": 0.907,
+      "step": 8013
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4922944671800009e-05,
+      "loss": 0.9595,
+      "step": 8014
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.492169191457074e-05,
+      "loss": 0.8623,
+      "step": 8015
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4920439055401036e-05,
+      "loss": 1.0239,
+      "step": 8016
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4919186094316844e-05,
+      "loss": 1.1123,
+      "step": 8017
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4917933031344119e-05,
+      "loss": 1.0117,
+      "step": 8018
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4916679866508815e-05,
+      "loss": 0.7793,
+      "step": 8019
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4915426599836884e-05,
+      "loss": 1.0088,
+      "step": 8020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4914173231354288e-05,
+      "loss": 0.9468,
+      "step": 8021
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4912919761086989e-05,
+      "loss": 1.0576,
+      "step": 8022
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4911666189060942e-05,
+      "loss": 0.9897,
+      "step": 8023
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4910412515302119e-05,
+      "loss": 1.0,
+      "step": 8024
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4909158739836487e-05,
+      "loss": 0.9121,
+      "step": 8025
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4907904862690008e-05,
+      "loss": 1.0337,
+      "step": 8026
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4906650883888658e-05,
+      "loss": 0.959,
+      "step": 8027
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4905396803458407e-05,
+      "loss": 0.9375,
+      "step": 8028
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4904142621425236e-05,
+      "loss": 0.9473,
+      "step": 8029
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4902888337815116e-05,
+      "loss": 0.8809,
+      "step": 8030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4901633952654029e-05,
+      "loss": 0.9609,
+      "step": 8031
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4900379465967954e-05,
+      "loss": 0.9028,
+      "step": 8032
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4899124877782881e-05,
+      "loss": 0.8984,
+      "step": 8033
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4897870188124786e-05,
+      "loss": 0.9512,
+      "step": 8034
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4896615397019662e-05,
+      "loss": 1.0806,
+      "step": 8035
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4895360504493503e-05,
+      "loss": 1.0225,
+      "step": 8036
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4894105510572295e-05,
+      "loss": 1.0361,
+      "step": 8037
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4892850415282032e-05,
+      "loss": 0.9624,
+      "step": 8038
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4891595218648714e-05,
+      "loss": 0.9741,
+      "step": 8039
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4890339920698334e-05,
+      "loss": 0.8892,
+      "step": 8040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4889084521456898e-05,
+      "loss": 0.8926,
+      "step": 8041
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4887829020950404e-05,
+      "loss": 0.9272,
+      "step": 8042
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.488657341920486e-05,
+      "loss": 0.8921,
+      "step": 8043
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.488531771624627e-05,
+      "loss": 0.9399,
+      "step": 8044
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4884061912100643e-05,
+      "loss": 0.9426,
+      "step": 8045
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4882806006793991e-05,
+      "loss": 1.0059,
+      "step": 8046
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4881550000352327e-05,
+      "loss": 0.916,
+      "step": 8047
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4880293892801664e-05,
+      "loss": 0.9634,
+      "step": 8048
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4879037684168021e-05,
+      "loss": 1.0513,
+      "step": 8049
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4877781374477418e-05,
+      "loss": 0.8503,
+      "step": 8050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4876524963755873e-05,
+      "loss": 1.1816,
+      "step": 8051
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4875268452029412e-05,
+      "loss": 0.9473,
+      "step": 8052
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.487401183932406e-05,
+      "loss": 0.939,
+      "step": 8053
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4872755125665845e-05,
+      "loss": 0.9541,
+      "step": 8054
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4871498311080794e-05,
+      "loss": 0.8516,
+      "step": 8055
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4870241395594941e-05,
+      "loss": 0.998,
+      "step": 8056
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4868984379234324e-05,
+      "loss": 1.0288,
+      "step": 8057
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.486772726202497e-05,
+      "loss": 0.9419,
+      "step": 8058
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4866470043992921e-05,
+      "loss": 0.9502,
+      "step": 8059
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4865212725164223e-05,
+      "loss": 0.9158,
+      "step": 8060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4863955305564907e-05,
+      "loss": 0.918,
+      "step": 8061
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4862697785221026e-05,
+      "loss": 0.8967,
+      "step": 8062
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4861440164158623e-05,
+      "loss": 0.9614,
+      "step": 8063
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4860182442403748e-05,
+      "loss": 0.9556,
+      "step": 8064
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4858924619982449e-05,
+      "loss": 0.9453,
+      "step": 8065
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4857666696920778e-05,
+      "loss": 0.9165,
+      "step": 8066
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4856408673244793e-05,
+      "loss": 0.8496,
+      "step": 8067
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.485515054898055e-05,
+      "loss": 0.9927,
+      "step": 8068
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4853892324154112e-05,
+      "loss": 0.9622,
+      "step": 8069
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.485263399879153e-05,
+      "loss": 0.9199,
+      "step": 8070
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4851375572918874e-05,
+      "loss": 0.8262,
+      "step": 8071
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4850117046562208e-05,
+      "loss": 1.0024,
+      "step": 8072
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4848858419747597e-05,
+      "loss": 0.9258,
+      "step": 8073
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4847599692501114e-05,
+      "loss": 0.9937,
+      "step": 8074
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4846340864848827e-05,
+      "loss": 0.9805,
+      "step": 8075
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.484508193681681e-05,
+      "loss": 0.9307,
+      "step": 8076
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4843822908431142e-05,
+      "loss": 1.042,
+      "step": 8077
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4842563779717895e-05,
+      "loss": 0.9023,
+      "step": 8078
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4841304550703152e-05,
+      "loss": 0.8354,
+      "step": 8079
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4840045221412996e-05,
+      "loss": 0.9561,
+      "step": 8080
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.483878579187351e-05,
+      "loss": 0.9404,
+      "step": 8081
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4837526262110775e-05,
+      "loss": 0.9111,
+      "step": 8082
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4836266632150886e-05,
+      "loss": 0.9927,
+      "step": 8083
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4835006902019933e-05,
+      "loss": 0.8833,
+      "step": 8084
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4833747071744003e-05,
+      "loss": 0.9634,
+      "step": 8085
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.483248714134919e-05,
+      "loss": 1.0386,
+      "step": 8086
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4831227110861595e-05,
+      "loss": 1.0103,
+      "step": 8087
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4829966980307312e-05,
+      "loss": 0.9697,
+      "step": 8088
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4828706749712447e-05,
+      "loss": 0.9375,
+      "step": 8089
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4827446419103097e-05,
+      "loss": 1.0391,
+      "step": 8090
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4826185988505369e-05,
+      "loss": 1.019,
+      "step": 8091
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4824925457945367e-05,
+      "loss": 0.9258,
+      "step": 8092
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4823664827449205e-05,
+      "loss": 0.9409,
+      "step": 8093
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4822404097042989e-05,
+      "loss": 0.9297,
+      "step": 8094
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4821143266752836e-05,
+      "loss": 0.9429,
+      "step": 8095
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4819882336604857e-05,
+      "loss": 1.1099,
+      "step": 8096
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.481862130662517e-05,
+      "loss": 1.0024,
+      "step": 8097
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4817360176839892e-05,
+      "loss": 0.9619,
+      "step": 8098
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4816098947275153e-05,
+      "loss": 0.8628,
+      "step": 8099
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4814837617957068e-05,
+      "loss": 0.8591,
+      "step": 8100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4813576188911762e-05,
+      "loss": 0.9487,
+      "step": 8101
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4812314660165364e-05,
+      "loss": 0.9512,
+      "step": 8102
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4811053031744002e-05,
+      "loss": 0.9668,
+      "step": 8103
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4809791303673813e-05,
+      "loss": 1.1299,
+      "step": 8104
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4808529475980927e-05,
+      "loss": 0.8491,
+      "step": 8105
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4807267548691478e-05,
+      "loss": 1.0332,
+      "step": 8106
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4806005521831605e-05,
+      "loss": 1.0181,
+      "step": 8107
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4804743395427448e-05,
+      "loss": 0.9805,
+      "step": 8108
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.480348116950515e-05,
+      "loss": 0.9268,
+      "step": 8109
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4802218844090849e-05,
+      "loss": 1.0586,
+      "step": 8110
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4800956419210701e-05,
+      "loss": 0.9766,
+      "step": 8111
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4799693894890845e-05,
+      "loss": 0.9375,
+      "step": 8112
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4798431271157434e-05,
+      "loss": 0.9844,
+      "step": 8113
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4797168548036623e-05,
+      "loss": 1.0596,
+      "step": 8114
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.479590572555456e-05,
+      "loss": 0.9575,
+      "step": 8115
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4794642803737404e-05,
+      "loss": 0.9438,
+      "step": 8116
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4793379782611315e-05,
+      "loss": 0.9512,
+      "step": 8117
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4792116662202454e-05,
+      "loss": 0.9214,
+      "step": 8118
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4790853442536979e-05,
+      "loss": 0.9482,
+      "step": 8119
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4789590123641058e-05,
+      "loss": 0.9756,
+      "step": 8120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4788326705540857e-05,
+      "loss": 0.9358,
+      "step": 8121
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4787063188262542e-05,
+      "loss": 0.9673,
+      "step": 8122
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4785799571832287e-05,
+      "loss": 1.019,
+      "step": 8123
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4784535856276262e-05,
+      "loss": 0.9434,
+      "step": 8124
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4783272041620642e-05,
+      "loss": 1.0415,
+      "step": 8125
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4782008127891604e-05,
+      "loss": 0.9517,
+      "step": 8126
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4780744115115331e-05,
+      "loss": 0.9976,
+      "step": 8127
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4779480003317996e-05,
+      "loss": 0.9243,
+      "step": 8128
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4778215792525789e-05,
+      "loss": 0.9746,
+      "step": 8129
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.477695148276489e-05,
+      "loss": 0.9922,
+      "step": 8130
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4775687074061489e-05,
+      "loss": 0.8923,
+      "step": 8131
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4774422566441773e-05,
+      "loss": 0.8477,
+      "step": 8132
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4773157959931934e-05,
+      "loss": 0.8726,
+      "step": 8133
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4771893254558165e-05,
+      "loss": 1.0522,
+      "step": 8134
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4770628450346663e-05,
+      "loss": 0.9189,
+      "step": 8135
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.476936354732362e-05,
+      "loss": 0.9673,
+      "step": 8136
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4768098545515242e-05,
+      "loss": 0.9429,
+      "step": 8137
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4766833444947728e-05,
+      "loss": 1.0771,
+      "step": 8138
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4765568245647277e-05,
+      "loss": 0.9771,
+      "step": 8139
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.47643029476401e-05,
+      "loss": 0.7812,
+      "step": 8140
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4763037550952403e-05,
+      "loss": 0.8833,
+      "step": 8141
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4761772055610393e-05,
+      "loss": 0.9849,
+      "step": 8142
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4760506461640283e-05,
+      "loss": 1.0713,
+      "step": 8143
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4759240769068291e-05,
+      "loss": 0.9834,
+      "step": 8144
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4757974977920624e-05,
+      "loss": 1.0073,
+      "step": 8145
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4756709088223508e-05,
+      "loss": 1.0557,
+      "step": 8146
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4755443100003158e-05,
+      "loss": 0.9531,
+      "step": 8147
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4754177013285795e-05,
+      "loss": 0.8843,
+      "step": 8148
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4752910828097646e-05,
+      "loss": 0.8945,
+      "step": 8149
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4751644544464937e-05,
+      "loss": 1.0684,
+      "step": 8150
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.475037816241389e-05,
+      "loss": 1.0327,
+      "step": 8151
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4749111681970743e-05,
+      "loss": 1.0518,
+      "step": 8152
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4747845103161723e-05,
+      "loss": 1.0239,
+      "step": 8153
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4746578426013063e-05,
+      "loss": 0.9189,
+      "step": 8154
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4745311650551005e-05,
+      "loss": 0.8484,
+      "step": 8155
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4744044776801778e-05,
+      "loss": 1.0122,
+      "step": 8156
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.474277780479163e-05,
+      "loss": 1.0352,
+      "step": 8157
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4741510734546804e-05,
+      "loss": 0.9165,
+      "step": 8158
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4740243566093538e-05,
+      "loss": 1.019,
+      "step": 8159
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4738976299458082e-05,
+      "loss": 0.9746,
+      "step": 8160
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.473770893466668e-05,
+      "loss": 0.9087,
+      "step": 8161
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4736441471745587e-05,
+      "loss": 0.9185,
+      "step": 8162
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4735173910721052e-05,
+      "loss": 0.8892,
+      "step": 8163
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4733906251619332e-05,
+      "loss": 0.8931,
+      "step": 8164
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4732638494466682e-05,
+      "loss": 0.8491,
+      "step": 8165
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4731370639289363e-05,
+      "loss": 0.9229,
+      "step": 8166
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.473010268611363e-05,
+      "loss": 0.9312,
+      "step": 8167
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.472883463496575e-05,
+      "loss": 0.9067,
+      "step": 8168
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4727566485871985e-05,
+      "loss": 0.9331,
+      "step": 8169
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4726298238858603e-05,
+      "loss": 0.9937,
+      "step": 8170
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4725029893951872e-05,
+      "loss": 0.9897,
+      "step": 8171
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4723761451178062e-05,
+      "loss": 0.8616,
+      "step": 8172
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.472249291056345e-05,
+      "loss": 0.8384,
+      "step": 8173
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4721224272134304e-05,
+      "loss": 1.0469,
+      "step": 8174
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4719955535916905e-05,
+      "loss": 0.9282,
+      "step": 8175
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4718686701937528e-05,
+      "loss": 1.0781,
+      "step": 8176
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4717417770222457e-05,
+      "loss": 0.9404,
+      "step": 8177
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4716148740797973e-05,
+      "loss": 0.9395,
+      "step": 8178
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4714879613690362e-05,
+      "loss": 0.9004,
+      "step": 8179
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.471361038892591e-05,
+      "loss": 0.9849,
+      "step": 8180
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4712341066530904e-05,
+      "loss": 0.8506,
+      "step": 8181
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4711071646531639e-05,
+      "loss": 0.8059,
+      "step": 8182
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4709802128954405e-05,
+      "loss": 1.0024,
+      "step": 8183
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4708532513825498e-05,
+      "loss": 0.9844,
+      "step": 8184
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4707262801171212e-05,
+      "loss": 0.8423,
+      "step": 8185
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.470599299101785e-05,
+      "loss": 0.9316,
+      "step": 8186
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4704723083391711e-05,
+      "loss": 0.9922,
+      "step": 8187
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4703453078319101e-05,
+      "loss": 0.939,
+      "step": 8188
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4702182975826318e-05,
+      "loss": 0.8906,
+      "step": 8189
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4700912775939676e-05,
+      "loss": 1.0508,
+      "step": 8190
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4699642478685477e-05,
+      "loss": 0.9683,
+      "step": 8191
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4698372084090036e-05,
+      "loss": 1.0127,
+      "step": 8192
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4697101592179668e-05,
+      "loss": 0.9834,
+      "step": 8193
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4695831002980684e-05,
+      "loss": 0.7964,
+      "step": 8194
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4694560316519406e-05,
+      "loss": 0.9258,
+      "step": 8195
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4693289532822147e-05,
+      "loss": 0.9175,
+      "step": 8196
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4692018651915234e-05,
+      "loss": 1.0068,
+      "step": 8197
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4690747673824986e-05,
+      "loss": 0.9619,
+      "step": 8198
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.468947659857773e-05,
+      "loss": 1.0225,
+      "step": 8199
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4688205426199792e-05,
+      "loss": 1.0337,
+      "step": 8200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.46869341567175e-05,
+      "loss": 0.9604,
+      "step": 8201
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4685662790157192e-05,
+      "loss": 1.0884,
+      "step": 8202
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4684391326545192e-05,
+      "loss": 0.9678,
+      "step": 8203
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.468311976590784e-05,
+      "loss": 0.9248,
+      "step": 8204
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.468184810827147e-05,
+      "loss": 0.8696,
+      "step": 8205
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4680576353662426e-05,
+      "loss": 0.8813,
+      "step": 8206
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4679304502107046e-05,
+      "loss": 0.7971,
+      "step": 8207
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4678032553631673e-05,
+      "loss": 0.9282,
+      "step": 8208
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4676760508262652e-05,
+      "loss": 0.8701,
+      "step": 8209
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4675488366026334e-05,
+      "loss": 0.8162,
+      "step": 8210
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4674216126949066e-05,
+      "loss": 0.9224,
+      "step": 8211
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4672943791057197e-05,
+      "loss": 0.9048,
+      "step": 8212
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4671671358377081e-05,
+      "loss": 0.8672,
+      "step": 8213
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4670398828935076e-05,
+      "loss": 0.9531,
+      "step": 8214
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4669126202757535e-05,
+      "loss": 0.9141,
+      "step": 8215
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4667853479870821e-05,
+      "loss": 0.9941,
+      "step": 8216
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4666580660301295e-05,
+      "loss": 1.0308,
+      "step": 8217
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.466530774407532e-05,
+      "loss": 0.9888,
+      "step": 8218
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4664034731219256e-05,
+      "loss": 1.0171,
+      "step": 8219
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4662761621759478e-05,
+      "loss": 0.9497,
+      "step": 8220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4661488415722352e-05,
+      "loss": 0.8838,
+      "step": 8221
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4660215113134247e-05,
+      "loss": 1.0752,
+      "step": 8222
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4658941714021539e-05,
+      "loss": 0.8564,
+      "step": 8223
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4657668218410606e-05,
+      "loss": 1.0474,
+      "step": 8224
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4656394626327816e-05,
+      "loss": 0.9038,
+      "step": 8225
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4655120937799559e-05,
+      "loss": 0.9824,
+      "step": 8226
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.465384715285221e-05,
+      "loss": 0.979,
+      "step": 8227
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4652573271512151e-05,
+      "loss": 0.896,
+      "step": 8228
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4651299293805774e-05,
+      "loss": 0.957,
+      "step": 8229
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4650025219759458e-05,
+      "loss": 0.8916,
+      "step": 8230
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.46487510493996e-05,
+      "loss": 0.9502,
+      "step": 8231
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4647476782752585e-05,
+      "loss": 1.0308,
+      "step": 8232
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.464620241984481e-05,
+      "loss": 1.04,
+      "step": 8233
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4644927960702671e-05,
+      "loss": 0.9336,
+      "step": 8234
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.464365340535256e-05,
+      "loss": 0.9575,
+      "step": 8235
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.464237875382088e-05,
+      "loss": 0.9409,
+      "step": 8236
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4641104006134032e-05,
+      "loss": 0.9512,
+      "step": 8237
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4639829162318417e-05,
+      "loss": 1.0215,
+      "step": 8238
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4638554222400445e-05,
+      "loss": 0.9929,
+      "step": 8239
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.463727918640652e-05,
+      "loss": 0.9863,
+      "step": 8240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4636004054363048e-05,
+      "loss": 0.9751,
+      "step": 8241
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4634728826296445e-05,
+      "loss": 1.0312,
+      "step": 8242
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4633453502233124e-05,
+      "loss": 1.0112,
+      "step": 8243
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4632178082199498e-05,
+      "loss": 0.9395,
+      "step": 8244
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4630902566221982e-05,
+      "loss": 1.0195,
+      "step": 8245
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4629626954327001e-05,
+      "loss": 1.0864,
+      "step": 8246
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.462835124654097e-05,
+      "loss": 1.0713,
+      "step": 8247
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4627075442890322e-05,
+      "loss": 1.0146,
+      "step": 8248
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4625799543401469e-05,
+      "loss": 0.833,
+      "step": 8249
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4624523548100844e-05,
+      "loss": 0.957,
+      "step": 8250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.462324745701488e-05,
+      "loss": 0.9224,
+      "step": 8251
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4621971270169999e-05,
+      "loss": 0.9609,
+      "step": 8252
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4620694987592643e-05,
+      "loss": 0.9995,
+      "step": 8253
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4619418609309241e-05,
+      "loss": 0.9302,
+      "step": 8254
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4618142135346234e-05,
+      "loss": 0.9702,
+      "step": 8255
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4616865565730058e-05,
+      "loss": 0.874,
+      "step": 8256
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4615588900487152e-05,
+      "loss": 0.9277,
+      "step": 8257
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4614312139643967e-05,
+      "loss": 1.001,
+      "step": 8258
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.461303528322694e-05,
+      "loss": 0.978,
+      "step": 8259
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.461175833126252e-05,
+      "loss": 0.9326,
+      "step": 8260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4610481283777159e-05,
+      "loss": 0.9453,
+      "step": 8261
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4609204140797305e-05,
+      "loss": 0.9062,
+      "step": 8262
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.460792690234941e-05,
+      "loss": 0.8896,
+      "step": 8263
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4606649568459927e-05,
+      "loss": 1.0156,
+      "step": 8264
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.460537213915532e-05,
+      "loss": 0.9565,
+      "step": 8265
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.460409461446204e-05,
+      "loss": 1.0107,
+      "step": 8266
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4602816994406555e-05,
+      "loss": 0.981,
+      "step": 8267
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.460153927901532e-05,
+      "loss": 0.9932,
+      "step": 8268
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.46002614683148e-05,
+      "loss": 0.9624,
+      "step": 8269
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4598983562331471e-05,
+      "loss": 0.8364,
+      "step": 8270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4597705561091793e-05,
+      "loss": 1.0635,
+      "step": 8271
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.459642746462224e-05,
+      "loss": 0.9814,
+      "step": 8272
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4595149272949285e-05,
+      "loss": 0.9419,
+      "step": 8273
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4593870986099398e-05,
+      "loss": 1.0137,
+      "step": 8274
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4592592604099061e-05,
+      "loss": 0.9775,
+      "step": 8275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4591314126974749e-05,
+      "loss": 0.9829,
+      "step": 8276
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4590035554752946e-05,
+      "loss": 1.0288,
+      "step": 8277
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4588756887460132e-05,
+      "loss": 0.9775,
+      "step": 8278
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4587478125122787e-05,
+      "loss": 1.0811,
+      "step": 8279
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4586199267767404e-05,
+      "loss": 1.0308,
+      "step": 8280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.458492031542047e-05,
+      "loss": 0.854,
+      "step": 8281
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.458364126810847e-05,
+      "loss": 0.8252,
+      "step": 8282
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4582362125857902e-05,
+      "loss": 0.9614,
+      "step": 8283
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4581082888695261e-05,
+      "loss": 0.9531,
+      "step": 8284
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4579803556647041e-05,
+      "loss": 0.9111,
+      "step": 8285
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.457852412973974e-05,
+      "loss": 0.9409,
+      "step": 8286
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4577244607999857e-05,
+      "loss": 0.9912,
+      "step": 8287
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4575964991453895e-05,
+      "loss": 0.9385,
+      "step": 8288
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4574685280128358e-05,
+      "loss": 1.001,
+      "step": 8289
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4573405474049753e-05,
+      "loss": 0.9966,
+      "step": 8290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4572125573244586e-05,
+      "loss": 0.8098,
+      "step": 8291
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4570845577739372e-05,
+      "loss": 0.9761,
+      "step": 8292
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4569565487560616e-05,
+      "loss": 0.8887,
+      "step": 8293
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4568285302734832e-05,
+      "loss": 0.9648,
+      "step": 8294
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.456700502328854e-05,
+      "loss": 0.8555,
+      "step": 8295
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4565724649248258e-05,
+      "loss": 0.9551,
+      "step": 8296
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4564444180640504e-05,
+      "loss": 0.9341,
+      "step": 8297
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.45631636174918e-05,
+      "loss": 0.9717,
+      "step": 8298
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4561882959828668e-05,
+      "loss": 0.8232,
+      "step": 8299
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4560602207677637e-05,
+      "loss": 1.0005,
+      "step": 8300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4559321361065228e-05,
+      "loss": 1.022,
+      "step": 8301
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4558040420017978e-05,
+      "loss": 0.9231,
+      "step": 8302
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4556759384562418e-05,
+      "loss": 1.0,
+      "step": 8303
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4555478254725075e-05,
+      "loss": 0.9995,
+      "step": 8304
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.455419703053249e-05,
+      "loss": 0.8984,
+      "step": 8305
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4552915712011198e-05,
+      "loss": 0.9438,
+      "step": 8306
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.455163429918774e-05,
+      "loss": 1.0474,
+      "step": 8307
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4550352792088655e-05,
+      "loss": 0.9077,
+      "step": 8308
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4549071190740486e-05,
+      "loss": 0.9041,
+      "step": 8309
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4547789495169782e-05,
+      "loss": 0.9424,
+      "step": 8310
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4546507705403085e-05,
+      "loss": 0.9185,
+      "step": 8311
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4545225821466951e-05,
+      "loss": 1.0361,
+      "step": 8312
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4543943843387921e-05,
+      "loss": 1.0659,
+      "step": 8313
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4542661771192559e-05,
+      "loss": 1.002,
+      "step": 8314
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.454137960490741e-05,
+      "loss": 1.0,
+      "step": 8315
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4540097344559037e-05,
+      "loss": 0.8931,
+      "step": 8316
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4538814990173997e-05,
+      "loss": 0.8572,
+      "step": 8317
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.453753254177885e-05,
+      "loss": 0.9395,
+      "step": 8318
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.453624999940016e-05,
+      "loss": 0.9541,
+      "step": 8319
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4534967363064491e-05,
+      "loss": 0.9795,
+      "step": 8320
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.453368463279841e-05,
+      "loss": 0.8745,
+      "step": 8321
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4532401808628485e-05,
+      "loss": 1.0044,
+      "step": 8322
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4531118890581288e-05,
+      "loss": 0.9155,
+      "step": 8323
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4529835878683387e-05,
+      "loss": 0.958,
+      "step": 8324
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4528552772961362e-05,
+      "loss": 0.9385,
+      "step": 8325
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4527269573441787e-05,
+      "loss": 0.9775,
+      "step": 8326
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4525986280151239e-05,
+      "loss": 0.9619,
+      "step": 8327
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4524702893116299e-05,
+      "loss": 0.9937,
+      "step": 8328
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.452341941236355e-05,
+      "loss": 0.9111,
+      "step": 8329
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4522135837919573e-05,
+      "loss": 0.8608,
+      "step": 8330
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4520852169810957e-05,
+      "loss": 0.9971,
+      "step": 8331
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.451956840806429e-05,
+      "loss": 0.9082,
+      "step": 8332
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4518284552706163e-05,
+      "loss": 1.0229,
+      "step": 8333
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4517000603763162e-05,
+      "loss": 1.0063,
+      "step": 8334
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4515716561261887e-05,
+      "loss": 0.9692,
+      "step": 8335
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4514432425228932e-05,
+      "loss": 0.9297,
+      "step": 8336
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4513148195690896e-05,
+      "loss": 1.0112,
+      "step": 8337
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4511863872674374e-05,
+      "loss": 1.0254,
+      "step": 8338
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4510579456205971e-05,
+      "loss": 0.8501,
+      "step": 8339
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4509294946312289e-05,
+      "loss": 0.8843,
+      "step": 8340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4508010343019934e-05,
+      "loss": 0.9539,
+      "step": 8341
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4506725646355512e-05,
+      "loss": 1.0142,
+      "step": 8342
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4505440856345639e-05,
+      "loss": 1.0552,
+      "step": 8343
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4504155973016916e-05,
+      "loss": 0.9248,
+      "step": 8344
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4502870996395962e-05,
+      "loss": 1.0693,
+      "step": 8345
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4501585926509387e-05,
+      "loss": 0.7783,
+      "step": 8346
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4500300763383816e-05,
+      "loss": 0.9624,
+      "step": 8347
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4499015507045866e-05,
+      "loss": 1.0229,
+      "step": 8348
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4497730157522152e-05,
+      "loss": 1.0259,
+      "step": 8349
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4496444714839301e-05,
+      "loss": 0.9365,
+      "step": 8350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4495159179023938e-05,
+      "loss": 0.8164,
+      "step": 8351
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4493873550102693e-05,
+      "loss": 0.8711,
+      "step": 8352
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4492587828102185e-05,
+      "loss": 0.9702,
+      "step": 8353
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4491302013049051e-05,
+      "loss": 0.8965,
+      "step": 8354
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4490016104969922e-05,
+      "loss": 0.9165,
+      "step": 8355
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4488730103891433e-05,
+      "loss": 1.019,
+      "step": 8356
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.448744400984022e-05,
+      "loss": 1.0649,
+      "step": 8357
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.448615782284292e-05,
+      "loss": 0.9365,
+      "step": 8358
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4484871542926176e-05,
+      "loss": 1.04,
+      "step": 8359
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4483585170116627e-05,
+      "loss": 1.0415,
+      "step": 8360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4482298704440919e-05,
+      "loss": 0.938,
+      "step": 8361
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4481012145925695e-05,
+      "loss": 1.0581,
+      "step": 8362
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4479725494597608e-05,
+      "loss": 1.0132,
+      "step": 8363
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4478438750483301e-05,
+      "loss": 0.8779,
+      "step": 8364
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4477151913609432e-05,
+      "loss": 0.9438,
+      "step": 8365
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4475864984002654e-05,
+      "loss": 0.9712,
+      "step": 8366
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.447457796168962e-05,
+      "loss": 0.939,
+      "step": 8367
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4473290846696984e-05,
+      "loss": 1.0806,
+      "step": 8368
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4472003639051411e-05,
+      "loss": 0.9624,
+      "step": 8369
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.447071633877956e-05,
+      "loss": 1.0532,
+      "step": 8370
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4469428945908093e-05,
+      "loss": 0.8999,
+      "step": 8371
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4468141460463678e-05,
+      "loss": 1.021,
+      "step": 8372
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4466853882472982e-05,
+      "loss": 0.9492,
+      "step": 8373
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4465566211962671e-05,
+      "loss": 0.9902,
+      "step": 8374
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4464278448959418e-05,
+      "loss": 0.8633,
+      "step": 8375
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4462990593489895e-05,
+      "loss": 0.877,
+      "step": 8376
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.446170264558078e-05,
+      "loss": 0.9585,
+      "step": 8377
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4460414605258744e-05,
+      "loss": 1.0308,
+      "step": 8378
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4459126472550469e-05,
+      "loss": 0.8896,
+      "step": 8379
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4457838247482632e-05,
+      "loss": 0.9287,
+      "step": 8380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4456549930081923e-05,
+      "loss": 0.9312,
+      "step": 8381
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4455261520375017e-05,
+      "loss": 1.0273,
+      "step": 8382
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4453973018388601e-05,
+      "loss": 0.9771,
+      "step": 8383
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.445268442414937e-05,
+      "loss": 1.064,
+      "step": 8384
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.445139573768401e-05,
+      "loss": 0.9927,
+      "step": 8385
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4450106959019214e-05,
+      "loss": 0.9702,
+      "step": 8386
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4448818088181673e-05,
+      "loss": 0.937,
+      "step": 8387
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4447529125198086e-05,
+      "loss": 0.978,
+      "step": 8388
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.444624007009515e-05,
+      "loss": 0.9253,
+      "step": 8389
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4444950922899564e-05,
+      "loss": 0.9434,
+      "step": 8390
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4443661683638025e-05,
+      "loss": 0.8018,
+      "step": 8391
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4442372352337243e-05,
+      "loss": 1.0278,
+      "step": 8392
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4441082929023924e-05,
+      "loss": 0.8296,
+      "step": 8393
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4439793413724767e-05,
+      "loss": 0.917,
+      "step": 8394
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.443850380646649e-05,
+      "loss": 0.9194,
+      "step": 8395
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4437214107275799e-05,
+      "loss": 0.9629,
+      "step": 8396
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4435924316179406e-05,
+      "loss": 0.9722,
+      "step": 8397
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4434634433204029e-05,
+      "loss": 1.0952,
+      "step": 8398
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4433344458376384e-05,
+      "loss": 0.9165,
+      "step": 8399
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4432054391723186e-05,
+      "loss": 1.0967,
+      "step": 8400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4430764233271162e-05,
+      "loss": 0.9148,
+      "step": 8401
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4429473983047029e-05,
+      "loss": 0.8984,
+      "step": 8402
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4428183641077512e-05,
+      "loss": 0.9697,
+      "step": 8403
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4426893207389339e-05,
+      "loss": 0.8584,
+      "step": 8404
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4425602682009237e-05,
+      "loss": 1.0137,
+      "step": 8405
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4424312064963937e-05,
+      "loss": 0.9971,
+      "step": 8406
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4423021356280169e-05,
+      "loss": 1.0356,
+      "step": 8407
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4421730555984668e-05,
+      "loss": 0.9512,
+      "step": 8408
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4420439664104171e-05,
+      "loss": 1.0361,
+      "step": 8409
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4419148680665414e-05,
+      "loss": 0.9033,
+      "step": 8410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4417857605695136e-05,
+      "loss": 1.001,
+      "step": 8411
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4416566439220079e-05,
+      "loss": 0.8862,
+      "step": 8412
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4415275181266983e-05,
+      "loss": 1.0552,
+      "step": 8413
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4413983831862602e-05,
+      "loss": 0.9053,
+      "step": 8414
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4412692391033673e-05,
+      "loss": 0.9224,
+      "step": 8415
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4411400858806951e-05,
+      "loss": 0.9312,
+      "step": 8416
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4410109235209187e-05,
+      "loss": 0.9038,
+      "step": 8417
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4408817520267128e-05,
+      "loss": 1.0122,
+      "step": 8418
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4407525714007535e-05,
+      "loss": 0.9751,
+      "step": 8419
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4406233816457163e-05,
+      "loss": 0.9556,
+      "step": 8420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4404941827642766e-05,
+      "loss": 0.9219,
+      "step": 8421
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4403649747591111e-05,
+      "loss": 0.9805,
+      "step": 8422
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4402357576328954e-05,
+      "loss": 0.9409,
+      "step": 8423
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4401065313883064e-05,
+      "loss": 1.0449,
+      "step": 8424
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4399772960280204e-05,
+      "loss": 0.855,
+      "step": 8425
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4398480515547147e-05,
+      "loss": 0.9204,
+      "step": 8426
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4397187979710654e-05,
+      "loss": 0.9473,
+      "step": 8427
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4395895352797502e-05,
+      "loss": 0.8743,
+      "step": 8428
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4394602634834463e-05,
+      "loss": 0.8535,
+      "step": 8429
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4393309825848314e-05,
+      "loss": 0.8926,
+      "step": 8430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.439201692586583e-05,
+      "loss": 1.0776,
+      "step": 8431
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4390723934913793e-05,
+      "loss": 0.9443,
+      "step": 8432
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4389430853018982e-05,
+      "loss": 1.0273,
+      "step": 8433
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.438813768020818e-05,
+      "loss": 0.8765,
+      "step": 8434
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4386844416508172e-05,
+      "loss": 0.939,
+      "step": 8435
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4385551061945746e-05,
+      "loss": 1.0835,
+      "step": 8436
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4384257616547687e-05,
+      "loss": 0.9077,
+      "step": 8437
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4382964080340791e-05,
+      "loss": 0.9829,
+      "step": 8438
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4381670453351847e-05,
+      "loss": 0.9458,
+      "step": 8439
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4380376735607648e-05,
+      "loss": 0.9453,
+      "step": 8440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4379082927134995e-05,
+      "loss": 0.876,
+      "step": 8441
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4377789027960681e-05,
+      "loss": 0.8301,
+      "step": 8442
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4376495038111507e-05,
+      "loss": 0.9409,
+      "step": 8443
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4375200957614275e-05,
+      "loss": 0.9302,
+      "step": 8444
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4373906786495788e-05,
+      "loss": 0.9336,
+      "step": 8445
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4372612524782854e-05,
+      "loss": 0.8984,
+      "step": 8446
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4371318172502278e-05,
+      "loss": 1.061,
+      "step": 8447
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4370023729680868e-05,
+      "loss": 0.9214,
+      "step": 8448
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4368729196345438e-05,
+      "loss": 1.0259,
+      "step": 8449
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4367434572522802e-05,
+      "loss": 0.916,
+      "step": 8450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4366139858239772e-05,
+      "loss": 0.957,
+      "step": 8451
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4364845053523166e-05,
+      "loss": 1.043,
+      "step": 8452
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4363550158399802e-05,
+      "loss": 0.9209,
+      "step": 8453
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.43622551728965e-05,
+      "loss": 0.8574,
+      "step": 8454
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4360960097040085e-05,
+      "loss": 0.8818,
+      "step": 8455
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4359664930857378e-05,
+      "loss": 0.8833,
+      "step": 8456
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4358369674375208e-05,
+      "loss": 0.9238,
+      "step": 8457
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.43570743276204e-05,
+      "loss": 1.0586,
+      "step": 8458
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4355778890619784e-05,
+      "loss": 0.8906,
+      "step": 8459
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4354483363400194e-05,
+      "loss": 0.832,
+      "step": 8460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4353187745988462e-05,
+      "loss": 0.9897,
+      "step": 8461
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4351892038411425e-05,
+      "loss": 0.8643,
+      "step": 8462
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.435059624069592e-05,
+      "loss": 0.8579,
+      "step": 8463
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4349300352868783e-05,
+      "loss": 0.9426,
+      "step": 8464
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4348004374956859e-05,
+      "loss": 1.0137,
+      "step": 8465
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.434670830698699e-05,
+      "loss": 0.9614,
+      "step": 8466
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.434541214898602e-05,
+      "loss": 0.9658,
+      "step": 8467
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4344115900980796e-05,
+      "loss": 0.9341,
+      "step": 8468
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4342819562998166e-05,
+      "loss": 0.9282,
+      "step": 8469
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4341523135064983e-05,
+      "loss": 1.0308,
+      "step": 8470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4340226617208095e-05,
+      "loss": 0.918,
+      "step": 8471
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4338930009454358e-05,
+      "loss": 0.9878,
+      "step": 8472
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4337633311830626e-05,
+      "loss": 0.8608,
+      "step": 8473
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4336336524363761e-05,
+      "loss": 0.8936,
+      "step": 8474
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.433503964708062e-05,
+      "loss": 0.9678,
+      "step": 8475
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4333742680008067e-05,
+      "loss": 1.021,
+      "step": 8476
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4332445623172961e-05,
+      "loss": 0.9033,
+      "step": 8477
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.433114847660217e-05,
+      "loss": 0.9004,
+      "step": 8478
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4329851240322564e-05,
+      "loss": 0.96,
+      "step": 8479
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4328553914361006e-05,
+      "loss": 0.9399,
+      "step": 8480
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.432725649874437e-05,
+      "loss": 0.8594,
+      "step": 8481
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.432595899349953e-05,
+      "loss": 1.0176,
+      "step": 8482
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4324661398653357e-05,
+      "loss": 0.9678,
+      "step": 8483
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.432336371423273e-05,
+      "loss": 1.0234,
+      "step": 8484
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4322065940264526e-05,
+      "loss": 1.0322,
+      "step": 8485
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4320768076775626e-05,
+      "loss": 1.0044,
+      "step": 8486
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4319470123792913e-05,
+      "loss": 0.8804,
+      "step": 8487
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4318172081343269e-05,
+      "loss": 0.8931,
+      "step": 8488
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.431687394945358e-05,
+      "loss": 0.9673,
+      "step": 8489
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4315575728150735e-05,
+      "loss": 1.002,
+      "step": 8490
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.431427741746162e-05,
+      "loss": 0.9751,
+      "step": 8491
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4312979017413128e-05,
+      "loss": 0.9453,
+      "step": 8492
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4311680528032153e-05,
+      "loss": 0.9468,
+      "step": 8493
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.431038194934559e-05,
+      "loss": 0.9414,
+      "step": 8494
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4309083281380335e-05,
+      "loss": 0.9209,
+      "step": 8495
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4307784524163287e-05,
+      "loss": 1.0361,
+      "step": 8496
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4306485677721347e-05,
+      "loss": 0.9424,
+      "step": 8497
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4305186742081415e-05,
+      "loss": 0.9648,
+      "step": 8498
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4303887717270396e-05,
+      "loss": 0.9482,
+      "step": 8499
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.43025886033152e-05,
+      "loss": 0.9453,
+      "step": 8500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4301289400242729e-05,
+      "loss": 0.9307,
+      "step": 8501
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4299990108079896e-05,
+      "loss": 1.0146,
+      "step": 8502
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.429869072685361e-05,
+      "loss": 0.9238,
+      "step": 8503
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.429739125659079e-05,
+      "loss": 1.0024,
+      "step": 8504
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4296091697318346e-05,
+      "loss": 1.0562,
+      "step": 8505
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4294792049063198e-05,
+      "loss": 0.9658,
+      "step": 8506
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.429349231185226e-05,
+      "loss": 0.8838,
+      "step": 8507
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.429219248571246e-05,
+      "loss": 0.9731,
+      "step": 8508
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4290892570670713e-05,
+      "loss": 0.9478,
+      "step": 8509
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428959256675395e-05,
+      "loss": 1.0063,
+      "step": 8510
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4288292473989097e-05,
+      "loss": 0.9604,
+      "step": 8511
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4286992292403077e-05,
+      "loss": 0.9473,
+      "step": 8512
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4285692022022822e-05,
+      "loss": 0.8755,
+      "step": 8513
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4284391662875266e-05,
+      "loss": 0.8711,
+      "step": 8514
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4283091214987344e-05,
+      "loss": 0.9092,
+      "step": 8515
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4281790678385986e-05,
+      "loss": 0.8936,
+      "step": 8516
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4280490053098133e-05,
+      "loss": 0.9785,
+      "step": 8517
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4279189339150724e-05,
+      "loss": 0.9468,
+      "step": 8518
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4277888536570696e-05,
+      "loss": 1.0024,
+      "step": 8519
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4276587645384998e-05,
+      "loss": 0.9268,
+      "step": 8520
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.427528666562057e-05,
+      "loss": 1.001,
+      "step": 8521
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4273985597304365e-05,
+      "loss": 0.8828,
+      "step": 8522
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4272684440463322e-05,
+      "loss": 0.8716,
+      "step": 8523
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4271383195124399e-05,
+      "loss": 0.9663,
+      "step": 8524
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4270081861314542e-05,
+      "loss": 0.9697,
+      "step": 8525
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4268780439060707e-05,
+      "loss": 0.9497,
+      "step": 8526
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4267478928389855e-05,
+      "loss": 0.9292,
+      "step": 8527
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4266177329328939e-05,
+      "loss": 0.8628,
+      "step": 8528
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4264875641904915e-05,
+      "loss": 0.9858,
+      "step": 8529
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4263573866144751e-05,
+      "loss": 0.8892,
+      "step": 8530
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4262272002075404e-05,
+      "loss": 0.8594,
+      "step": 8531
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4260970049723842e-05,
+      "loss": 1.0693,
+      "step": 8532
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4259668009117031e-05,
+      "loss": 0.9272,
+      "step": 8533
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425836588028194e-05,
+      "loss": 0.9753,
+      "step": 8534
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4257063663245539e-05,
+      "loss": 0.9624,
+      "step": 8535
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.42557613580348e-05,
+      "loss": 1.0654,
+      "step": 8536
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4254458964676692e-05,
+      "loss": 0.8696,
+      "step": 8537
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.42531564831982e-05,
+      "loss": 0.876,
+      "step": 8538
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4251853913626296e-05,
+      "loss": 1.0283,
+      "step": 8539
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4250551255987963e-05,
+      "loss": 0.7578,
+      "step": 8540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4249248510310178e-05,
+      "loss": 0.98,
+      "step": 8541
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.424794567661993e-05,
+      "loss": 1.0444,
+      "step": 8542
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4246642754944196e-05,
+      "loss": 0.9824,
+      "step": 8543
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4245339745309968e-05,
+      "loss": 1.0605,
+      "step": 8544
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4244036647744234e-05,
+      "loss": 0.875,
+      "step": 8545
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4242733462273983e-05,
+      "loss": 1.0112,
+      "step": 8546
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4241430188926209e-05,
+      "loss": 0.9565,
+      "step": 8547
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4240126827727905e-05,
+      "loss": 0.9521,
+      "step": 8548
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4238823378706064e-05,
+      "loss": 0.9707,
+      "step": 8549
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4237519841887689e-05,
+      "loss": 0.9551,
+      "step": 8550
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4236216217299777e-05,
+      "loss": 0.832,
+      "step": 8551
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4234912504969328e-05,
+      "loss": 0.9766,
+      "step": 8552
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4233608704923345e-05,
+      "loss": 0.978,
+      "step": 8553
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4232304817188838e-05,
+      "loss": 0.9893,
+      "step": 8554
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4231000841792808e-05,
+      "loss": 0.938,
+      "step": 8555
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4229696778762266e-05,
+      "loss": 0.7676,
+      "step": 8556
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4228392628124222e-05,
+      "loss": 1.0444,
+      "step": 8557
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4227088389905688e-05,
+      "loss": 0.9072,
+      "step": 8558
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4225784064133677e-05,
+      "loss": 0.978,
+      "step": 8559
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4224479650835207e-05,
+      "loss": 0.9697,
+      "step": 8560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4223175150037297e-05,
+      "loss": 0.9697,
+      "step": 8561
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.422187056176696e-05,
+      "loss": 0.8018,
+      "step": 8562
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4220565886051223e-05,
+      "loss": 1.0718,
+      "step": 8563
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4219261122917108e-05,
+      "loss": 0.9722,
+      "step": 8564
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4217956272391638e-05,
+      "loss": 0.9946,
+      "step": 8565
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.421665133450184e-05,
+      "loss": 1.019,
+      "step": 8566
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4215346309274746e-05,
+      "loss": 0.8809,
+      "step": 8567
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4214041196737383e-05,
+      "loss": 1.0034,
+      "step": 8568
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4212735996916784e-05,
+      "loss": 0.9951,
+      "step": 8569
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4211430709839983e-05,
+      "loss": 1.0015,
+      "step": 8570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4210125335534013e-05,
+      "loss": 0.9536,
+      "step": 8571
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.420881987402592e-05,
+      "loss": 1.0078,
+      "step": 8572
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.420751432534273e-05,
+      "loss": 0.9668,
+      "step": 8573
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4206208689511495e-05,
+      "loss": 0.938,
+      "step": 8574
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4204902966559254e-05,
+      "loss": 0.895,
+      "step": 8575
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4203597156513053e-05,
+      "loss": 0.9585,
+      "step": 8576
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4202291259399939e-05,
+      "loss": 0.877,
+      "step": 8577
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4200985275246957e-05,
+      "loss": 0.8188,
+      "step": 8578
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.419967920408116e-05,
+      "loss": 0.9785,
+      "step": 8579
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4198373045929602e-05,
+      "loss": 0.9907,
+      "step": 8580
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4197066800819328e-05,
+      "loss": 0.9927,
+      "step": 8581
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4195760468777404e-05,
+      "loss": 1.0195,
+      "step": 8582
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.419445404983088e-05,
+      "loss": 0.9495,
+      "step": 8583
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4193147544006824e-05,
+      "loss": 0.9014,
+      "step": 8584
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4191840951332288e-05,
+      "loss": 1.0459,
+      "step": 8585
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4190534271834335e-05,
+      "loss": 0.9897,
+      "step": 8586
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4189227505540037e-05,
+      "loss": 0.9502,
+      "step": 8587
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4187920652476455e-05,
+      "loss": 0.8782,
+      "step": 8588
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4186613712670657e-05,
+      "loss": 0.9883,
+      "step": 8589
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4185306686149714e-05,
+      "loss": 0.9136,
+      "step": 8590
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4183999572940696e-05,
+      "loss": 0.9199,
+      "step": 8591
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4182692373070681e-05,
+      "loss": 1.0029,
+      "step": 8592
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4181385086566742e-05,
+      "loss": 0.8711,
+      "step": 8593
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4180077713455954e-05,
+      "loss": 0.9917,
+      "step": 8594
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.41787702537654e-05,
+      "loss": 0.8784,
+      "step": 8595
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4177462707522156e-05,
+      "loss": 0.939,
+      "step": 8596
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4176155074753309e-05,
+      "loss": 0.981,
+      "step": 8597
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4174847355485939e-05,
+      "loss": 0.9521,
+      "step": 8598
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4173539549747135e-05,
+      "loss": 0.9863,
+      "step": 8599
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4172231657563985e-05,
+      "loss": 0.9414,
+      "step": 8600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.417092367896358e-05,
+      "loss": 0.947,
+      "step": 8601
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4169615613973004e-05,
+      "loss": 0.9648,
+      "step": 8602
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416830746261936e-05,
+      "loss": 0.9155,
+      "step": 8603
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416699922492974e-05,
+      "loss": 0.9355,
+      "step": 8604
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416569090093124e-05,
+      "loss": 0.853,
+      "step": 8605
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4164382490650957e-05,
+      "loss": 0.9556,
+      "step": 8606
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416307399411599e-05,
+      "loss": 1.021,
+      "step": 8607
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4161765411353448e-05,
+      "loss": 0.9658,
+      "step": 8608
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4160456742390426e-05,
+      "loss": 0.9893,
+      "step": 8609
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4159147987254039e-05,
+      "loss": 1.0239,
+      "step": 8610
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4157839145971389e-05,
+      "loss": 0.8376,
+      "step": 8611
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4156530218569586e-05,
+      "loss": 0.9565,
+      "step": 8612
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4155221205075742e-05,
+      "loss": 0.8564,
+      "step": 8613
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4153912105516967e-05,
+      "loss": 1.0146,
+      "step": 8614
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4152602919920383e-05,
+      "loss": 0.8457,
+      "step": 8615
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.41512936483131e-05,
+      "loss": 0.8928,
+      "step": 8616
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.414998429072224e-05,
+      "loss": 0.9805,
+      "step": 8617
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4148674847174916e-05,
+      "loss": 0.873,
+      "step": 8618
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.414736531769826e-05,
+      "loss": 1.0093,
+      "step": 8619
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.414605570231939e-05,
+      "loss": 1.0273,
+      "step": 8620
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.414474600106543e-05,
+      "loss": 0.9126,
+      "step": 8621
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.414343621396351e-05,
+      "loss": 0.8999,
+      "step": 8622
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4142126341040758e-05,
+      "loss": 0.9468,
+      "step": 8623
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4140816382324305e-05,
+      "loss": 1.082,
+      "step": 8624
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4139506337841282e-05,
+      "loss": 1.1206,
+      "step": 8625
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4138196207618824e-05,
+      "loss": 0.9033,
+      "step": 8626
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4136885991684068e-05,
+      "loss": 0.9336,
+      "step": 8627
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4135575690064155e-05,
+      "loss": 1.0146,
+      "step": 8628
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4134265302786217e-05,
+      "loss": 0.9478,
+      "step": 8629
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4132954829877399e-05,
+      "loss": 0.8259,
+      "step": 8630
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4131644271364846e-05,
+      "loss": 0.9282,
+      "step": 8631
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4130333627275703e-05,
+      "loss": 0.9619,
+      "step": 8632
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4129022897637113e-05,
+      "loss": 0.9619,
+      "step": 8633
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4127712082476228e-05,
+      "loss": 0.9077,
+      "step": 8634
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4126401181820198e-05,
+      "loss": 0.9038,
+      "step": 8635
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.412509019569617e-05,
+      "loss": 1.062,
+      "step": 8636
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4123779124131302e-05,
+      "loss": 0.9478,
+      "step": 8637
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4122467967152752e-05,
+      "loss": 0.877,
+      "step": 8638
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4121156724787671e-05,
+      "loss": 0.9907,
+      "step": 8639
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4119845397063221e-05,
+      "loss": 1.0024,
+      "step": 8640
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4118533984006566e-05,
+      "loss": 0.9971,
+      "step": 8641
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4117222485644866e-05,
+      "loss": 0.9502,
+      "step": 8642
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4115910902005283e-05,
+      "loss": 0.9434,
+      "step": 8643
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4114599233114988e-05,
+      "loss": 1.0093,
+      "step": 8644
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4113287479001145e-05,
+      "loss": 1.0205,
+      "step": 8645
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4111975639690925e-05,
+      "loss": 0.9248,
+      "step": 8646
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4110663715211499e-05,
+      "loss": 1.0381,
+      "step": 8647
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4109351705590043e-05,
+      "loss": 0.9526,
+      "step": 8648
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4108039610853727e-05,
+      "loss": 0.9761,
+      "step": 8649
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4106727431029734e-05,
+      "loss": 0.9683,
+      "step": 8650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4105415166145233e-05,
+      "loss": 0.8262,
+      "step": 8651
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4104102816227414e-05,
+      "loss": 0.9653,
+      "step": 8652
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4102790381303454e-05,
+      "loss": 0.9297,
+      "step": 8653
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.410147786140054e-05,
+      "loss": 0.9609,
+      "step": 8654
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4100165256545855e-05,
+      "loss": 1.0312,
+      "step": 8655
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4098852566766586e-05,
+      "loss": 0.9009,
+      "step": 8656
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.409753979208992e-05,
+      "loss": 0.8848,
+      "step": 8657
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4096226932543056e-05,
+      "loss": 1.0298,
+      "step": 8658
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4094913988153178e-05,
+      "loss": 0.9023,
+      "step": 8659
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4093600958947484e-05,
+      "loss": 0.8232,
+      "step": 8660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.409228784495317e-05,
+      "loss": 1.0195,
+      "step": 8661
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4090974646197432e-05,
+      "loss": 1.085,
+      "step": 8662
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4089661362707473e-05,
+      "loss": 0.9575,
+      "step": 8663
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4088347994510491e-05,
+      "loss": 0.9556,
+      "step": 8664
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4087034541633692e-05,
+      "loss": 0.8276,
+      "step": 8665
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.408572100410428e-05,
+      "loss": 0.9507,
+      "step": 8666
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.408440738194946e-05,
+      "loss": 0.8794,
+      "step": 8667
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4083093675196442e-05,
+      "loss": 0.9229,
+      "step": 8668
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4081779883872435e-05,
+      "loss": 0.9614,
+      "step": 8669
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4080466008004652e-05,
+      "loss": 0.8794,
+      "step": 8670
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4079152047620303e-05,
+      "loss": 0.9292,
+      "step": 8671
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4077838002746611e-05,
+      "loss": 0.894,
+      "step": 8672
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4076523873410786e-05,
+      "loss": 0.8906,
+      "step": 8673
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4075209659640049e-05,
+      "loss": 0.939,
+      "step": 8674
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4073895361461622e-05,
+      "loss": 0.8889,
+      "step": 8675
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4072580978902726e-05,
+      "loss": 1.0234,
+      "step": 8676
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4071266511990587e-05,
+      "loss": 0.9624,
+      "step": 8677
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4069951960752426e-05,
+      "loss": 0.8716,
+      "step": 8678
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4068637325215477e-05,
+      "loss": 0.999,
+      "step": 8679
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4067322605406965e-05,
+      "loss": 1.0342,
+      "step": 8680
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4066007801354125e-05,
+      "loss": 0.9517,
+      "step": 8681
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4064692913084184e-05,
+      "loss": 0.9146,
+      "step": 8682
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4063377940624381e-05,
+      "loss": 0.9937,
+      "step": 8683
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4062062884001952e-05,
+      "loss": 0.9458,
+      "step": 8684
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4060747743244134e-05,
+      "loss": 0.875,
+      "step": 8685
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4059432518378167e-05,
+      "loss": 0.979,
+      "step": 8686
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4058117209431292e-05,
+      "loss": 0.8909,
+      "step": 8687
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4056801816430755e-05,
+      "loss": 0.9561,
+      "step": 8688
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.40554863394038e-05,
+      "loss": 1.0444,
+      "step": 8689
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4054170778377669e-05,
+      "loss": 0.9536,
+      "step": 8690
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4052855133379613e-05,
+      "loss": 0.9204,
+      "step": 8691
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4051539404436886e-05,
+      "loss": 0.9077,
+      "step": 8692
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.405022359157674e-05,
+      "loss": 0.9722,
+      "step": 8693
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4048907694826428e-05,
+      "loss": 0.769,
+      "step": 8694
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.40475917142132e-05,
+      "loss": 0.8994,
+      "step": 8695
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4046275649764318e-05,
+      "loss": 0.9805,
+      "step": 8696
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.404495950150704e-05,
+      "loss": 0.9473,
+      "step": 8697
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4043643269468628e-05,
+      "loss": 0.9766,
+      "step": 8698
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4042326953676342e-05,
+      "loss": 0.9282,
+      "step": 8699
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4041010554157449e-05,
+      "loss": 0.9121,
+      "step": 8700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403969407093921e-05,
+      "loss": 0.9951,
+      "step": 8701
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4038377504048897e-05,
+      "loss": 0.9287,
+      "step": 8702
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4037060853513778e-05,
+      "loss": 1.0142,
+      "step": 8703
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4035744119361124e-05,
+      "loss": 0.9683,
+      "step": 8704
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403442730161821e-05,
+      "loss": 0.8628,
+      "step": 8705
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4033110400312308e-05,
+      "loss": 1.1328,
+      "step": 8706
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4031793415470697e-05,
+      "loss": 1.0059,
+      "step": 8707
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4030476347120651e-05,
+      "loss": 1.0312,
+      "step": 8708
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4029159195289452e-05,
+      "loss": 0.9692,
+      "step": 8709
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4027841960004383e-05,
+      "loss": 0.9692,
+      "step": 8710
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4026524641292723e-05,
+      "loss": 0.9292,
+      "step": 8711
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.402520723918176e-05,
+      "loss": 0.917,
+      "step": 8712
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4023889753698781e-05,
+      "loss": 0.9985,
+      "step": 8713
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4022572184871074e-05,
+      "loss": 0.8994,
+      "step": 8714
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4021254532725926e-05,
+      "loss": 0.9209,
+      "step": 8715
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.401993679729063e-05,
+      "loss": 0.8569,
+      "step": 8716
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4018618978592484e-05,
+      "loss": 0.9165,
+      "step": 8717
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.401730107665878e-05,
+      "loss": 0.9146,
+      "step": 8718
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4015983091516815e-05,
+      "loss": 0.9331,
+      "step": 8719
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4014665023193889e-05,
+      "loss": 1.0625,
+      "step": 8720
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4013346871717301e-05,
+      "loss": 0.9731,
+      "step": 8721
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4012028637114355e-05,
+      "loss": 0.8628,
+      "step": 8722
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4010710319412354e-05,
+      "loss": 0.8755,
+      "step": 8723
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4009391918638602e-05,
+      "loss": 1.0327,
+      "step": 8724
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4008073434820406e-05,
+      "loss": 0.9414,
+      "step": 8725
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4006754867985077e-05,
+      "loss": 0.9502,
+      "step": 8726
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4005436218159927e-05,
+      "loss": 0.938,
+      "step": 8727
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4004117485372261e-05,
+      "loss": 1.0039,
+      "step": 8728
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4002798669649407e-05,
+      "loss": 0.9648,
+      "step": 8729
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.400147977101867e-05,
+      "loss": 0.9526,
+      "step": 8730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4000160789507371e-05,
+      "loss": 0.9736,
+      "step": 8731
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3998841725142829e-05,
+      "loss": 0.9663,
+      "step": 8732
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3997522577952365e-05,
+      "loss": 0.9897,
+      "step": 8733
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3996203347963301e-05,
+      "loss": 0.9829,
+      "step": 8734
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3994884035202965e-05,
+      "loss": 0.824,
+      "step": 8735
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.399356463969868e-05,
+      "loss": 0.9736,
+      "step": 8736
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3992245161477773e-05,
+      "loss": 1.0776,
+      "step": 8737
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.399092560056758e-05,
+      "loss": 0.9932,
+      "step": 8738
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3989605956995425e-05,
+      "loss": 0.897,
+      "step": 8739
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.398828623078864e-05,
+      "loss": 0.9478,
+      "step": 8740
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3986966421974567e-05,
+      "loss": 1.0,
+      "step": 8741
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3985646530580538e-05,
+      "loss": 0.9531,
+      "step": 8742
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3984326556633895e-05,
+      "loss": 0.9731,
+      "step": 8743
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3983006500161972e-05,
+      "loss": 0.8472,
+      "step": 8744
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3981686361192117e-05,
+      "loss": 1.0327,
+      "step": 8745
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.398036613975167e-05,
+      "loss": 0.9541,
+      "step": 8746
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3979045835867975e-05,
+      "loss": 0.9595,
+      "step": 8747
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3977725449568381e-05,
+      "loss": 0.9199,
+      "step": 8748
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3976404980880236e-05,
+      "loss": 0.9888,
+      "step": 8749
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3975084429830888e-05,
+      "loss": 0.9321,
+      "step": 8750
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3973763796447693e-05,
+      "loss": 1.021,
+      "step": 8751
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3972443080758e-05,
+      "loss": 0.9565,
+      "step": 8752
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3971122282789167e-05,
+      "loss": 0.9014,
+      "step": 8753
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3969801402568552e-05,
+      "loss": 0.8428,
+      "step": 8754
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.396848044012351e-05,
+      "loss": 0.8921,
+      "step": 8755
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3967159395481406e-05,
+      "loss": 0.9473,
+      "step": 8756
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.39658382686696e-05,
+      "loss": 0.9331,
+      "step": 8757
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3964517059715455e-05,
+      "loss": 0.8804,
+      "step": 8758
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3963195768646336e-05,
+      "loss": 0.9919,
+      "step": 8759
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3961874395489612e-05,
+      "loss": 0.8682,
+      "step": 8760
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3960552940272652e-05,
+      "loss": 0.8457,
+      "step": 8761
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3959231403022823e-05,
+      "loss": 0.8945,
+      "step": 8762
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3957909783767503e-05,
+      "loss": 0.9517,
+      "step": 8763
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3956588082534062e-05,
+      "loss": 0.9834,
+      "step": 8764
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3955266299349877e-05,
+      "loss": 0.9272,
+      "step": 8765
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3953944434242326e-05,
+      "loss": 0.9194,
+      "step": 8766
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3952622487238787e-05,
+      "loss": 0.8721,
+      "step": 8767
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.395130045836664e-05,
+      "loss": 1.0132,
+      "step": 8768
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3949978347653267e-05,
+      "loss": 0.9761,
+      "step": 8769
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3948656155126058e-05,
+      "loss": 0.9731,
+      "step": 8770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3947333880812392e-05,
+      "loss": 0.854,
+      "step": 8771
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3946011524739662e-05,
+      "loss": 0.8643,
+      "step": 8772
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3944689086935252e-05,
+      "loss": 0.9683,
+      "step": 8773
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3943366567426558e-05,
+      "loss": 1.1006,
+      "step": 8774
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3942043966240968e-05,
+      "loss": 0.9214,
+      "step": 8775
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.394072128340588e-05,
+      "loss": 0.9971,
+      "step": 8776
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3939398518948688e-05,
+      "loss": 0.8872,
+      "step": 8777
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3938075672896791e-05,
+      "loss": 0.9692,
+      "step": 8778
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3936752745277586e-05,
+      "loss": 0.9697,
+      "step": 8779
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3935429736118475e-05,
+      "loss": 0.937,
+      "step": 8780
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3934106645446864e-05,
+      "loss": 0.8462,
+      "step": 8781
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3932783473290154e-05,
+      "loss": 1.0005,
+      "step": 8782
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3931460219675755e-05,
+      "loss": 0.9897,
+      "step": 8783
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3930136884631069e-05,
+      "loss": 0.9155,
+      "step": 8784
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3928813468183511e-05,
+      "loss": 1.0381,
+      "step": 8785
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3927489970360489e-05,
+      "loss": 0.9272,
+      "step": 8786
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3926166391189418e-05,
+      "loss": 0.9604,
+      "step": 8787
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.392484273069771e-05,
+      "loss": 1.0605,
+      "step": 8788
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3923518988912781e-05,
+      "loss": 0.8506,
+      "step": 8789
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3922195165862053e-05,
+      "loss": 0.9282,
+      "step": 8790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3920871261572944e-05,
+      "loss": 0.8777,
+      "step": 8791
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3919547276072874e-05,
+      "loss": 0.9893,
+      "step": 8792
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3918223209389263e-05,
+      "loss": 0.9331,
+      "step": 8793
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3916899061549542e-05,
+      "loss": 0.8972,
+      "step": 8794
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3915574832581139e-05,
+      "loss": 0.8867,
+      "step": 8795
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3914250522511474e-05,
+      "loss": 0.9429,
+      "step": 8796
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3912926131367983e-05,
+      "loss": 0.9536,
+      "step": 8797
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3911601659178094e-05,
+      "loss": 1.0537,
+      "step": 8798
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.391027710596924e-05,
+      "loss": 0.7683,
+      "step": 8799
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3908952471768858e-05,
+      "loss": 0.9448,
+      "step": 8800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3907627756604383e-05,
+      "loss": 0.856,
+      "step": 8801
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3906302960503253e-05,
+      "loss": 0.8623,
+      "step": 8802
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3904978083492908e-05,
+      "loss": 1.0625,
+      "step": 8803
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3903653125600792e-05,
+      "loss": 0.8569,
+      "step": 8804
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3902328086854342e-05,
+      "loss": 0.96,
+      "step": 8805
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.390100296728101e-05,
+      "loss": 0.9897,
+      "step": 8806
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3899677766908237e-05,
+      "loss": 0.9243,
+      "step": 8807
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3898352485763477e-05,
+      "loss": 0.8804,
+      "step": 8808
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3897027123874174e-05,
+      "loss": 0.9307,
+      "step": 8809
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3895701681267784e-05,
+      "loss": 1.0186,
+      "step": 8810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3894376157971756e-05,
+      "loss": 0.9526,
+      "step": 8811
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.389305055401355e-05,
+      "loss": 0.9316,
+      "step": 8812
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3891724869420618e-05,
+      "loss": 1.0347,
+      "step": 8813
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3890399104220418e-05,
+      "loss": 0.9927,
+      "step": 8814
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3889073258440413e-05,
+      "loss": 0.9922,
+      "step": 8815
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3887747332108064e-05,
+      "loss": 1.0005,
+      "step": 8816
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.388642132525083e-05,
+      "loss": 0.939,
+      "step": 8817
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.388509523789618e-05,
+      "loss": 0.96,
+      "step": 8818
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3883769070071583e-05,
+      "loss": 0.9902,
+      "step": 8819
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3882442821804502e-05,
+      "loss": 0.8203,
+      "step": 8820
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3881116493122411e-05,
+      "loss": 0.8857,
+      "step": 8821
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3879790084052778e-05,
+      "loss": 0.9038,
+      "step": 8822
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3878463594623077e-05,
+      "loss": 0.8379,
+      "step": 8823
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3877137024860786e-05,
+      "loss": 1.0288,
+      "step": 8824
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3875810374793374e-05,
+      "loss": 0.8513,
+      "step": 8825
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3874483644448329e-05,
+      "loss": 0.9077,
+      "step": 8826
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3873156833853127e-05,
+      "loss": 1.0146,
+      "step": 8827
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3871829943035247e-05,
+      "loss": 0.8691,
+      "step": 8828
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3870502972022175e-05,
+      "loss": 0.9521,
+      "step": 8829
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3869175920841392e-05,
+      "loss": 0.9111,
+      "step": 8830
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3867848789520389e-05,
+      "loss": 1.001,
+      "step": 8831
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3866521578086652e-05,
+      "loss": 0.8506,
+      "step": 8832
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3865194286567671e-05,
+      "loss": 0.9819,
+      "step": 8833
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.386386691499094e-05,
+      "loss": 1.1157,
+      "step": 8834
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3862539463383948e-05,
+      "loss": 1.0,
+      "step": 8835
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3861211931774193e-05,
+      "loss": 0.9683,
+      "step": 8836
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.385988432018917e-05,
+      "loss": 0.7773,
+      "step": 8837
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3858556628656377e-05,
+      "loss": 0.8691,
+      "step": 8838
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3857228857203316e-05,
+      "loss": 0.8989,
+      "step": 8839
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3855901005857484e-05,
+      "loss": 0.8564,
+      "step": 8840
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3854573074646388e-05,
+      "loss": 0.9409,
+      "step": 8841
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3853245063597532e-05,
+      "loss": 1.0071,
+      "step": 8842
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3851916972738421e-05,
+      "loss": 0.9092,
+      "step": 8843
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3850588802096563e-05,
+      "loss": 0.9275,
+      "step": 8844
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.384926055169947e-05,
+      "loss": 0.8774,
+      "step": 8845
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3847932221574652e-05,
+      "loss": 0.9707,
+      "step": 8846
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3846603811749621e-05,
+      "loss": 0.8706,
+      "step": 8847
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3845275322251897e-05,
+      "loss": 1.043,
+      "step": 8848
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3843946753108988e-05,
+      "loss": 1.0012,
+      "step": 8849
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3842618104348416e-05,
+      "loss": 0.9194,
+      "step": 8850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3841289375997699e-05,
+      "loss": 1.0205,
+      "step": 8851
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3839960568084363e-05,
+      "loss": 1.0366,
+      "step": 8852
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3838631680635928e-05,
+      "loss": 0.9429,
+      "step": 8853
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3837302713679916e-05,
+      "loss": 0.9077,
+      "step": 8854
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3835973667243855e-05,
+      "loss": 0.9028,
+      "step": 8855
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3834644541355275e-05,
+      "loss": 0.8691,
+      "step": 8856
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3833315336041704e-05,
+      "loss": 0.998,
+      "step": 8857
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3831986051330672e-05,
+      "loss": 0.9448,
+      "step": 8858
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3830656687249713e-05,
+      "loss": 0.9346,
+      "step": 8859
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.382932724382636e-05,
+      "loss": 1.0117,
+      "step": 8860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.382799772108815e-05,
+      "loss": 0.9863,
+      "step": 8861
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3826668119062626e-05,
+      "loss": 0.8931,
+      "step": 8862
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3825338437777319e-05,
+      "loss": 0.9194,
+      "step": 8863
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.382400867725977e-05,
+      "loss": 0.9883,
+      "step": 8864
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3822678837537529e-05,
+      "loss": 0.9004,
+      "step": 8865
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3821348918638133e-05,
+      "loss": 1.0562,
+      "step": 8866
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3820018920589133e-05,
+      "loss": 0.9468,
+      "step": 8867
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3818688843418076e-05,
+      "loss": 1.0488,
+      "step": 8868
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381735868715251e-05,
+      "loss": 0.8857,
+      "step": 8869
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381602845181998e-05,
+      "loss": 0.9487,
+      "step": 8870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3814698137448048e-05,
+      "loss": 0.9658,
+      "step": 8871
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3813367744064268e-05,
+      "loss": 0.8091,
+      "step": 8872
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3812037271696186e-05,
+      "loss": 0.8975,
+      "step": 8873
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381070672037137e-05,
+      "loss": 0.9072,
+      "step": 8874
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.380937609011737e-05,
+      "loss": 0.9751,
+      "step": 8875
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3808045380961754e-05,
+      "loss": 0.9009,
+      "step": 8876
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.380671459293208e-05,
+      "loss": 0.8911,
+      "step": 8877
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3805383726055914e-05,
+      "loss": 0.9868,
+      "step": 8878
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3804052780360822e-05,
+      "loss": 0.9146,
+      "step": 8879
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3802721755874369e-05,
+      "loss": 0.8848,
+      "step": 8880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3801390652624123e-05,
+      "loss": 1.0039,
+      "step": 8881
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3800059470637655e-05,
+      "loss": 0.8833,
+      "step": 8882
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3798728209942543e-05,
+      "loss": 0.9565,
+      "step": 8883
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3797396870566353e-05,
+      "loss": 0.74,
+      "step": 8884
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3796065452536665e-05,
+      "loss": 0.9912,
+      "step": 8885
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3794733955881052e-05,
+      "loss": 0.8254,
+      "step": 8886
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3793402380627099e-05,
+      "loss": 0.7603,
+      "step": 8887
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.379207072680238e-05,
+      "loss": 0.9492,
+      "step": 8888
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3790738994434479e-05,
+      "loss": 0.9658,
+      "step": 8889
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3789407183550979e-05,
+      "loss": 0.9614,
+      "step": 8890
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3788075294179467e-05,
+      "loss": 0.8652,
+      "step": 8891
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3786743326347524e-05,
+      "loss": 0.9551,
+      "step": 8892
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3785411280082747e-05,
+      "loss": 1.0225,
+      "step": 8893
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3784079155412719e-05,
+      "loss": 0.874,
+      "step": 8894
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3782746952365033e-05,
+      "loss": 0.8418,
+      "step": 8895
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3781414670967288e-05,
+      "loss": 0.979,
+      "step": 8896
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3780082311247072e-05,
+      "loss": 0.8171,
+      "step": 8897
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3778749873231984e-05,
+      "loss": 0.981,
+      "step": 8898
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3777417356949622e-05,
+      "loss": 0.9971,
+      "step": 8899
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3776084762427586e-05,
+      "loss": 0.9453,
+      "step": 8900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.377475208969348e-05,
+      "loss": 0.999,
+      "step": 8901
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3773419338774901e-05,
+      "loss": 0.8799,
+      "step": 8902
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3772086509699453e-05,
+      "loss": 1.0835,
+      "step": 8903
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3770753602494749e-05,
+      "loss": 0.98,
+      "step": 8904
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3769420617188394e-05,
+      "loss": 0.8024,
+      "step": 8905
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3768087553807995e-05,
+      "loss": 0.9478,
+      "step": 8906
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3766754412381164e-05,
+      "loss": 1.0039,
+      "step": 8907
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3765421192935518e-05,
+      "loss": 1.021,
+      "step": 8908
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3764087895498665e-05,
+      "loss": 1.0938,
+      "step": 8909
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3762754520098225e-05,
+      "loss": 0.9097,
+      "step": 8910
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3761421066761815e-05,
+      "loss": 1.0103,
+      "step": 8911
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3760087535517053e-05,
+      "loss": 0.9277,
+      "step": 8912
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.375875392639156e-05,
+      "loss": 0.8425,
+      "step": 8913
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3757420239412957e-05,
+      "loss": 0.897,
+      "step": 8914
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3756086474608871e-05,
+      "loss": 1.0708,
+      "step": 8915
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.375475263200693e-05,
+      "loss": 0.9771,
+      "step": 8916
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3753418711634751e-05,
+      "loss": 0.9224,
+      "step": 8917
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3752084713519973e-05,
+      "loss": 0.9907,
+      "step": 8918
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3750750637690218e-05,
+      "loss": 1.0352,
+      "step": 8919
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3749416484173126e-05,
+      "loss": 0.9551,
+      "step": 8920
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3748082252996324e-05,
+      "loss": 0.96,
+      "step": 8921
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3746747944187454e-05,
+      "loss": 0.9341,
+      "step": 8922
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3745413557774148e-05,
+      "loss": 1.0054,
+      "step": 8923
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3744079093784044e-05,
+      "loss": 0.8232,
+      "step": 8924
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3742744552244784e-05,
+      "loss": 0.957,
+      "step": 8925
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3741409933184011e-05,
+      "loss": 0.978,
+      "step": 8926
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3740075236629364e-05,
+      "loss": 1.0093,
+      "step": 8927
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3738740462608493e-05,
+      "loss": 0.9902,
+      "step": 8928
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3737405611149042e-05,
+      "loss": 0.8555,
+      "step": 8929
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3736070682278658e-05,
+      "loss": 1.0845,
+      "step": 8930
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3734735676024991e-05,
+      "loss": 0.9629,
+      "step": 8931
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3733400592415693e-05,
+      "loss": 0.9473,
+      "step": 8932
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3732065431478418e-05,
+      "loss": 0.9683,
+      "step": 8933
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.373073019324082e-05,
+      "loss": 0.9146,
+      "step": 8934
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3729394877730553e-05,
+      "loss": 0.896,
+      "step": 8935
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3728059484975278e-05,
+      "loss": 0.8896,
+      "step": 8936
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.372672401500265e-05,
+      "loss": 1.0444,
+      "step": 8937
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3725388467840333e-05,
+      "loss": 0.8931,
+      "step": 8938
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3724052843515989e-05,
+      "loss": 0.8726,
+      "step": 8939
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3722717142057283e-05,
+      "loss": 0.9126,
+      "step": 8940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3721381363491877e-05,
+      "loss": 0.9421,
+      "step": 8941
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3720045507847444e-05,
+      "loss": 0.9644,
+      "step": 8942
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3718709575151647e-05,
+      "loss": 0.9141,
+      "step": 8943
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3717373565432161e-05,
+      "loss": 0.8418,
+      "step": 8944
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3716037478716657e-05,
+      "loss": 0.9863,
+      "step": 8945
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3714701315032807e-05,
+      "loss": 0.877,
+      "step": 8946
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3713365074408288e-05,
+      "loss": 0.8643,
+      "step": 8947
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3712028756870774e-05,
+      "loss": 0.8237,
+      "step": 8948
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3710692362447946e-05,
+      "loss": 0.9165,
+      "step": 8949
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3709355891167487e-05,
+      "loss": 0.8572,
+      "step": 8950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.370801934305707e-05,
+      "loss": 0.9678,
+      "step": 8951
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3706682718144387e-05,
+      "loss": 0.9685,
+      "step": 8952
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3705346016457117e-05,
+      "loss": 0.8533,
+      "step": 8953
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.370400923802295e-05,
+      "loss": 0.8711,
+      "step": 8954
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3702672382869573e-05,
+      "loss": 0.9668,
+      "step": 8955
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3701335451024676e-05,
+      "loss": 0.7915,
+      "step": 8956
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3699998442515947e-05,
+      "loss": 0.9766,
+      "step": 8957
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3698661357371084e-05,
+      "loss": 0.8896,
+      "step": 8958
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3697324195617774e-05,
+      "loss": 0.9956,
+      "step": 8959
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.369598695728372e-05,
+      "loss": 0.8862,
+      "step": 8960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.369464964239662e-05,
+      "loss": 0.9531,
+      "step": 8961
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3693312250984168e-05,
+      "loss": 1.0435,
+      "step": 8962
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3691974783074065e-05,
+      "loss": 0.9424,
+      "step": 8963
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3690637238694015e-05,
+      "loss": 1.0054,
+      "step": 8964
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3689299617871726e-05,
+      "loss": 1.0322,
+      "step": 8965
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3687961920634896e-05,
+      "loss": 1.0171,
+      "step": 8966
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3686624147011236e-05,
+      "loss": 0.8379,
+      "step": 8967
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3685286297028453e-05,
+      "loss": 0.8647,
+      "step": 8968
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368394837071426e-05,
+      "loss": 0.9194,
+      "step": 8969
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3682610368096366e-05,
+      "loss": 1.0166,
+      "step": 8970
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368127228920248e-05,
+      "loss": 0.9385,
+      "step": 8971
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3679934134060329e-05,
+      "loss": 0.8708,
+      "step": 8972
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3678595902697623e-05,
+      "loss": 0.938,
+      "step": 8973
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3677257595142077e-05,
+      "loss": 0.9971,
+      "step": 8974
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3675919211421413e-05,
+      "loss": 0.9053,
+      "step": 8975
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3674580751563357e-05,
+      "loss": 0.8767,
+      "step": 8976
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3673242215595626e-05,
+      "loss": 0.9536,
+      "step": 8977
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3671903603545944e-05,
+      "loss": 1.0352,
+      "step": 8978
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.367056491544204e-05,
+      "loss": 0.9214,
+      "step": 8979
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3669226151311639e-05,
+      "loss": 0.8066,
+      "step": 8980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3667887311182472e-05,
+      "loss": 1.0591,
+      "step": 8981
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.366654839508227e-05,
+      "loss": 0.9512,
+      "step": 8982
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3665209403038767e-05,
+      "loss": 0.9434,
+      "step": 8983
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3663870335079687e-05,
+      "loss": 1.0078,
+      "step": 8984
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3662531191232777e-05,
+      "loss": 0.9697,
+      "step": 8985
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3661191971525772e-05,
+      "loss": 0.9448,
+      "step": 8986
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3659852675986407e-05,
+      "loss": 0.9399,
+      "step": 8987
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3658513304642422e-05,
+      "loss": 0.9214,
+      "step": 8988
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3657173857521561e-05,
+      "loss": 1.0049,
+      "step": 8989
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3655834334651567e-05,
+      "loss": 0.8188,
+      "step": 8990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3654494736060186e-05,
+      "loss": 0.9375,
+      "step": 8991
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3653155061775163e-05,
+      "loss": 1.0356,
+      "step": 8992
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.365181531182424e-05,
+      "loss": 0.9028,
+      "step": 8993
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3650475486235178e-05,
+      "loss": 0.9849,
+      "step": 8994
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.364913558503572e-05,
+      "loss": 0.8657,
+      "step": 8995
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3647795608253617e-05,
+      "loss": 0.9146,
+      "step": 8996
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3646455555916633e-05,
+      "loss": 0.9487,
+      "step": 8997
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3645115428052518e-05,
+      "loss": 0.9116,
+      "step": 8998
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3643775224689026e-05,
+      "loss": 0.9946,
+      "step": 8999
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3642434945853921e-05,
+      "loss": 0.9609,
+      "step": 9000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3641094591574962e-05,
+      "loss": 0.9199,
+      "step": 9001
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.363975416187991e-05,
+      "loss": 0.8853,
+      "step": 9002
+    },
+    {
+      "epoch": 2.0,
+      "eval_cIoU": 0.7799340486526489,
+      "eval_gIoU": 0.7781411409378052,
+      "eval_runtime": 120.2967,
+      "eval_samples_per_second": 31.68,
+      "eval_steps_per_second": 3.965,
+      "step": 9002
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3638413656796532e-05,
+      "loss": 0.7603,
+      "step": 9003
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.363707307635259e-05,
+      "loss": 0.916,
+      "step": 9004
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3635732420575851e-05,
+      "loss": 0.8218,
+      "step": 9005
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3634391689494082e-05,
+      "loss": 0.7554,
+      "step": 9006
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3633050883135057e-05,
+      "loss": 0.9092,
+      "step": 9007
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.363171000152654e-05,
+      "loss": 0.8262,
+      "step": 9008
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3630369044696315e-05,
+      "loss": 0.7754,
+      "step": 9009
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3629028012672148e-05,
+      "loss": 0.7224,
+      "step": 9010
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3627686905481819e-05,
+      "loss": 0.8149,
+      "step": 9011
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3626345723153103e-05,
+      "loss": 0.7466,
+      "step": 9012
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.362500446571378e-05,
+      "loss": 0.8159,
+      "step": 9013
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3623663133191632e-05,
+      "loss": 0.8149,
+      "step": 9014
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3622321725614441e-05,
+      "loss": 0.7837,
+      "step": 9015
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.362098024300999e-05,
+      "loss": 0.7798,
+      "step": 9016
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3619638685406066e-05,
+      "loss": 0.7622,
+      "step": 9017
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3618297052830451e-05,
+      "loss": 0.7466,
+      "step": 9018
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3616955345310942e-05,
+      "loss": 0.7192,
+      "step": 9019
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3615613562875323e-05,
+      "loss": 0.7402,
+      "step": 9020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3614271705551388e-05,
+      "loss": 0.7617,
+      "step": 9021
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3612929773366928e-05,
+      "loss": 0.7371,
+      "step": 9022
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3611587766349738e-05,
+      "loss": 0.7871,
+      "step": 9023
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3610245684527618e-05,
+      "loss": 0.8662,
+      "step": 9024
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3608903527928361e-05,
+      "loss": 0.8088,
+      "step": 9025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.360756129657977e-05,
+      "loss": 0.8394,
+      "step": 9026
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3606218990509645e-05,
+      "loss": 0.7378,
+      "step": 9027
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3604876609745788e-05,
+      "loss": 0.834,
+      "step": 9028
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3603534154316e-05,
+      "loss": 0.7334,
+      "step": 9029
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3602191624248092e-05,
+      "loss": 0.8076,
+      "step": 9030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3600849019569869e-05,
+      "loss": 0.6545,
+      "step": 9031
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3599506340309138e-05,
+      "loss": 0.7847,
+      "step": 9032
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3598163586493714e-05,
+      "loss": 0.8232,
+      "step": 9033
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3596820758151401e-05,
+      "loss": 0.8088,
+      "step": 9034
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.359547785531002e-05,
+      "loss": 0.7402,
+      "step": 9035
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.359413487799738e-05,
+      "loss": 0.7183,
+      "step": 9036
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.35927918262413e-05,
+      "loss": 0.7441,
+      "step": 9037
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.35914487000696e-05,
+      "loss": 0.834,
+      "step": 9038
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3590105499510097e-05,
+      "loss": 0.8267,
+      "step": 9039
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3588762224590612e-05,
+      "loss": 0.7429,
+      "step": 9040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.358741887533897e-05,
+      "loss": 0.8037,
+      "step": 9041
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3586075451782988e-05,
+      "loss": 0.7397,
+      "step": 9042
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3584731953950501e-05,
+      "loss": 0.7734,
+      "step": 9043
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3583388381869329e-05,
+      "loss": 0.8408,
+      "step": 9044
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3582044735567306e-05,
+      "loss": 0.7041,
+      "step": 9045
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3580701015072258e-05,
+      "loss": 0.8203,
+      "step": 9046
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3579357220412017e-05,
+      "loss": 0.7756,
+      "step": 9047
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3578013351614418e-05,
+      "loss": 0.6992,
+      "step": 9048
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3576669408707299e-05,
+      "loss": 0.8306,
+      "step": 9049
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3575325391718493e-05,
+      "loss": 0.7861,
+      "step": 9050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3573981300675836e-05,
+      "loss": 0.7815,
+      "step": 9051
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.357263713560717e-05,
+      "loss": 0.7637,
+      "step": 9052
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3571292896540335e-05,
+      "loss": 0.8809,
+      "step": 9053
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3569948583503174e-05,
+      "loss": 0.8672,
+      "step": 9054
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3568604196523531e-05,
+      "loss": 0.677,
+      "step": 9055
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3567259735629254e-05,
+      "loss": 0.7031,
+      "step": 9056
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3565915200848186e-05,
+      "loss": 0.7778,
+      "step": 9057
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3564570592208178e-05,
+      "loss": 0.7905,
+      "step": 9058
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3563225909737076e-05,
+      "loss": 0.8042,
+      "step": 9059
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3561881153462739e-05,
+      "loss": 0.7388,
+      "step": 9060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3560536323413014e-05,
+      "loss": 0.7104,
+      "step": 9061
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355919141961576e-05,
+      "loss": 0.7681,
+      "step": 9062
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355784644209883e-05,
+      "loss": 0.6316,
+      "step": 9063
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3556501390890087e-05,
+      "loss": 0.7124,
+      "step": 9064
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3555156266017383e-05,
+      "loss": 0.7129,
+      "step": 9065
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3553811067508586e-05,
+      "loss": 0.7856,
+      "step": 9066
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3552465795391551e-05,
+      "loss": 0.6968,
+      "step": 9067
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355112044969415e-05,
+      "loss": 0.8267,
+      "step": 9068
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3549775030444241e-05,
+      "loss": 0.792,
+      "step": 9069
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3548429537669696e-05,
+      "loss": 0.8813,
+      "step": 9070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.354708397139838e-05,
+      "loss": 0.7769,
+      "step": 9071
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3545738331658167e-05,
+      "loss": 0.7051,
+      "step": 9072
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3544392618476924e-05,
+      "loss": 0.7866,
+      "step": 9073
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3543046831882528e-05,
+      "loss": 0.7568,
+      "step": 9074
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3541700971902852e-05,
+      "loss": 0.7173,
+      "step": 9075
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3540355038565771e-05,
+      "loss": 0.6794,
+      "step": 9076
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3539009031899166e-05,
+      "loss": 0.8315,
+      "step": 9077
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3537662951930912e-05,
+      "loss": 0.8184,
+      "step": 9078
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.353631679868889e-05,
+      "loss": 0.7949,
+      "step": 9079
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.353497057220099e-05,
+      "loss": 0.7068,
+      "step": 9080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3533624272495084e-05,
+      "loss": 0.7959,
+      "step": 9081
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3532277899599064e-05,
+      "loss": 0.7222,
+      "step": 9082
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3530931453540814e-05,
+      "loss": 0.6807,
+      "step": 9083
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3529584934348225e-05,
+      "loss": 0.7483,
+      "step": 9084
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.352823834204918e-05,
+      "loss": 0.7319,
+      "step": 9085
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.352689167667158e-05,
+      "loss": 0.7622,
+      "step": 9086
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3525544938243313e-05,
+      "loss": 0.7126,
+      "step": 9087
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3524198126792273e-05,
+      "loss": 0.6682,
+      "step": 9088
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.352285124234636e-05,
+      "loss": 0.7661,
+      "step": 9089
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3521504284933464e-05,
+      "loss": 0.8389,
+      "step": 9090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3520157254581487e-05,
+      "loss": 0.7949,
+      "step": 9091
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.351881015131833e-05,
+      "loss": 0.7266,
+      "step": 9092
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3517462975171898e-05,
+      "loss": 0.6899,
+      "step": 9093
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3516115726170089e-05,
+      "loss": 0.7925,
+      "step": 9094
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.351476840434081e-05,
+      "loss": 0.7205,
+      "step": 9095
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3513421009711969e-05,
+      "loss": 0.7549,
+      "step": 9096
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3512073542311467e-05,
+      "loss": 0.79,
+      "step": 9097
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3510726002167222e-05,
+      "loss": 0.7383,
+      "step": 9098
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3509378389307142e-05,
+      "loss": 0.8267,
+      "step": 9099
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3508030703759136e-05,
+      "loss": 0.8149,
+      "step": 9100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3506682945551124e-05,
+      "loss": 0.8379,
+      "step": 9101
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3505335114711016e-05,
+      "loss": 0.8623,
+      "step": 9102
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3503987211266733e-05,
+      "loss": 0.7271,
+      "step": 9103
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3502639235246191e-05,
+      "loss": 0.7754,
+      "step": 9104
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.350129118667731e-05,
+      "loss": 0.8315,
+      "step": 9105
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3499943065588013e-05,
+      "loss": 0.8569,
+      "step": 9106
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.349859487200622e-05,
+      "loss": 0.729,
+      "step": 9107
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3497246605959857e-05,
+      "loss": 0.7583,
+      "step": 9108
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3495898267476852e-05,
+      "loss": 0.6382,
+      "step": 9109
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.349454985658513e-05,
+      "loss": 0.6792,
+      "step": 9110
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3493201373312618e-05,
+      "loss": 0.7642,
+      "step": 9111
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3491852817687253e-05,
+      "loss": 0.7649,
+      "step": 9112
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3490504189736959e-05,
+      "loss": 0.7881,
+      "step": 9113
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3489155489489676e-05,
+      "loss": 0.7554,
+      "step": 9114
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3487806716973335e-05,
+      "loss": 0.6812,
+      "step": 9115
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3486457872215875e-05,
+      "loss": 0.7979,
+      "step": 9116
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3485108955245232e-05,
+      "loss": 0.8247,
+      "step": 9117
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3483759966089344e-05,
+      "loss": 0.751,
+      "step": 9118
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3482410904776155e-05,
+      "loss": 0.7131,
+      "step": 9119
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3481061771333608e-05,
+      "loss": 0.8525,
+      "step": 9120
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3479712565789644e-05,
+      "loss": 0.8286,
+      "step": 9121
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.347836328817221e-05,
+      "loss": 0.7891,
+      "step": 9122
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3477013938509252e-05,
+      "loss": 0.7498,
+      "step": 9123
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3475664516828718e-05,
+      "loss": 0.6841,
+      "step": 9124
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.347431502315856e-05,
+      "loss": 0.7446,
+      "step": 9125
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3472965457526727e-05,
+      "loss": 0.6919,
+      "step": 9126
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3471615819961174e-05,
+      "loss": 0.7354,
+      "step": 9127
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3470266110489853e-05,
+      "loss": 0.7061,
+      "step": 9128
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3468916329140723e-05,
+      "loss": 0.7031,
+      "step": 9129
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3467566475941738e-05,
+      "loss": 0.7512,
+      "step": 9130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3466216550920858e-05,
+      "loss": 0.7744,
+      "step": 9131
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3464866554106044e-05,
+      "loss": 0.7734,
+      "step": 9132
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3463516485525258e-05,
+      "loss": 0.6904,
+      "step": 9133
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3462166345206462e-05,
+      "loss": 0.7617,
+      "step": 9134
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3460816133177622e-05,
+      "loss": 0.7417,
+      "step": 9135
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.34594658494667e-05,
+      "loss": 0.8325,
+      "step": 9136
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.345811549410167e-05,
+      "loss": 0.7644,
+      "step": 9137
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3456765067110497e-05,
+      "loss": 0.8472,
+      "step": 9138
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3455414568521157e-05,
+      "loss": 0.8008,
+      "step": 9139
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3454063998361615e-05,
+      "loss": 0.7129,
+      "step": 9140
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3452713356659848e-05,
+      "loss": 0.7168,
+      "step": 9141
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3451362643443832e-05,
+      "loss": 0.7681,
+      "step": 9142
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3450011858741543e-05,
+      "loss": 0.7764,
+      "step": 9143
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3448661002580959e-05,
+      "loss": 0.8774,
+      "step": 9144
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.344731007499006e-05,
+      "loss": 0.7954,
+      "step": 9145
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3445959075996822e-05,
+      "loss": 0.8613,
+      "step": 9146
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3444608005629235e-05,
+      "loss": 0.7568,
+      "step": 9147
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3443256863915281e-05,
+      "loss": 0.6626,
+      "step": 9148
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3441905650882942e-05,
+      "loss": 0.7622,
+      "step": 9149
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3440554366560208e-05,
+      "loss": 0.8037,
+      "step": 9150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3439203010975068e-05,
+      "loss": 0.7979,
+      "step": 9151
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3437851584155513e-05,
+      "loss": 0.7498,
+      "step": 9152
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.343650008612953e-05,
+      "loss": 0.6926,
+      "step": 9153
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3435148516925116e-05,
+      "loss": 0.8652,
+      "step": 9154
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3433796876570263e-05,
+      "loss": 0.8359,
+      "step": 9155
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3432445165092964e-05,
+      "loss": 0.8047,
+      "step": 9156
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.343109338252122e-05,
+      "loss": 0.7393,
+      "step": 9157
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3429741528883034e-05,
+      "loss": 0.7881,
+      "step": 9158
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3428389604206397e-05,
+      "loss": 0.793,
+      "step": 9159
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3427037608519318e-05,
+      "loss": 0.6973,
+      "step": 9160
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3425685541849795e-05,
+      "loss": 0.7383,
+      "step": 9161
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3424333404225833e-05,
+      "loss": 0.7158,
+      "step": 9162
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3422981195675444e-05,
+      "loss": 0.8062,
+      "step": 9163
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3421628916226632e-05,
+      "loss": 0.6633,
+      "step": 9164
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3420276565907407e-05,
+      "loss": 0.7024,
+      "step": 9165
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3418924144745776e-05,
+      "loss": 0.7749,
+      "step": 9166
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3417571652769755e-05,
+      "loss": 0.7417,
+      "step": 9167
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3416219090007355e-05,
+      "loss": 0.7964,
+      "step": 9168
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3414866456486592e-05,
+      "loss": 0.751,
+      "step": 9169
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3413513752235483e-05,
+      "loss": 0.8018,
+      "step": 9170
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3412160977282044e-05,
+      "loss": 0.7783,
+      "step": 9171
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3410808131654296e-05,
+      "loss": 0.8721,
+      "step": 9172
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3409455215380258e-05,
+      "loss": 0.7319,
+      "step": 9173
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3408102228487954e-05,
+      "loss": 0.7046,
+      "step": 9174
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3406749171005407e-05,
+      "loss": 0.8184,
+      "step": 9175
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3405396042960642e-05,
+      "loss": 0.7422,
+      "step": 9176
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3404042844381686e-05,
+      "loss": 0.8574,
+      "step": 9177
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3402689575296568e-05,
+      "loss": 0.7671,
+      "step": 9178
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3401336235733315e-05,
+      "loss": 0.7329,
+      "step": 9179
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3399982825719961e-05,
+      "loss": 0.8398,
+      "step": 9180
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3398629345284538e-05,
+      "loss": 0.6311,
+      "step": 9181
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3397275794455076e-05,
+      "loss": 0.7344,
+      "step": 9182
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3395922173259614e-05,
+      "loss": 0.811,
+      "step": 9183
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3394568481726193e-05,
+      "loss": 0.7529,
+      "step": 9184
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3393214719882843e-05,
+      "loss": 0.7515,
+      "step": 9185
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3391860887757609e-05,
+      "loss": 0.7476,
+      "step": 9186
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3390506985378525e-05,
+      "loss": 0.7588,
+      "step": 9187
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3389153012773642e-05,
+      "loss": 0.7847,
+      "step": 9188
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3387798969971005e-05,
+      "loss": 0.7593,
+      "step": 9189
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.338644485699865e-05,
+      "loss": 0.7,
+      "step": 9190
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3385090673884635e-05,
+      "loss": 0.8013,
+      "step": 9191
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3383736420657001e-05,
+      "loss": 0.7319,
+      "step": 9192
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3382382097343803e-05,
+      "loss": 0.6787,
+      "step": 9193
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3381027703973086e-05,
+      "loss": 0.8438,
+      "step": 9194
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3379673240572908e-05,
+      "loss": 0.7983,
+      "step": 9195
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3378318707171322e-05,
+      "loss": 0.8521,
+      "step": 9196
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3376964103796384e-05,
+      "loss": 0.8213,
+      "step": 9197
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3375609430476151e-05,
+      "loss": 0.7456,
+      "step": 9198
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3374254687238682e-05,
+      "loss": 0.748,
+      "step": 9199
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3372899874112036e-05,
+      "loss": 0.729,
+      "step": 9200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3371544991124274e-05,
+      "loss": 0.7124,
+      "step": 9201
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.337019003830346e-05,
+      "loss": 0.7925,
+      "step": 9202
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3368835015677662e-05,
+      "loss": 0.8286,
+      "step": 9203
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.336747992327494e-05,
+      "loss": 0.7197,
+      "step": 9204
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3366124761123362e-05,
+      "loss": 0.6968,
+      "step": 9205
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3364769529251e-05,
+      "loss": 0.8057,
+      "step": 9206
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3363414227685925e-05,
+      "loss": 0.8384,
+      "step": 9207
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3362058856456205e-05,
+      "loss": 0.7764,
+      "step": 9208
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3360703415589913e-05,
+      "loss": 0.7637,
+      "step": 9209
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3359347905115126e-05,
+      "loss": 0.6775,
+      "step": 9210
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3357992325059918e-05,
+      "loss": 0.7046,
+      "step": 9211
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3356636675452368e-05,
+      "loss": 0.8608,
+      "step": 9212
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3355280956320553e-05,
+      "loss": 0.7705,
+      "step": 9213
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3353925167692557e-05,
+      "loss": 0.8149,
+      "step": 9214
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.335256930959646e-05,
+      "loss": 0.7905,
+      "step": 9215
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3351213382060342e-05,
+      "loss": 0.8052,
+      "step": 9216
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3349857385112291e-05,
+      "loss": 0.8428,
+      "step": 9217
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3348501318780391e-05,
+      "loss": 0.7471,
+      "step": 9218
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3347145183092734e-05,
+      "loss": 0.8066,
+      "step": 9219
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.33457889780774e-05,
+      "loss": 0.7668,
+      "step": 9220
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3344432703762489e-05,
+      "loss": 0.8184,
+      "step": 9221
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3343076360176089e-05,
+      "loss": 0.7627,
+      "step": 9222
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3341719947346292e-05,
+      "loss": 0.8525,
+      "step": 9223
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3340363465301194e-05,
+      "loss": 0.7007,
+      "step": 9224
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.333900691406889e-05,
+      "loss": 0.6255,
+      "step": 9225
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.333765029367748e-05,
+      "loss": 0.8252,
+      "step": 9226
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3336293604155058e-05,
+      "loss": 0.7939,
+      "step": 9227
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3334936845529732e-05,
+      "loss": 0.7388,
+      "step": 9228
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3333580017829597e-05,
+      "loss": 0.8628,
+      "step": 9229
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3332223121082759e-05,
+      "loss": 0.6902,
+      "step": 9230
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3330866155317321e-05,
+      "loss": 0.8125,
+      "step": 9231
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3329509120561392e-05,
+      "loss": 0.7539,
+      "step": 9232
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3328152016843078e-05,
+      "loss": 0.7393,
+      "step": 9233
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3326794844190489e-05,
+      "loss": 0.812,
+      "step": 9234
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3325437602631735e-05,
+      "loss": 0.7427,
+      "step": 9235
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3324080292194925e-05,
+      "loss": 0.8096,
+      "step": 9236
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3322722912908174e-05,
+      "loss": 0.7871,
+      "step": 9237
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3321365464799603e-05,
+      "loss": 0.6992,
+      "step": 9238
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3320007947897316e-05,
+      "loss": 0.728,
+      "step": 9239
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.331865036222944e-05,
+      "loss": 0.7935,
+      "step": 9240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.331729270782409e-05,
+      "loss": 0.7979,
+      "step": 9241
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3315934984709386e-05,
+      "loss": 0.7412,
+      "step": 9242
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3314577192913456e-05,
+      "loss": 0.7515,
+      "step": 9243
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3313219332464415e-05,
+      "loss": 0.7527,
+      "step": 9244
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3311861403390394e-05,
+      "loss": 0.7368,
+      "step": 9245
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3310503405719512e-05,
+      "loss": 0.8149,
+      "step": 9246
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3309145339479903e-05,
+      "loss": 0.7349,
+      "step": 9247
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3307787204699692e-05,
+      "loss": 0.7612,
+      "step": 9248
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3306429001407011e-05,
+      "loss": 0.7224,
+      "step": 9249
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3305070729629992e-05,
+      "loss": 0.8003,
+      "step": 9250
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3303712389396766e-05,
+      "loss": 0.7576,
+      "step": 9251
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3302353980735471e-05,
+      "loss": 0.7432,
+      "step": 9252
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3300995503674244e-05,
+      "loss": 0.7598,
+      "step": 9253
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3299636958241215e-05,
+      "loss": 0.7725,
+      "step": 9254
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3298278344464533e-05,
+      "loss": 0.8228,
+      "step": 9255
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.329691966237233e-05,
+      "loss": 0.7217,
+      "step": 9256
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3295560911992748e-05,
+      "loss": 0.8755,
+      "step": 9257
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3294202093353938e-05,
+      "loss": 0.7046,
+      "step": 9258
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3292843206484039e-05,
+      "loss": 0.8635,
+      "step": 9259
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3291484251411192e-05,
+      "loss": 0.7012,
+      "step": 9260
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3290125228163552e-05,
+      "loss": 0.7568,
+      "step": 9261
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3288766136769267e-05,
+      "loss": 0.709,
+      "step": 9262
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3287406977256483e-05,
+      "loss": 0.6321,
+      "step": 9263
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3286047749653352e-05,
+      "loss": 0.7285,
+      "step": 9264
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.328468845398803e-05,
+      "loss": 0.7104,
+      "step": 9265
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3283329090288672e-05,
+      "loss": 0.8149,
+      "step": 9266
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3281969658583434e-05,
+      "loss": 0.8091,
+      "step": 9267
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3280610158900467e-05,
+      "loss": 0.8638,
+      "step": 9268
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3279250591267935e-05,
+      "loss": 0.7354,
+      "step": 9269
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3277890955714e-05,
+      "loss": 0.7891,
+      "step": 9270
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3276531252266816e-05,
+      "loss": 0.8242,
+      "step": 9271
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3275171480954552e-05,
+      "loss": 0.7344,
+      "step": 9272
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3273811641805372e-05,
+      "loss": 0.7424,
+      "step": 9273
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3272451734847438e-05,
+      "loss": 0.7734,
+      "step": 9274
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3271091760108918e-05,
+      "loss": 0.7458,
+      "step": 9275
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3269731717617982e-05,
+      "loss": 0.7524,
+      "step": 9276
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3268371607402799e-05,
+      "loss": 0.7622,
+      "step": 9277
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3267011429491541e-05,
+      "loss": 0.7207,
+      "step": 9278
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3265651183912383e-05,
+      "loss": 0.7915,
+      "step": 9279
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3264290870693493e-05,
+      "loss": 0.8726,
+      "step": 9280
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.326293048986305e-05,
+      "loss": 0.7217,
+      "step": 9281
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.326157004144923e-05,
+      "loss": 0.7764,
+      "step": 9282
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3260209525480215e-05,
+      "loss": 0.8267,
+      "step": 9283
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3258848941984179e-05,
+      "loss": 0.6987,
+      "step": 9284
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3257488290989304e-05,
+      "loss": 0.7437,
+      "step": 9285
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3256127572523778e-05,
+      "loss": 0.7612,
+      "step": 9286
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3254766786615779e-05,
+      "loss": 0.7344,
+      "step": 9287
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3253405933293493e-05,
+      "loss": 0.8018,
+      "step": 9288
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3252045012585108e-05,
+      "loss": 0.7827,
+      "step": 9289
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3250684024518813e-05,
+      "loss": 0.8911,
+      "step": 9290
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3249322969122796e-05,
+      "loss": 0.8389,
+      "step": 9291
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3247961846425248e-05,
+      "loss": 0.6736,
+      "step": 9292
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.324660065645436e-05,
+      "loss": 0.7109,
+      "step": 9293
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3245239399238327e-05,
+      "loss": 0.7524,
+      "step": 9294
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3243878074805347e-05,
+      "loss": 0.7344,
+      "step": 9295
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.324251668318361e-05,
+      "loss": 0.6948,
+      "step": 9296
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3241155224401318e-05,
+      "loss": 0.8477,
+      "step": 9297
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3239793698486671e-05,
+      "loss": 0.7505,
+      "step": 9298
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3238432105467866e-05,
+      "loss": 0.812,
+      "step": 9299
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3237070445373109e-05,
+      "loss": 0.6992,
+      "step": 9300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3235708718230598e-05,
+      "loss": 0.7461,
+      "step": 9301
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3234346924068544e-05,
+      "loss": 0.8867,
+      "step": 9302
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3232985062915147e-05,
+      "loss": 0.7925,
+      "step": 9303
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3231623134798619e-05,
+      "loss": 0.772,
+      "step": 9304
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.323026113974717e-05,
+      "loss": 0.7205,
+      "step": 9305
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3228899077789004e-05,
+      "loss": 0.8057,
+      "step": 9306
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3227536948952338e-05,
+      "loss": 0.7822,
+      "step": 9307
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3226174753265381e-05,
+      "loss": 0.7295,
+      "step": 9308
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3224812490756353e-05,
+      "loss": 0.7412,
+      "step": 9309
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3223450161453468e-05,
+      "loss": 0.7241,
+      "step": 9310
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3222087765384939e-05,
+      "loss": 0.7266,
+      "step": 9311
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.322072530257899e-05,
+      "loss": 0.8428,
+      "step": 9312
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3219362773063837e-05,
+      "loss": 0.752,
+      "step": 9313
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3218000176867702e-05,
+      "loss": 0.812,
+      "step": 9314
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3216637514018811e-05,
+      "loss": 0.7563,
+      "step": 9315
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3215274784545383e-05,
+      "loss": 0.7629,
+      "step": 9316
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.321391198847565e-05,
+      "loss": 0.7881,
+      "step": 9317
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3212549125837832e-05,
+      "loss": 0.8325,
+      "step": 9318
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3211186196660162e-05,
+      "loss": 0.697,
+      "step": 9319
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3209823200970869e-05,
+      "loss": 0.7427,
+      "step": 9320
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3208460138798183e-05,
+      "loss": 0.8281,
+      "step": 9321
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3207097010170336e-05,
+      "loss": 0.7778,
+      "step": 9322
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.320573381511556e-05,
+      "loss": 0.7744,
+      "step": 9323
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3204370553662096e-05,
+      "loss": 0.752,
+      "step": 9324
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3203007225838174e-05,
+      "loss": 0.8384,
+      "step": 9325
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3201643831672037e-05,
+      "loss": 0.8291,
+      "step": 9326
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.320028037119192e-05,
+      "loss": 0.6758,
+      "step": 9327
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3198916844426067e-05,
+      "loss": 0.7944,
+      "step": 9328
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.319755325140272e-05,
+      "loss": 0.7744,
+      "step": 9329
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3196189592150119e-05,
+      "loss": 0.8179,
+      "step": 9330
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3194825866696512e-05,
+      "loss": 0.8286,
+      "step": 9331
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3193462075070148e-05,
+      "loss": 0.8267,
+      "step": 9332
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3192098217299265e-05,
+      "loss": 0.7402,
+      "step": 9333
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.319073429341212e-05,
+      "loss": 0.7651,
+      "step": 9334
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3189370303436962e-05,
+      "loss": 0.7727,
+      "step": 9335
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3188006247402038e-05,
+      "loss": 0.6042,
+      "step": 9336
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3186642125335607e-05,
+      "loss": 0.7422,
+      "step": 9337
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.318527793726592e-05,
+      "loss": 0.7114,
+      "step": 9338
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3183913683221232e-05,
+      "loss": 0.752,
+      "step": 9339
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3182549363229802e-05,
+      "loss": 0.8369,
+      "step": 9340
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3181184977319886e-05,
+      "loss": 0.751,
+      "step": 9341
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3179820525519747e-05,
+      "loss": 0.8325,
+      "step": 9342
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3178456007857646e-05,
+      "loss": 0.7275,
+      "step": 9343
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3177091424361845e-05,
+      "loss": 0.7429,
+      "step": 9344
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3175726775060606e-05,
+      "loss": 0.835,
+      "step": 9345
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3174362059982198e-05,
+      "loss": 0.7939,
+      "step": 9346
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3172997279154884e-05,
+      "loss": 0.7964,
+      "step": 9347
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3171632432606933e-05,
+      "loss": 0.8052,
+      "step": 9348
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3170267520366613e-05,
+      "loss": 0.6846,
+      "step": 9349
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3168902542462199e-05,
+      "loss": 0.8071,
+      "step": 9350
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3167537498921959e-05,
+      "loss": 0.6838,
+      "step": 9351
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3166172389774169e-05,
+      "loss": 0.6909,
+      "step": 9352
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3164807215047097e-05,
+      "loss": 0.8228,
+      "step": 9353
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3163441974769031e-05,
+      "loss": 0.8091,
+      "step": 9354
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.316207666896824e-05,
+      "loss": 0.7537,
+      "step": 9355
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3160711297673005e-05,
+      "loss": 0.696,
+      "step": 9356
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3159345860911607e-05,
+      "loss": 0.6912,
+      "step": 9357
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3157980358712328e-05,
+      "loss": 0.8315,
+      "step": 9358
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3156614791103448e-05,
+      "loss": 0.7114,
+      "step": 9359
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3155249158113255e-05,
+      "loss": 0.8091,
+      "step": 9360
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3153883459770032e-05,
+      "loss": 0.7295,
+      "step": 9361
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3152517696102067e-05,
+      "loss": 0.8633,
+      "step": 9362
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.315115186713765e-05,
+      "loss": 0.7886,
+      "step": 9363
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3149785972905065e-05,
+      "loss": 0.8062,
+      "step": 9364
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3148420013432608e-05,
+      "loss": 0.729,
+      "step": 9365
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.314705398874857e-05,
+      "loss": 0.6875,
+      "step": 9366
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3145687898881245e-05,
+      "loss": 0.8159,
+      "step": 9367
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.314432174385893e-05,
+      "loss": 0.7505,
+      "step": 9368
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3142955523709919e-05,
+      "loss": 0.8569,
+      "step": 9369
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.314158923846251e-05,
+      "loss": 0.731,
+      "step": 9370
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3140222888145004e-05,
+      "loss": 0.708,
+      "step": 9371
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3138856472785699e-05,
+      "loss": 0.7583,
+      "step": 9372
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3137489992412897e-05,
+      "loss": 0.8176,
+      "step": 9373
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3136123447054903e-05,
+      "loss": 0.8188,
+      "step": 9374
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3134756836740021e-05,
+      "loss": 0.7407,
+      "step": 9375
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3133390161496555e-05,
+      "loss": 0.7046,
+      "step": 9376
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3132023421352818e-05,
+      "loss": 0.7437,
+      "step": 9377
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.313065661633711e-05,
+      "loss": 0.7368,
+      "step": 9378
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3129289746477747e-05,
+      "loss": 0.811,
+      "step": 9379
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3127922811803038e-05,
+      "loss": 0.8135,
+      "step": 9380
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3126555812341298e-05,
+      "loss": 0.8242,
+      "step": 9381
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3125188748120836e-05,
+      "loss": 0.749,
+      "step": 9382
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3123821619169972e-05,
+      "loss": 0.7744,
+      "step": 9383
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3122454425517019e-05,
+      "loss": 0.7886,
+      "step": 9384
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.31210871671903e-05,
+      "loss": 0.6714,
+      "step": 9385
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.311971984421813e-05,
+      "loss": 0.751,
+      "step": 9386
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3118352456628833e-05,
+      "loss": 0.7344,
+      "step": 9387
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3116985004450727e-05,
+      "loss": 0.6548,
+      "step": 9388
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3115617487712139e-05,
+      "loss": 0.8101,
+      "step": 9389
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3114249906441388e-05,
+      "loss": 0.7705,
+      "step": 9390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3112882260666805e-05,
+      "loss": 0.7549,
+      "step": 9391
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.311151455041672e-05,
+      "loss": 0.7681,
+      "step": 9392
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3110146775719454e-05,
+      "loss": 0.7383,
+      "step": 9393
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3108778936603343e-05,
+      "loss": 0.7642,
+      "step": 9394
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3107411033096716e-05,
+      "loss": 0.7554,
+      "step": 9395
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3106043065227905e-05,
+      "loss": 0.7881,
+      "step": 9396
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3104675033025245e-05,
+      "loss": 0.834,
+      "step": 9397
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3103306936517072e-05,
+      "loss": 0.689,
+      "step": 9398
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3101938775731723e-05,
+      "loss": 0.6667,
+      "step": 9399
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3100570550697533e-05,
+      "loss": 0.6514,
+      "step": 9400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3099202261442842e-05,
+      "loss": 0.8535,
+      "step": 9401
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3097833907995996e-05,
+      "loss": 0.7185,
+      "step": 9402
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.309646549038533e-05,
+      "loss": 0.7354,
+      "step": 9403
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.309509700863919e-05,
+      "loss": 0.8726,
+      "step": 9404
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.309372846278592e-05,
+      "loss": 0.749,
+      "step": 9405
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.309235985285387e-05,
+      "loss": 0.7583,
+      "step": 9406
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3090991178871384e-05,
+      "loss": 0.7837,
+      "step": 9407
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.308962244086681e-05,
+      "loss": 0.7451,
+      "step": 9408
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3088253638868498e-05,
+      "loss": 0.7524,
+      "step": 9409
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3086884772904803e-05,
+      "loss": 0.8032,
+      "step": 9410
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.308551584300407e-05,
+      "loss": 0.7725,
+      "step": 9411
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3084146849194662e-05,
+      "loss": 0.7422,
+      "step": 9412
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3082777791504926e-05,
+      "loss": 0.7827,
+      "step": 9413
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3081408669963222e-05,
+      "loss": 0.6904,
+      "step": 9414
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3080039484597914e-05,
+      "loss": 0.8135,
+      "step": 9415
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307867023543735e-05,
+      "loss": 0.771,
+      "step": 9416
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.30773009225099e-05,
+      "loss": 0.7583,
+      "step": 9417
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3075931545843916e-05,
+      "loss": 0.8462,
+      "step": 9418
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3074562105467772e-05,
+      "loss": 0.8511,
+      "step": 9419
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3073192601409825e-05,
+      "loss": 0.7251,
+      "step": 9420
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3071823033698447e-05,
+      "loss": 0.7271,
+      "step": 9421
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3070453402362e-05,
+      "loss": 0.7847,
+      "step": 9422
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3069083707428853e-05,
+      "loss": 0.7788,
+      "step": 9423
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3067713948927378e-05,
+      "loss": 0.8076,
+      "step": 9424
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3066344126885943e-05,
+      "loss": 0.7939,
+      "step": 9425
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3064974241332924e-05,
+      "loss": 0.7998,
+      "step": 9426
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3063604292296695e-05,
+      "loss": 0.7949,
+      "step": 9427
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3062234279805626e-05,
+      "loss": 0.8516,
+      "step": 9428
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3060864203888097e-05,
+      "loss": 0.8472,
+      "step": 9429
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3059494064572484e-05,
+      "loss": 0.6543,
+      "step": 9430
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3058123861887168e-05,
+      "loss": 0.7822,
+      "step": 9431
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3056753595860529e-05,
+      "loss": 0.6985,
+      "step": 9432
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3055383266520947e-05,
+      "loss": 0.7261,
+      "step": 9433
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3054012873896809e-05,
+      "loss": 0.8599,
+      "step": 9434
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3052642418016493e-05,
+      "loss": 0.7754,
+      "step": 9435
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3051271898908391e-05,
+      "loss": 0.7588,
+      "step": 9436
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3049901316600883e-05,
+      "loss": 0.7334,
+      "step": 9437
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3048530671122362e-05,
+      "loss": 0.7632,
+      "step": 9438
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3047159962501216e-05,
+      "loss": 0.75,
+      "step": 9439
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3045789190765836e-05,
+      "loss": 0.7246,
+      "step": 9440
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3044418355944614e-05,
+      "loss": 0.7617,
+      "step": 9441
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.304304745806594e-05,
+      "loss": 0.8911,
+      "step": 9442
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3041676497158216e-05,
+      "loss": 0.7607,
+      "step": 9443
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3040305473249832e-05,
+      "loss": 0.7871,
+      "step": 9444
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3038934386369191e-05,
+      "loss": 0.7544,
+      "step": 9445
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3037563236544687e-05,
+      "loss": 0.6875,
+      "step": 9446
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3036192023804716e-05,
+      "loss": 0.7214,
+      "step": 9447
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.303482074817769e-05,
+      "loss": 0.7612,
+      "step": 9448
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3033449409692003e-05,
+      "loss": 0.6743,
+      "step": 9449
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3032078008376061e-05,
+      "loss": 0.8589,
+      "step": 9450
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3030706544258274e-05,
+      "loss": 0.6919,
+      "step": 9451
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.302933501736704e-05,
+      "loss": 0.7095,
+      "step": 9452
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.302796342773077e-05,
+      "loss": 0.769,
+      "step": 9453
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3026591775377875e-05,
+      "loss": 0.7754,
+      "step": 9454
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.302522006033676e-05,
+      "loss": 0.7622,
+      "step": 9455
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3023848282635845e-05,
+      "loss": 0.7764,
+      "step": 9456
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.302247644230354e-05,
+      "loss": 0.7595,
+      "step": 9457
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3021104539368254e-05,
+      "loss": 0.8413,
+      "step": 9458
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3019732573858407e-05,
+      "loss": 0.7607,
+      "step": 9459
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3018360545802418e-05,
+      "loss": 0.7266,
+      "step": 9460
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.30169884552287e-05,
+      "loss": 0.6624,
+      "step": 9461
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3015616302165675e-05,
+      "loss": 0.8086,
+      "step": 9462
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3014244086641764e-05,
+      "loss": 0.7153,
+      "step": 9463
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3012871808685388e-05,
+      "loss": 0.8213,
+      "step": 9464
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3011499468324968e-05,
+      "loss": 0.7451,
+      "step": 9465
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3010127065588936e-05,
+      "loss": 0.7866,
+      "step": 9466
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3008754600505708e-05,
+      "loss": 0.7656,
+      "step": 9467
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3007382073103718e-05,
+      "loss": 0.8481,
+      "step": 9468
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3006009483411393e-05,
+      "loss": 0.8325,
+      "step": 9469
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3004636831457163e-05,
+      "loss": 0.8208,
+      "step": 9470
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3003264117269456e-05,
+      "loss": 0.8413,
+      "step": 9471
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3001891340876709e-05,
+      "loss": 0.7427,
+      "step": 9472
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3000518502307351e-05,
+      "loss": 0.7136,
+      "step": 9473
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2999145601589822e-05,
+      "loss": 0.8379,
+      "step": 9474
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2997772638752553e-05,
+      "loss": 0.7402,
+      "step": 9475
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2996399613823984e-05,
+      "loss": 0.7695,
+      "step": 9476
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2995026526832555e-05,
+      "loss": 0.7061,
+      "step": 9477
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2993653377806705e-05,
+      "loss": 0.7476,
+      "step": 9478
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2992280166774873e-05,
+      "loss": 0.7373,
+      "step": 9479
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2990906893765503e-05,
+      "loss": 0.7397,
+      "step": 9480
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2989533558807041e-05,
+      "loss": 0.9326,
+      "step": 9481
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.298816016192793e-05,
+      "loss": 0.7881,
+      "step": 9482
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2986786703156618e-05,
+      "loss": 0.7559,
+      "step": 9483
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.298541318252155e-05,
+      "loss": 0.7388,
+      "step": 9484
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2984039600051177e-05,
+      "loss": 0.7437,
+      "step": 9485
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.298266595577395e-05,
+      "loss": 0.7637,
+      "step": 9486
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2981292249718319e-05,
+      "loss": 0.8975,
+      "step": 9487
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2979918481912736e-05,
+      "loss": 0.7937,
+      "step": 9488
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.297854465238566e-05,
+      "loss": 0.7378,
+      "step": 9489
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2977170761165541e-05,
+      "loss": 0.7791,
+      "step": 9490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2975796808280837e-05,
+      "loss": 0.7275,
+      "step": 9491
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2974422793760009e-05,
+      "loss": 0.7969,
+      "step": 9492
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2973048717631511e-05,
+      "loss": 0.7051,
+      "step": 9493
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2971674579923808e-05,
+      "loss": 0.8027,
+      "step": 9494
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2970300380665356e-05,
+      "loss": 0.7334,
+      "step": 9495
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.296892611988463e-05,
+      "loss": 0.7559,
+      "step": 9496
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2967551797610084e-05,
+      "loss": 0.7734,
+      "step": 9497
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2966177413870185e-05,
+      "loss": 0.7971,
+      "step": 9498
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2964802968693401e-05,
+      "loss": 0.7847,
+      "step": 9499
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2963428462108203e-05,
+      "loss": 0.7676,
+      "step": 9500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2962053894143053e-05,
+      "loss": 0.8462,
+      "step": 9501
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.296067926482643e-05,
+      "loss": 0.73,
+      "step": 9502
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2959304574186802e-05,
+      "loss": 0.7622,
+      "step": 9503
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2957929822252644e-05,
+      "loss": 0.731,
+      "step": 9504
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2956555009052427e-05,
+      "loss": 0.7964,
+      "step": 9505
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2955180134614631e-05,
+      "loss": 0.8062,
+      "step": 9506
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2953805198967726e-05,
+      "loss": 0.729,
+      "step": 9507
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.29524302021402e-05,
+      "loss": 0.7168,
+      "step": 9508
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2951055144160527e-05,
+      "loss": 0.8101,
+      "step": 9509
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2949680025057189e-05,
+      "loss": 0.8584,
+      "step": 9510
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.294830484485867e-05,
+      "loss": 0.7458,
+      "step": 9511
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.294692960359345e-05,
+      "loss": 0.8208,
+      "step": 9512
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2945554301290013e-05,
+      "loss": 0.8159,
+      "step": 9513
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.294417893797685e-05,
+      "loss": 0.8013,
+      "step": 9514
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2942803513682442e-05,
+      "loss": 0.6917,
+      "step": 9515
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2941428028435283e-05,
+      "loss": 0.7104,
+      "step": 9516
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.294005248226386e-05,
+      "loss": 0.748,
+      "step": 9517
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2938676875196664e-05,
+      "loss": 0.8535,
+      "step": 9518
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2937301207262184e-05,
+      "loss": 0.7803,
+      "step": 9519
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.293592547848892e-05,
+      "loss": 0.7773,
+      "step": 9520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2934549688905364e-05,
+      "loss": 0.728,
+      "step": 9521
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2933173838540012e-05,
+      "loss": 0.7939,
+      "step": 9522
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2931797927421362e-05,
+      "loss": 0.8032,
+      "step": 9523
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2930421955577907e-05,
+      "loss": 0.8901,
+      "step": 9524
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.292904592303816e-05,
+      "loss": 0.7358,
+      "step": 9525
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2927669829830608e-05,
+      "loss": 0.7446,
+      "step": 9526
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2926293675983757e-05,
+      "loss": 0.7139,
+      "step": 9527
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2924917461526114e-05,
+      "loss": 0.6545,
+      "step": 9528
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2923541186486182e-05,
+      "loss": 0.686,
+      "step": 9529
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2922164850892464e-05,
+      "loss": 0.7446,
+      "step": 9530
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2920788454773474e-05,
+      "loss": 0.7939,
+      "step": 9531
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2919411998157717e-05,
+      "loss": 0.7812,
+      "step": 9532
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.29180354810737e-05,
+      "loss": 0.7666,
+      "step": 9533
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.291665890354994e-05,
+      "loss": 0.8018,
+      "step": 9534
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2915282265614945e-05,
+      "loss": 0.7576,
+      "step": 9535
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2913905567297232e-05,
+      "loss": 0.7642,
+      "step": 9536
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2912528808625311e-05,
+      "loss": 0.8154,
+      "step": 9537
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.29111519896277e-05,
+      "loss": 0.7537,
+      "step": 9538
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2909775110332919e-05,
+      "loss": 0.6963,
+      "step": 9539
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2908398170769485e-05,
+      "loss": 0.8232,
+      "step": 9540
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2907021170965919e-05,
+      "loss": 0.6987,
+      "step": 9541
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2905644110950738e-05,
+      "loss": 0.7781,
+      "step": 9542
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2904266990752468e-05,
+      "loss": 0.6455,
+      "step": 9543
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2902889810399626e-05,
+      "loss": 0.7578,
+      "step": 9544
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2901512569920746e-05,
+      "loss": 0.7944,
+      "step": 9545
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2900135269344354e-05,
+      "loss": 0.7815,
+      "step": 9546
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2898757908698972e-05,
+      "loss": 0.7627,
+      "step": 9547
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2897380488013126e-05,
+      "loss": 0.8701,
+      "step": 9548
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2896003007315354e-05,
+      "loss": 0.7175,
+      "step": 9549
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2894625466634183e-05,
+      "loss": 0.8462,
+      "step": 9550
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2893247865998144e-05,
+      "loss": 0.7251,
+      "step": 9551
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2891870205435772e-05,
+      "loss": 0.7109,
+      "step": 9552
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2890492484975603e-05,
+      "loss": 0.8379,
+      "step": 9553
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.288911470464617e-05,
+      "loss": 0.7485,
+      "step": 9554
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2887736864476014e-05,
+      "loss": 0.832,
+      "step": 9555
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2886358964493668e-05,
+      "loss": 0.8394,
+      "step": 9556
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2884981004727676e-05,
+      "loss": 0.7959,
+      "step": 9557
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2883602985206581e-05,
+      "loss": 0.7031,
+      "step": 9558
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.288222490595892e-05,
+      "loss": 0.7417,
+      "step": 9559
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2880846767013238e-05,
+      "loss": 0.8218,
+      "step": 9560
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2879468568398081e-05,
+      "loss": 0.8364,
+      "step": 9561
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2878090310141995e-05,
+      "loss": 0.8081,
+      "step": 9562
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2876711992273526e-05,
+      "loss": 0.7129,
+      "step": 9563
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2875333614821222e-05,
+      "loss": 0.7166,
+      "step": 9564
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2873955177813633e-05,
+      "loss": 0.8262,
+      "step": 9565
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2872576681279311e-05,
+      "loss": 0.7419,
+      "step": 9566
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2871198125246809e-05,
+      "loss": 0.7412,
+      "step": 9567
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2869819509744677e-05,
+      "loss": 0.7549,
+      "step": 9568
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.286844083480147e-05,
+      "loss": 0.7974,
+      "step": 9569
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2867062100445746e-05,
+      "loss": 0.8188,
+      "step": 9570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2865683306706061e-05,
+      "loss": 0.8315,
+      "step": 9571
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2864304453610974e-05,
+      "loss": 0.6689,
+      "step": 9572
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2862925541189042e-05,
+      "loss": 0.7217,
+      "step": 9573
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.286154656946883e-05,
+      "loss": 0.6887,
+      "step": 9574
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2860167538478895e-05,
+      "loss": 0.835,
+      "step": 9575
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2858788448247803e-05,
+      "loss": 0.7476,
+      "step": 9576
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2857409298804118e-05,
+      "loss": 0.7905,
+      "step": 9577
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2856030090176404e-05,
+      "loss": 0.698,
+      "step": 9578
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.285465082239323e-05,
+      "loss": 0.7581,
+      "step": 9579
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2853271495483164e-05,
+      "loss": 0.6899,
+      "step": 9580
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2851892109474773e-05,
+      "loss": 0.7729,
+      "step": 9581
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2850512664396631e-05,
+      "loss": 0.7197,
+      "step": 9582
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2849133160277308e-05,
+      "loss": 0.6702,
+      "step": 9583
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2847753597145372e-05,
+      "loss": 0.7056,
+      "step": 9584
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.284637397502941e-05,
+      "loss": 0.793,
+      "step": 9585
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2844994293957982e-05,
+      "loss": 0.7979,
+      "step": 9586
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2843614553959676e-05,
+      "loss": 0.7114,
+      "step": 9587
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2842234755063064e-05,
+      "loss": 0.7188,
+      "step": 9588
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2840854897296726e-05,
+      "loss": 0.8545,
+      "step": 9589
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2839474980689246e-05,
+      "loss": 0.7295,
+      "step": 9590
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.28380950052692e-05,
+      "loss": 0.6956,
+      "step": 9591
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2836714971065175e-05,
+      "loss": 0.7173,
+      "step": 9592
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2835334878105751e-05,
+      "loss": 0.6377,
+      "step": 9593
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2833954726419518e-05,
+      "loss": 0.7366,
+      "step": 9594
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2832574516035057e-05,
+      "loss": 0.6689,
+      "step": 9595
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2831194246980959e-05,
+      "loss": 0.7847,
+      "step": 9596
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2829813919285812e-05,
+      "loss": 0.822,
+      "step": 9597
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2828433532978207e-05,
+      "loss": 0.8667,
+      "step": 9598
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2827053088086733e-05,
+      "loss": 0.8394,
+      "step": 9599
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2825672584639987e-05,
+      "loss": 0.813,
+      "step": 9600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2824292022666558e-05,
+      "loss": 0.8008,
+      "step": 9601
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2822911402195042e-05,
+      "loss": 0.7432,
+      "step": 9602
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2821530723254035e-05,
+      "loss": 0.6787,
+      "step": 9603
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2820149985872132e-05,
+      "loss": 0.72,
+      "step": 9604
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2818769190077937e-05,
+      "loss": 0.7632,
+      "step": 9605
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2817388335900047e-05,
+      "loss": 0.7241,
+      "step": 9606
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.281600742336706e-05,
+      "loss": 0.7617,
+      "step": 9607
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2814626452507586e-05,
+      "loss": 0.79,
+      "step": 9608
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2813245423350216e-05,
+      "loss": 0.7705,
+      "step": 9609
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2811864335923567e-05,
+      "loss": 0.7878,
+      "step": 9610
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2810483190256238e-05,
+      "loss": 0.7822,
+      "step": 9611
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2809101986376837e-05,
+      "loss": 0.7705,
+      "step": 9612
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2807720724313973e-05,
+      "loss": 0.7666,
+      "step": 9613
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2806339404096252e-05,
+      "loss": 0.7759,
+      "step": 9614
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2804958025752294e-05,
+      "loss": 0.8423,
+      "step": 9615
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2803576589310703e-05,
+      "loss": 0.7925,
+      "step": 9616
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2802195094800089e-05,
+      "loss": 0.7249,
+      "step": 9617
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2800813542249073e-05,
+      "loss": 0.647,
+      "step": 9618
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2799431931686267e-05,
+      "loss": 0.687,
+      "step": 9619
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2798050263140287e-05,
+      "loss": 0.8296,
+      "step": 9620
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2796668536639751e-05,
+      "loss": 0.688,
+      "step": 9621
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2795286752213283e-05,
+      "loss": 0.7124,
+      "step": 9622
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2793904909889496e-05,
+      "loss": 0.7324,
+      "step": 9623
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2792523009697017e-05,
+      "loss": 0.8242,
+      "step": 9624
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2791141051664467e-05,
+      "loss": 0.7236,
+      "step": 9625
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2789759035820466e-05,
+      "loss": 0.8198,
+      "step": 9626
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2788376962193643e-05,
+      "loss": 0.7332,
+      "step": 9627
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2786994830812625e-05,
+      "loss": 0.9014,
+      "step": 9628
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2785612641706036e-05,
+      "loss": 0.8115,
+      "step": 9629
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2784230394902509e-05,
+      "loss": 0.741,
+      "step": 9630
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2782848090430666e-05,
+      "loss": 0.8042,
+      "step": 9631
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2781465728319144e-05,
+      "loss": 0.7456,
+      "step": 9632
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2780083308596574e-05,
+      "loss": 0.7058,
+      "step": 9633
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.277870083129159e-05,
+      "loss": 0.7832,
+      "step": 9634
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2777318296432825e-05,
+      "loss": 0.8633,
+      "step": 9635
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2775935704048919e-05,
+      "loss": 0.7314,
+      "step": 9636
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2774553054168502e-05,
+      "loss": 0.7437,
+      "step": 9637
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2773170346820216e-05,
+      "loss": 0.7832,
+      "step": 9638
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.27717875820327e-05,
+      "loss": 0.8076,
+      "step": 9639
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2770404759834593e-05,
+      "loss": 0.6877,
+      "step": 9640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2769021880254542e-05,
+      "loss": 0.8325,
+      "step": 9641
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2767638943321181e-05,
+      "loss": 0.8213,
+      "step": 9642
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2766255949063162e-05,
+      "loss": 0.7642,
+      "step": 9643
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2764872897509128e-05,
+      "loss": 0.6936,
+      "step": 9644
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2763489788687723e-05,
+      "loss": 0.7915,
+      "step": 9645
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2762106622627595e-05,
+      "loss": 0.7568,
+      "step": 9646
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2760723399357396e-05,
+      "loss": 0.7759,
+      "step": 9647
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2759340118905771e-05,
+      "loss": 0.7678,
+      "step": 9648
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2757956781301376e-05,
+      "loss": 0.7812,
+      "step": 9649
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2756573386572864e-05,
+      "loss": 0.7922,
+      "step": 9650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2755189934748882e-05,
+      "loss": 0.8047,
+      "step": 9651
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.275380642585809e-05,
+      "loss": 0.6589,
+      "step": 9652
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2752422859929143e-05,
+      "loss": 0.71,
+      "step": 9653
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2751039236990698e-05,
+      "loss": 0.7539,
+      "step": 9654
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2749655557071414e-05,
+      "loss": 0.7466,
+      "step": 9655
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2748271820199948e-05,
+      "loss": 0.6084,
+      "step": 9656
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2746888026404962e-05,
+      "loss": 0.7778,
+      "step": 9657
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2745504175715118e-05,
+      "loss": 0.7158,
+      "step": 9658
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.274412026815908e-05,
+      "loss": 0.7881,
+      "step": 9659
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2742736303765508e-05,
+      "loss": 0.8428,
+      "step": 9660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2741352282563073e-05,
+      "loss": 0.7651,
+      "step": 9661
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2739968204580438e-05,
+      "loss": 0.8423,
+      "step": 9662
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2738584069846272e-05,
+      "loss": 0.7156,
+      "step": 9663
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.273719987838924e-05,
+      "loss": 0.8301,
+      "step": 9664
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2735815630238019e-05,
+      "loss": 0.752,
+      "step": 9665
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2734431325421277e-05,
+      "loss": 0.7563,
+      "step": 9666
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2733046963967683e-05,
+      "loss": 0.8398,
+      "step": 9667
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2731662545905916e-05,
+      "loss": 0.8213,
+      "step": 9668
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2730278071264648e-05,
+      "loss": 0.7461,
+      "step": 9669
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2728893540072553e-05,
+      "loss": 0.7195,
+      "step": 9670
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2727508952358311e-05,
+      "loss": 0.7559,
+      "step": 9671
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2726124308150602e-05,
+      "loss": 0.8232,
+      "step": 9672
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2724739607478098e-05,
+      "loss": 0.7358,
+      "step": 9673
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.272335485036949e-05,
+      "loss": 0.772,
+      "step": 9674
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2721970036853452e-05,
+      "loss": 0.8486,
+      "step": 9675
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.272058516695867e-05,
+      "loss": 0.6956,
+      "step": 9676
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2719200240713824e-05,
+      "loss": 0.7686,
+      "step": 9677
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2717815258147605e-05,
+      "loss": 0.7529,
+      "step": 9678
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2716430219288698e-05,
+      "loss": 0.7246,
+      "step": 9679
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2715045124165788e-05,
+      "loss": 0.7842,
+      "step": 9680
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2713659972807565e-05,
+      "loss": 0.6326,
+      "step": 9681
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2712274765242718e-05,
+      "loss": 0.7788,
+      "step": 9682
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2710889501499941e-05,
+      "loss": 0.8184,
+      "step": 9683
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2709504181607923e-05,
+      "loss": 0.7812,
+      "step": 9684
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2708118805595362e-05,
+      "loss": 0.7415,
+      "step": 9685
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2706733373490945e-05,
+      "loss": 0.7881,
+      "step": 9686
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2705347885323374e-05,
+      "loss": 0.6809,
+      "step": 9687
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2703962341121345e-05,
+      "loss": 0.7793,
+      "step": 9688
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2702576740913556e-05,
+      "loss": 0.7607,
+      "step": 9689
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2701191084728706e-05,
+      "loss": 0.7241,
+      "step": 9690
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2699805372595492e-05,
+      "loss": 0.8604,
+      "step": 9691
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.269841960454262e-05,
+      "loss": 0.7876,
+      "step": 9692
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2697033780598789e-05,
+      "loss": 0.7837,
+      "step": 9693
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2695647900792707e-05,
+      "loss": 0.7847,
+      "step": 9694
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2694261965153076e-05,
+      "loss": 0.6973,
+      "step": 9695
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2692875973708603e-05,
+      "loss": 0.8154,
+      "step": 9696
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2691489926487994e-05,
+      "loss": 0.8662,
+      "step": 9697
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2690103823519957e-05,
+      "loss": 0.7168,
+      "step": 9698
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2688717664833208e-05,
+      "loss": 0.74,
+      "step": 9699
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.268733145045645e-05,
+      "loss": 0.8262,
+      "step": 9700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.26859451804184e-05,
+      "loss": 0.8027,
+      "step": 9701
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2684558854747768e-05,
+      "loss": 0.8535,
+      "step": 9702
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2683172473473272e-05,
+      "loss": 0.7419,
+      "step": 9703
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2681786036623622e-05,
+      "loss": 0.7681,
+      "step": 9704
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2680399544227539e-05,
+      "loss": 0.7617,
+      "step": 9705
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2679012996313737e-05,
+      "loss": 0.8071,
+      "step": 9706
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.267762639291094e-05,
+      "loss": 0.6199,
+      "step": 9707
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2676239734047864e-05,
+      "loss": 0.8306,
+      "step": 9708
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2674853019753228e-05,
+      "loss": 0.7305,
+      "step": 9709
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2673466250055758e-05,
+      "loss": 0.7527,
+      "step": 9710
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2672079424984178e-05,
+      "loss": 0.8374,
+      "step": 9711
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2670692544567211e-05,
+      "loss": 0.7185,
+      "step": 9712
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2669305608833584e-05,
+      "loss": 0.7722,
+      "step": 9713
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2667918617812022e-05,
+      "loss": 0.8599,
+      "step": 9714
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2666531571531256e-05,
+      "loss": 0.8003,
+      "step": 9715
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2665144470020013e-05,
+      "loss": 0.7019,
+      "step": 9716
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.266375731330702e-05,
+      "loss": 0.73,
+      "step": 9717
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2662370101421013e-05,
+      "loss": 0.7603,
+      "step": 9718
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2660982834390727e-05,
+      "loss": 0.7178,
+      "step": 9719
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265959551224489e-05,
+      "loss": 0.8018,
+      "step": 9720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265820813501224e-05,
+      "loss": 0.7073,
+      "step": 9721
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2656820702721508e-05,
+      "loss": 0.7739,
+      "step": 9722
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2655433215401438e-05,
+      "loss": 0.7827,
+      "step": 9723
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2654045673080767e-05,
+      "loss": 0.8159,
+      "step": 9724
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265265807578823e-05,
+      "loss": 0.7849,
+      "step": 9725
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2651270423552572e-05,
+      "loss": 0.7305,
+      "step": 9726
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2649882716402535e-05,
+      "loss": 0.7598,
+      "step": 9727
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.264849495436686e-05,
+      "loss": 0.7998,
+      "step": 9728
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2647107137474288e-05,
+      "loss": 0.7578,
+      "step": 9729
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.264571926575357e-05,
+      "loss": 0.7002,
+      "step": 9730
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.264433133923345e-05,
+      "loss": 0.814,
+      "step": 9731
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2642943357942674e-05,
+      "loss": 0.897,
+      "step": 9732
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.264155532190999e-05,
+      "loss": 0.6973,
+      "step": 9733
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2640167231164156e-05,
+      "loss": 0.7793,
+      "step": 9734
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2638779085733908e-05,
+      "loss": 0.7468,
+      "step": 9735
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2637390885648007e-05,
+      "loss": 0.749,
+      "step": 9736
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2636002630935208e-05,
+      "loss": 0.813,
+      "step": 9737
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2634614321624259e-05,
+      "loss": 0.7495,
+      "step": 9738
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2633225957743919e-05,
+      "loss": 0.7422,
+      "step": 9739
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2631837539322946e-05,
+      "loss": 0.7485,
+      "step": 9740
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2630449066390096e-05,
+      "loss": 0.7363,
+      "step": 9741
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2629060538974125e-05,
+      "loss": 0.897,
+      "step": 9742
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2627671957103795e-05,
+      "loss": 0.8242,
+      "step": 9743
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2626283320807869e-05,
+      "loss": 0.7524,
+      "step": 9744
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2624894630115106e-05,
+      "loss": 0.8325,
+      "step": 9745
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.262350588505427e-05,
+      "loss": 0.8022,
+      "step": 9746
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2622117085654127e-05,
+      "loss": 0.7542,
+      "step": 9747
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.262072823194344e-05,
+      "loss": 0.771,
+      "step": 9748
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2619339323950976e-05,
+      "loss": 0.729,
+      "step": 9749
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2617950361705505e-05,
+      "loss": 0.7646,
+      "step": 9750
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2616561345235793e-05,
+      "loss": 0.7505,
+      "step": 9751
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2615172274570613e-05,
+      "loss": 0.7153,
+      "step": 9752
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2613783149738735e-05,
+      "loss": 0.7578,
+      "step": 9753
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.261239397076893e-05,
+      "loss": 0.7217,
+      "step": 9754
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2611004737689972e-05,
+      "loss": 0.7698,
+      "step": 9755
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2609615450530634e-05,
+      "loss": 0.7778,
+      "step": 9756
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2608226109319695e-05,
+      "loss": 0.8286,
+      "step": 9757
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2606836714085928e-05,
+      "loss": 0.772,
+      "step": 9758
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2605447264858112e-05,
+      "loss": 0.7983,
+      "step": 9759
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2604057761665029e-05,
+      "loss": 0.7009,
+      "step": 9760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2602668204535456e-05,
+      "loss": 0.853,
+      "step": 9761
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2601278593498174e-05,
+      "loss": 0.8135,
+      "step": 9762
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2599888928581967e-05,
+      "loss": 0.668,
+      "step": 9763
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2598499209815616e-05,
+      "loss": 0.677,
+      "step": 9764
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2597109437227907e-05,
+      "loss": 0.7461,
+      "step": 9765
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2595719610847626e-05,
+      "loss": 0.6797,
+      "step": 9766
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2594329730703558e-05,
+      "loss": 0.7668,
+      "step": 9767
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2592939796824494e-05,
+      "loss": 0.8062,
+      "step": 9768
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2591549809239223e-05,
+      "loss": 0.7598,
+      "step": 9769
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2590159767976528e-05,
+      "loss": 0.6973,
+      "step": 9770
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2588769673065211e-05,
+      "loss": 0.8208,
+      "step": 9771
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2587379524534055e-05,
+      "loss": 0.8623,
+      "step": 9772
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2585989322411857e-05,
+      "loss": 0.7427,
+      "step": 9773
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2584599066727416e-05,
+      "loss": 0.8218,
+      "step": 9774
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2583208757509517e-05,
+      "loss": 0.7627,
+      "step": 9775
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2581818394786967e-05,
+      "loss": 0.7935,
+      "step": 9776
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2580427978588561e-05,
+      "loss": 0.7588,
+      "step": 9777
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2579037508943098e-05,
+      "loss": 0.8252,
+      "step": 9778
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2577646985879373e-05,
+      "loss": 0.8428,
+      "step": 9779
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2576256409426192e-05,
+      "loss": 0.6899,
+      "step": 9780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2574865779612357e-05,
+      "loss": 0.6575,
+      "step": 9781
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2573475096466672e-05,
+      "loss": 0.9272,
+      "step": 9782
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2572084360017938e-05,
+      "loss": 0.7734,
+      "step": 9783
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2570693570294965e-05,
+      "loss": 0.6765,
+      "step": 9784
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2569302727326557e-05,
+      "loss": 0.6946,
+      "step": 9785
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2567911831141522e-05,
+      "loss": 0.8037,
+      "step": 9786
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2566520881768666e-05,
+      "loss": 0.8447,
+      "step": 9787
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2565129879236808e-05,
+      "loss": 0.7373,
+      "step": 9788
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2563738823574751e-05,
+      "loss": 0.7642,
+      "step": 9789
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2562347714811311e-05,
+      "loss": 0.7896,
+      "step": 9790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2560956552975299e-05,
+      "loss": 0.7305,
+      "step": 9791
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.255956533809553e-05,
+      "loss": 0.8223,
+      "step": 9792
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2558174070200822e-05,
+      "loss": 0.6694,
+      "step": 9793
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.255678274931999e-05,
+      "loss": 0.7925,
+      "step": 9794
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2555391375481847e-05,
+      "loss": 0.709,
+      "step": 9795
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2553999948715218e-05,
+      "loss": 0.8687,
+      "step": 9796
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2552608469048922e-05,
+      "loss": 0.7979,
+      "step": 9797
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2551216936511778e-05,
+      "loss": 0.6582,
+      "step": 9798
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2549825351132607e-05,
+      "loss": 0.8091,
+      "step": 9799
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2548433712940236e-05,
+      "loss": 0.7324,
+      "step": 9800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2547042021963487e-05,
+      "loss": 0.7871,
+      "step": 9801
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2545650278231188e-05,
+      "loss": 0.8467,
+      "step": 9802
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.254425848177216e-05,
+      "loss": 0.7524,
+      "step": 9803
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2542866632615235e-05,
+      "loss": 0.7798,
+      "step": 9804
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2541474730789243e-05,
+      "loss": 0.8345,
+      "step": 9805
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2540082776323009e-05,
+      "loss": 0.7871,
+      "step": 9806
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2538690769245365e-05,
+      "loss": 0.8247,
+      "step": 9807
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2537298709585149e-05,
+      "loss": 0.7837,
+      "step": 9808
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2535906597371183e-05,
+      "loss": 0.8169,
+      "step": 9809
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2534514432632312e-05,
+      "loss": 0.8574,
+      "step": 9810
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2533122215397362e-05,
+      "loss": 0.7012,
+      "step": 9811
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2531729945695172e-05,
+      "loss": 0.8066,
+      "step": 9812
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2530337623554585e-05,
+      "loss": 0.7954,
+      "step": 9813
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2528945249004433e-05,
+      "loss": 0.793,
+      "step": 9814
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2527552822073562e-05,
+      "loss": 0.8149,
+      "step": 9815
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2526160342790803e-05,
+      "loss": 0.8813,
+      "step": 9816
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2524767811185007e-05,
+      "loss": 0.6968,
+      "step": 9817
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.252337522728501e-05,
+      "loss": 0.7627,
+      "step": 9818
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2521982591119661e-05,
+      "loss": 0.8154,
+      "step": 9819
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2520589902717801e-05,
+      "loss": 0.7324,
+      "step": 9820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2519197162108281e-05,
+      "loss": 0.7822,
+      "step": 9821
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2517804369319941e-05,
+      "loss": 0.8022,
+      "step": 9822
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2516411524381637e-05,
+      "loss": 0.7324,
+      "step": 9823
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.251501862732221e-05,
+      "loss": 0.8325,
+      "step": 9824
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2513625678170517e-05,
+      "loss": 0.7039,
+      "step": 9825
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2512232676955407e-05,
+      "loss": 0.8945,
+      "step": 9826
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2510839623705733e-05,
+      "loss": 0.748,
+      "step": 9827
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2509446518450348e-05,
+      "loss": 0.8081,
+      "step": 9828
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2508053361218106e-05,
+      "loss": 0.7461,
+      "step": 9829
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2506660152037864e-05,
+      "loss": 0.8652,
+      "step": 9830
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2505266890938479e-05,
+      "loss": 0.7344,
+      "step": 9831
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2503873577948808e-05,
+      "loss": 0.7349,
+      "step": 9832
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2502480213097709e-05,
+      "loss": 0.6714,
+      "step": 9833
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2501086796414046e-05,
+      "loss": 0.8989,
+      "step": 9834
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2499693327926674e-05,
+      "loss": 0.8374,
+      "step": 9835
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2498299807664463e-05,
+      "loss": 0.6831,
+      "step": 9836
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2496906235656267e-05,
+      "loss": 0.7537,
+      "step": 9837
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2495512611930957e-05,
+      "loss": 0.6968,
+      "step": 9838
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.24941189365174e-05,
+      "loss": 0.8096,
+      "step": 9839
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2492725209444454e-05,
+      "loss": 0.7581,
+      "step": 9840
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2491331430740994e-05,
+      "loss": 0.6521,
+      "step": 9841
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2489937600435886e-05,
+      "loss": 0.793,
+      "step": 9842
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2488543718558002e-05,
+      "loss": 0.7373,
+      "step": 9843
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2487149785136208e-05,
+      "loss": 0.8069,
+      "step": 9844
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2485755800199378e-05,
+      "loss": 0.8301,
+      "step": 9845
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2484361763776387e-05,
+      "loss": 0.793,
+      "step": 9846
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2482967675896107e-05,
+      "loss": 0.6873,
+      "step": 9847
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2481573536587412e-05,
+      "loss": 0.8481,
+      "step": 9848
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2480179345879182e-05,
+      "loss": 0.7495,
+      "step": 9849
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2478785103800288e-05,
+      "loss": 0.7354,
+      "step": 9850
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2477390810379615e-05,
+      "loss": 0.7812,
+      "step": 9851
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2475996465646037e-05,
+      "loss": 0.8276,
+      "step": 9852
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2474602069628437e-05,
+      "loss": 0.7192,
+      "step": 9853
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2473207622355696e-05,
+      "loss": 0.7896,
+      "step": 9854
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2471813123856695e-05,
+      "loss": 0.7837,
+      "step": 9855
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.247041857416032e-05,
+      "loss": 0.739,
+      "step": 9856
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2469023973295456e-05,
+      "loss": 0.7607,
+      "step": 9857
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2467629321290984e-05,
+      "loss": 0.7671,
+      "step": 9858
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2466234618175794e-05,
+      "loss": 0.6848,
+      "step": 9859
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2464839863978774e-05,
+      "loss": 0.7632,
+      "step": 9860
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.246344505872881e-05,
+      "loss": 0.7827,
+      "step": 9861
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2462050202454798e-05,
+      "loss": 0.7642,
+      "step": 9862
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2460655295185625e-05,
+      "loss": 0.7905,
+      "step": 9863
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2459260336950179e-05,
+      "loss": 0.813,
+      "step": 9864
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2457865327777358e-05,
+      "loss": 0.7766,
+      "step": 9865
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2456470267696057e-05,
+      "loss": 0.8145,
+      "step": 9866
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2455075156735172e-05,
+      "loss": 0.8022,
+      "step": 9867
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2453679994923594e-05,
+      "loss": 0.7642,
+      "step": 9868
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.245228478229022e-05,
+      "loss": 0.7656,
+      "step": 9869
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2450889518863953e-05,
+      "loss": 0.8408,
+      "step": 9870
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.244949420467369e-05,
+      "loss": 0.739,
+      "step": 9871
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2448098839748335e-05,
+      "loss": 0.697,
+      "step": 9872
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2446703424116783e-05,
+      "loss": 0.8647,
+      "step": 9873
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.244530795780794e-05,
+      "loss": 0.7046,
+      "step": 9874
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2443912440850709e-05,
+      "loss": 0.688,
+      "step": 9875
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2442516873273996e-05,
+      "loss": 0.7415,
+      "step": 9876
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2441121255106706e-05,
+      "loss": 0.7441,
+      "step": 9877
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2439725586377744e-05,
+      "loss": 0.8232,
+      "step": 9878
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.243832986711602e-05,
+      "loss": 0.7441,
+      "step": 9879
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2436934097350441e-05,
+      "loss": 0.8062,
+      "step": 9880
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2435538277109919e-05,
+      "loss": 0.7583,
+      "step": 9881
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2434142406423369e-05,
+      "loss": 0.6958,
+      "step": 9882
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2432746485319692e-05,
+      "loss": 0.7842,
+      "step": 9883
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2431350513827805e-05,
+      "loss": 0.8152,
+      "step": 9884
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2429954491976626e-05,
+      "loss": 0.7612,
+      "step": 9885
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2428558419795067e-05,
+      "loss": 0.8579,
+      "step": 9886
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2427162297312045e-05,
+      "loss": 0.8447,
+      "step": 9887
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2425766124556477e-05,
+      "loss": 0.7322,
+      "step": 9888
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.242436990155728e-05,
+      "loss": 0.698,
+      "step": 9889
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2422973628343376e-05,
+      "loss": 0.7512,
+      "step": 9890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2421577304943684e-05,
+      "loss": 0.7666,
+      "step": 9891
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2420180931387123e-05,
+      "loss": 0.7207,
+      "step": 9892
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2418784507702619e-05,
+      "loss": 0.8149,
+      "step": 9893
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2417388033919093e-05,
+      "loss": 0.7744,
+      "step": 9894
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2415991510065473e-05,
+      "loss": 0.7661,
+      "step": 9895
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2414594936170678e-05,
+      "loss": 0.8228,
+      "step": 9896
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.241319831226364e-05,
+      "loss": 0.7461,
+      "step": 9897
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2411801638373287e-05,
+      "loss": 0.7754,
+      "step": 9898
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2410404914528542e-05,
+      "loss": 0.772,
+      "step": 9899
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2409008140758339e-05,
+      "loss": 0.688,
+      "step": 9900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2407611317091606e-05,
+      "loss": 0.7871,
+      "step": 9901
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2406214443557276e-05,
+      "loss": 0.7954,
+      "step": 9902
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2404817520184285e-05,
+      "loss": 0.8203,
+      "step": 9903
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2403420547001562e-05,
+      "loss": 0.7131,
+      "step": 9904
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2402023524038043e-05,
+      "loss": 0.7227,
+      "step": 9905
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2400626451322665e-05,
+      "loss": 0.874,
+      "step": 9906
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2399229328884363e-05,
+      "loss": 0.8188,
+      "step": 9907
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.239783215675208e-05,
+      "loss": 0.7646,
+      "step": 9908
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2396434934954747e-05,
+      "loss": 0.6382,
+      "step": 9909
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2395037663521308e-05,
+      "loss": 0.8169,
+      "step": 9910
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2393640342480703e-05,
+      "loss": 0.7373,
+      "step": 9911
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2392242971861878e-05,
+      "loss": 0.7812,
+      "step": 9912
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2390845551693771e-05,
+      "loss": 0.7664,
+      "step": 9913
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2389448082005329e-05,
+      "loss": 0.6943,
+      "step": 9914
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2388050562825493e-05,
+      "loss": 0.6831,
+      "step": 9915
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2386652994183214e-05,
+      "loss": 0.7939,
+      "step": 9916
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2385255376107437e-05,
+      "loss": 0.6641,
+      "step": 9917
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2383857708627109e-05,
+      "loss": 0.8154,
+      "step": 9918
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2382459991771181e-05,
+      "loss": 0.8779,
+      "step": 9919
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2381062225568603e-05,
+      "loss": 0.9019,
+      "step": 9920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2379664410048324e-05,
+      "loss": 0.6189,
+      "step": 9921
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2378266545239301e-05,
+      "loss": 0.7036,
+      "step": 9922
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2376868631170483e-05,
+      "loss": 0.7974,
+      "step": 9923
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2375470667870823e-05,
+      "loss": 0.7686,
+      "step": 9924
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2374072655369282e-05,
+      "loss": 0.8789,
+      "step": 9925
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2372674593694812e-05,
+      "loss": 0.7441,
+      "step": 9926
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.237127648287637e-05,
+      "loss": 0.7329,
+      "step": 9927
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2369878322942919e-05,
+      "loss": 0.7676,
+      "step": 9928
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2368480113923411e-05,
+      "loss": 0.7764,
+      "step": 9929
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2367081855846812e-05,
+      "loss": 0.752,
+      "step": 9930
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2365683548742083e-05,
+      "loss": 0.792,
+      "step": 9931
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2364285192638184e-05,
+      "loss": 0.7949,
+      "step": 9932
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2362886787564082e-05,
+      "loss": 0.7881,
+      "step": 9933
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2361488333548736e-05,
+      "loss": 0.7373,
+      "step": 9934
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2360089830621115e-05,
+      "loss": 0.7456,
+      "step": 9935
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2358691278810186e-05,
+      "loss": 0.8228,
+      "step": 9936
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2357292678144917e-05,
+      "loss": 0.7451,
+      "step": 9937
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2355894028654272e-05,
+      "loss": 0.7217,
+      "step": 9938
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2354495330367225e-05,
+      "loss": 0.7441,
+      "step": 9939
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2353096583312742e-05,
+      "loss": 0.7759,
+      "step": 9940
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2351697787519798e-05,
+      "loss": 0.7754,
+      "step": 9941
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2350298943017371e-05,
+      "loss": 0.689,
+      "step": 9942
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2348900049834424e-05,
+      "loss": 0.7085,
+      "step": 9943
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2347501107999937e-05,
+      "loss": 0.8413,
+      "step": 9944
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2346102117542885e-05,
+      "loss": 0.8306,
+      "step": 9945
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2344703078492241e-05,
+      "loss": 0.75,
+      "step": 9946
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.234330399087699e-05,
+      "loss": 0.8774,
+      "step": 9947
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2341904854726104e-05,
+      "loss": 0.814,
+      "step": 9948
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2340505670068566e-05,
+      "loss": 0.7761,
+      "step": 9949
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2339106436933355e-05,
+      "loss": 0.6643,
+      "step": 9950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2337707155349452e-05,
+      "loss": 0.8599,
+      "step": 9951
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2336307825345841e-05,
+      "loss": 0.9116,
+      "step": 9952
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2334908446951503e-05,
+      "loss": 0.7559,
+      "step": 9953
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2333509020195428e-05,
+      "loss": 0.7041,
+      "step": 9954
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.23321095451066e-05,
+      "loss": 0.769,
+      "step": 9955
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2330710021714002e-05,
+      "loss": 0.7627,
+      "step": 9956
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2329310450046626e-05,
+      "loss": 0.8521,
+      "step": 9957
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2327910830133456e-05,
+      "loss": 0.7808,
+      "step": 9958
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2326511162003482e-05,
+      "loss": 0.7964,
+      "step": 9959
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2325111445685701e-05,
+      "loss": 0.7158,
+      "step": 9960
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2323711681209096e-05,
+      "loss": 0.8784,
+      "step": 9961
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2322311868602666e-05,
+      "loss": 0.7671,
+      "step": 9962
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.23209120078954e-05,
+      "loss": 0.7795,
+      "step": 9963
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2319512099116298e-05,
+      "loss": 0.8044,
+      "step": 9964
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.231811214229435e-05,
+      "loss": 0.6934,
+      "step": 9965
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2316712137458553e-05,
+      "loss": 0.7349,
+      "step": 9966
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2315312084637909e-05,
+      "loss": 0.7549,
+      "step": 9967
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2313911983861414e-05,
+      "loss": 0.7832,
+      "step": 9968
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2312511835158067e-05,
+      "loss": 0.7051,
+      "step": 9969
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.231111163855687e-05,
+      "loss": 0.8511,
+      "step": 9970
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2309711394086826e-05,
+      "loss": 0.7158,
+      "step": 9971
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2308311101776933e-05,
+      "loss": 0.6975,
+      "step": 9972
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2306910761656195e-05,
+      "loss": 0.7759,
+      "step": 9973
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.230551037375362e-05,
+      "loss": 0.7246,
+      "step": 9974
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2304109938098213e-05,
+      "loss": 0.813,
+      "step": 9975
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2302709454718977e-05,
+      "loss": 0.7539,
+      "step": 9976
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2301308923644923e-05,
+      "loss": 0.8892,
+      "step": 9977
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2299908344905056e-05,
+      "loss": 0.8223,
+      "step": 9978
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.229850771852839e-05,
+      "loss": 0.7222,
+      "step": 9979
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2297107044543932e-05,
+      "loss": 0.7476,
+      "step": 9980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2295706322980696e-05,
+      "loss": 0.6602,
+      "step": 9981
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.229430555386769e-05,
+      "loss": 0.7925,
+      "step": 9982
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2292904737233935e-05,
+      "loss": 0.8359,
+      "step": 9983
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2291503873108437e-05,
+      "loss": 0.8105,
+      "step": 9984
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2290102961520217e-05,
+      "loss": 0.7593,
+      "step": 9985
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2288702002498292e-05,
+      "loss": 0.7476,
+      "step": 9986
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2287300996071674e-05,
+      "loss": 0.7634,
+      "step": 9987
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2285899942269384e-05,
+      "loss": 0.7666,
+      "step": 9988
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.228449884112044e-05,
+      "loss": 0.7217,
+      "step": 9989
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2283097692653865e-05,
+      "loss": 0.7517,
+      "step": 9990
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2281696496898679e-05,
+      "loss": 0.7803,
+      "step": 9991
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2280295253883906e-05,
+      "loss": 0.7656,
+      "step": 9992
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2278893963638563e-05,
+      "loss": 0.7319,
+      "step": 9993
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2277492626191683e-05,
+      "loss": 0.7217,
+      "step": 9994
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2276091241572283e-05,
+      "loss": 0.7915,
+      "step": 9995
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2274689809809398e-05,
+      "loss": 0.7852,
+      "step": 9996
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2273288330932044e-05,
+      "loss": 0.8613,
+      "step": 9997
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2271886804969259e-05,
+      "loss": 0.7466,
+      "step": 9998
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2270485231950069e-05,
+      "loss": 0.7905,
+      "step": 9999
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.22690836119035e-05,
+      "loss": 0.8569,
+      "step": 10000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.226768194485859e-05,
+      "loss": 0.6929,
+      "step": 10001
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2266280230844366e-05,
+      "loss": 0.7939,
+      "step": 10002
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.226487846988986e-05,
+      "loss": 0.8486,
+      "step": 10003
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2263476662024112e-05,
+      "loss": 0.8291,
+      "step": 10004
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.226207480727615e-05,
+      "loss": 0.7295,
+      "step": 10005
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2260672905675015e-05,
+      "loss": 0.7275,
+      "step": 10006
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2259270957249741e-05,
+      "loss": 0.6587,
+      "step": 10007
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2257868962029366e-05,
+      "loss": 0.7578,
+      "step": 10008
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.225646692004293e-05,
+      "loss": 0.8501,
+      "step": 10009
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2255064831319472e-05,
+      "loss": 0.7302,
+      "step": 10010
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2253662695888036e-05,
+      "loss": 0.7791,
+      "step": 10011
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2252260513777657e-05,
+      "loss": 0.8125,
+      "step": 10012
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2250858285017381e-05,
+      "loss": 0.7329,
+      "step": 10013
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2249456009636257e-05,
+      "loss": 0.8477,
+      "step": 10014
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2248053687663319e-05,
+      "loss": 0.7637,
+      "step": 10015
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.224665131912762e-05,
+      "loss": 0.8062,
+      "step": 10016
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2245248904058204e-05,
+      "loss": 0.7927,
+      "step": 10017
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2243846442484122e-05,
+      "loss": 0.7163,
+      "step": 10018
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2242443934434415e-05,
+      "loss": 0.7974,
+      "step": 10019
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.224104137993814e-05,
+      "loss": 0.8477,
+      "step": 10020
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2239638779024346e-05,
+      "loss": 0.8745,
+      "step": 10021
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2238236131722081e-05,
+      "loss": 0.7549,
+      "step": 10022
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.22368334380604e-05,
+      "loss": 0.7617,
+      "step": 10023
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2235430698068353e-05,
+      "loss": 0.7773,
+      "step": 10024
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2234027911775e-05,
+      "loss": 0.7598,
+      "step": 10025
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2232625079209389e-05,
+      "loss": 0.8262,
+      "step": 10026
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2231222200400581e-05,
+      "loss": 0.6794,
+      "step": 10027
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2229819275377633e-05,
+      "loss": 0.7759,
+      "step": 10028
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2228416304169601e-05,
+      "loss": 0.7271,
+      "step": 10029
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2227013286805545e-05,
+      "loss": 0.6738,
+      "step": 10030
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2225610223314525e-05,
+      "loss": 0.7537,
+      "step": 10031
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2224207113725603e-05,
+      "loss": 0.7954,
+      "step": 10032
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.222280395806784e-05,
+      "loss": 0.6448,
+      "step": 10033
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.22214007563703e-05,
+      "loss": 0.7227,
+      "step": 10034
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2219997508662043e-05,
+      "loss": 0.8052,
+      "step": 10035
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2218594214972134e-05,
+      "loss": 0.749,
+      "step": 10036
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2217190875329645e-05,
+      "loss": 0.8354,
+      "step": 10037
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2215787489763637e-05,
+      "loss": 0.7388,
+      "step": 10038
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.221438405830318e-05,
+      "loss": 0.6963,
+      "step": 10039
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2212980580977339e-05,
+      "loss": 0.792,
+      "step": 10040
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2211577057815186e-05,
+      "loss": 0.6997,
+      "step": 10041
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2210173488845794e-05,
+      "loss": 0.7656,
+      "step": 10042
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.220876987409823e-05,
+      "loss": 0.7876,
+      "step": 10043
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2207366213601568e-05,
+      "loss": 0.7917,
+      "step": 10044
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2205962507384883e-05,
+      "loss": 0.8472,
+      "step": 10045
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2204558755477251e-05,
+      "loss": 0.8137,
+      "step": 10046
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2203154957907742e-05,
+      "loss": 0.7183,
+      "step": 10047
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2201751114705432e-05,
+      "loss": 0.7339,
+      "step": 10048
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.22003472258994e-05,
+      "loss": 0.7268,
+      "step": 10049
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2198943291518725e-05,
+      "loss": 0.7568,
+      "step": 10050
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2197539311592486e-05,
+      "loss": 0.7954,
+      "step": 10051
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2196135286149762e-05,
+      "loss": 0.7651,
+      "step": 10052
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2194731215219634e-05,
+      "loss": 0.7427,
+      "step": 10053
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2193327098831182e-05,
+      "loss": 0.8887,
+      "step": 10054
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2191922937013489e-05,
+      "loss": 0.6953,
+      "step": 10055
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2190518729795641e-05,
+      "loss": 0.6665,
+      "step": 10056
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2189114477206724e-05,
+      "loss": 0.877,
+      "step": 10057
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2187710179275822e-05,
+      "loss": 0.8223,
+      "step": 10058
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.218630583603202e-05,
+      "loss": 0.6909,
+      "step": 10059
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2184901447504406e-05,
+      "loss": 0.7168,
+      "step": 10060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2183497013722067e-05,
+      "loss": 0.8086,
+      "step": 10061
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2182092534714096e-05,
+      "loss": 0.854,
+      "step": 10062
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.218068801050958e-05,
+      "loss": 0.9258,
+      "step": 10063
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2179283441137612e-05,
+      "loss": 0.731,
+      "step": 10064
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2177878826627284e-05,
+      "loss": 0.7683,
+      "step": 10065
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2176474167007685e-05,
+      "loss": 0.7573,
+      "step": 10066
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2175069462307914e-05,
+      "loss": 0.6934,
+      "step": 10067
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2173664712557063e-05,
+      "loss": 0.7969,
+      "step": 10068
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.217225991778423e-05,
+      "loss": 0.7715,
+      "step": 10069
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2170855078018513e-05,
+      "loss": 0.7227,
+      "step": 10070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2169450193289008e-05,
+      "loss": 0.8433,
+      "step": 10071
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2168045263624811e-05,
+      "loss": 0.6973,
+      "step": 10072
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2166640289055025e-05,
+      "loss": 0.7866,
+      "step": 10073
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2165235269608748e-05,
+      "loss": 0.7913,
+      "step": 10074
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2163830205315088e-05,
+      "loss": 0.8081,
+      "step": 10075
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2162425096203136e-05,
+      "loss": 0.7603,
+      "step": 10076
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2161019942302005e-05,
+      "loss": 0.8057,
+      "step": 10077
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2159614743640795e-05,
+      "loss": 0.7935,
+      "step": 10078
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2158209500248608e-05,
+      "loss": 0.8237,
+      "step": 10079
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2156804212154558e-05,
+      "loss": 0.7393,
+      "step": 10080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2155398879387748e-05,
+      "loss": 0.7214,
+      "step": 10081
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2153993501977283e-05,
+      "loss": 0.708,
+      "step": 10082
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2152588079952277e-05,
+      "loss": 0.7998,
+      "step": 10083
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2151182613341836e-05,
+      "loss": 0.7163,
+      "step": 10084
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2149777102175076e-05,
+      "loss": 0.7466,
+      "step": 10085
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2148371546481102e-05,
+      "loss": 0.8096,
+      "step": 10086
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2146965946289029e-05,
+      "loss": 0.6626,
+      "step": 10087
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.214556030162797e-05,
+      "loss": 0.8496,
+      "step": 10088
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2144154612527041e-05,
+      "loss": 0.7185,
+      "step": 10089
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.214274887901536e-05,
+      "loss": 0.7563,
+      "step": 10090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2141343101122037e-05,
+      "loss": 0.8066,
+      "step": 10091
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2139937278876189e-05,
+      "loss": 0.749,
+      "step": 10092
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2138531412306937e-05,
+      "loss": 0.6926,
+      "step": 10093
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2137125501443403e-05,
+      "loss": 0.7891,
+      "step": 10094
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2135719546314704e-05,
+      "loss": 0.6919,
+      "step": 10095
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2134313546949959e-05,
+      "loss": 0.7581,
+      "step": 10096
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2132907503378294e-05,
+      "loss": 0.7915,
+      "step": 10097
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2131501415628825e-05,
+      "loss": 0.7827,
+      "step": 10098
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2130095283730681e-05,
+      "loss": 0.7324,
+      "step": 10099
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2128689107712985e-05,
+      "loss": 0.7622,
+      "step": 10100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2127282887604864e-05,
+      "loss": 0.7861,
+      "step": 10101
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2125876623435443e-05,
+      "loss": 0.7583,
+      "step": 10102
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2124470315233845e-05,
+      "loss": 0.7891,
+      "step": 10103
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2123063963029205e-05,
+      "loss": 0.8428,
+      "step": 10104
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2121657566850649e-05,
+      "loss": 0.7852,
+      "step": 10105
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2120251126727305e-05,
+      "loss": 0.6895,
+      "step": 10106
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2118844642688307e-05,
+      "loss": 0.6611,
+      "step": 10107
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2117438114762787e-05,
+      "loss": 0.7456,
+      "step": 10108
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2116031542979876e-05,
+      "loss": 0.7373,
+      "step": 10109
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2114624927368708e-05,
+      "loss": 0.7437,
+      "step": 10110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2113218267958416e-05,
+      "loss": 0.6953,
+      "step": 10111
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2111811564778137e-05,
+      "loss": 0.8267,
+      "step": 10112
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2110404817857009e-05,
+      "loss": 0.8887,
+      "step": 10113
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2108998027224166e-05,
+      "loss": 0.8403,
+      "step": 10114
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2107591192908748e-05,
+      "loss": 0.7866,
+      "step": 10115
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2106184314939893e-05,
+      "loss": 0.8364,
+      "step": 10116
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2104777393346742e-05,
+      "loss": 0.7681,
+      "step": 10117
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2103370428158434e-05,
+      "loss": 0.7866,
+      "step": 10118
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2101963419404114e-05,
+      "loss": 0.7207,
+      "step": 10119
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2100556367112921e-05,
+      "loss": 0.7378,
+      "step": 10120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2099149271314e-05,
+      "loss": 0.8276,
+      "step": 10121
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2097742132036498e-05,
+      "loss": 0.7832,
+      "step": 10122
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2096334949309556e-05,
+      "loss": 0.7734,
+      "step": 10123
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2094927723162321e-05,
+      "loss": 0.7686,
+      "step": 10124
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2093520453623943e-05,
+      "loss": 0.7822,
+      "step": 10125
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.209211314072357e-05,
+      "loss": 0.7808,
+      "step": 10126
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2090705784490348e-05,
+      "loss": 0.7637,
+      "step": 10127
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2089298384953428e-05,
+      "loss": 0.7866,
+      "step": 10128
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.208789094214196e-05,
+      "loss": 0.7476,
+      "step": 10129
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2086483456085096e-05,
+      "loss": 0.8066,
+      "step": 10130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2085075926811991e-05,
+      "loss": 0.6941,
+      "step": 10131
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2083668354351793e-05,
+      "loss": 0.7139,
+      "step": 10132
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2082260738733664e-05,
+      "loss": 0.6812,
+      "step": 10133
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2080853079986752e-05,
+      "loss": 0.7974,
+      "step": 10134
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.207944537814022e-05,
+      "loss": 0.792,
+      "step": 10135
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2078037633223218e-05,
+      "loss": 0.7603,
+      "step": 10136
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2076629845264908e-05,
+      "loss": 0.8911,
+      "step": 10137
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2075222014294448e-05,
+      "loss": 0.8423,
+      "step": 10138
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2073814140340999e-05,
+      "loss": 0.8535,
+      "step": 10139
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2072406223433715e-05,
+      "loss": 0.7891,
+      "step": 10140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2070998263601766e-05,
+      "loss": 0.6577,
+      "step": 10141
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.206959026087431e-05,
+      "loss": 0.7988,
+      "step": 10142
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2068182215280511e-05,
+      "loss": 0.8369,
+      "step": 10143
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.206677412684953e-05,
+      "loss": 0.7622,
+      "step": 10144
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2065365995610538e-05,
+      "loss": 0.7622,
+      "step": 10145
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2063957821592698e-05,
+      "loss": 0.8159,
+      "step": 10146
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2062549604825178e-05,
+      "loss": 0.7705,
+      "step": 10147
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2061141345337141e-05,
+      "loss": 0.7822,
+      "step": 10148
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2059733043157761e-05,
+      "loss": 0.8223,
+      "step": 10149
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2058324698316208e-05,
+      "loss": 0.7651,
+      "step": 10150
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2056916310841649e-05,
+      "loss": 0.7632,
+      "step": 10151
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2055507880763254e-05,
+      "loss": 0.8276,
+      "step": 10152
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2054099408110195e-05,
+      "loss": 0.7854,
+      "step": 10153
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2052690892911652e-05,
+      "loss": 0.7603,
+      "step": 10154
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.205128233519679e-05,
+      "loss": 0.7686,
+      "step": 10155
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2049873734994785e-05,
+      "loss": 0.7356,
+      "step": 10156
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2048465092334818e-05,
+      "loss": 0.8232,
+      "step": 10157
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2047056407246063e-05,
+      "loss": 0.9263,
+      "step": 10158
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2045647679757697e-05,
+      "loss": 0.771,
+      "step": 10159
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2044238909898897e-05,
+      "loss": 0.7988,
+      "step": 10160
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2042830097698845e-05,
+      "loss": 0.6724,
+      "step": 10161
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2041421243186719e-05,
+      "loss": 0.8413,
+      "step": 10162
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.20400123463917e-05,
+      "loss": 0.7146,
+      "step": 10163
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2038603407342973e-05,
+      "loss": 0.8013,
+      "step": 10164
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2037194426069713e-05,
+      "loss": 0.811,
+      "step": 10165
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.203578540260111e-05,
+      "loss": 0.79,
+      "step": 10166
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2034376336966344e-05,
+      "loss": 0.771,
+      "step": 10167
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2032967229194606e-05,
+      "loss": 0.7896,
+      "step": 10168
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2031558079315073e-05,
+      "loss": 0.7446,
+      "step": 10169
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2030148887356943e-05,
+      "loss": 0.7261,
+      "step": 10170
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2028739653349395e-05,
+      "loss": 0.771,
+      "step": 10171
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2027330377321624e-05,
+      "loss": 0.8511,
+      "step": 10172
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2025921059302817e-05,
+      "loss": 0.8086,
+      "step": 10173
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2024511699322163e-05,
+      "loss": 0.6567,
+      "step": 10174
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2023102297408855e-05,
+      "loss": 0.7886,
+      "step": 10175
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2021692853592086e-05,
+      "loss": 0.6934,
+      "step": 10176
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2020283367901047e-05,
+      "loss": 0.8574,
+      "step": 10177
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2018873840364933e-05,
+      "loss": 0.7534,
+      "step": 10178
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2017464271012941e-05,
+      "loss": 0.8306,
+      "step": 10179
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2016054659874263e-05,
+      "loss": 0.6228,
+      "step": 10180
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2014645006978094e-05,
+      "loss": 0.8091,
+      "step": 10181
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2013235312353638e-05,
+      "loss": 0.772,
+      "step": 10182
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2011825576030088e-05,
+      "loss": 0.7524,
+      "step": 10183
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2010415798036644e-05,
+      "loss": 0.7212,
+      "step": 10184
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2009005978402506e-05,
+      "loss": 0.7031,
+      "step": 10185
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2007596117156878e-05,
+      "loss": 0.7305,
+      "step": 10186
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2006186214328958e-05,
+      "loss": 0.7417,
+      "step": 10187
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.200477626994795e-05,
+      "loss": 0.7744,
+      "step": 10188
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2003366284043059e-05,
+      "loss": 0.6301,
+      "step": 10189
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2001956256643488e-05,
+      "loss": 0.833,
+      "step": 10190
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.200054618777844e-05,
+      "loss": 0.8774,
+      "step": 10191
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1999136077477123e-05,
+      "loss": 0.832,
+      "step": 10192
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1997725925768744e-05,
+      "loss": 0.7422,
+      "step": 10193
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1996315732682508e-05,
+      "loss": 0.7578,
+      "step": 10194
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.199490549824763e-05,
+      "loss": 0.7271,
+      "step": 10195
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1993495222493315e-05,
+      "loss": 0.7937,
+      "step": 10196
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.199208490544877e-05,
+      "loss": 0.6829,
+      "step": 10197
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1990674547143215e-05,
+      "loss": 0.8306,
+      "step": 10198
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1989264147605854e-05,
+      "loss": 0.8372,
+      "step": 10199
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1987853706865904e-05,
+      "loss": 0.6997,
+      "step": 10200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1986443224952576e-05,
+      "loss": 0.749,
+      "step": 10201
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1985032701895087e-05,
+      "loss": 0.6885,
+      "step": 10202
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1983622137722653e-05,
+      "loss": 0.7617,
+      "step": 10203
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1982211532464488e-05,
+      "loss": 0.7871,
+      "step": 10204
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1980800886149809e-05,
+      "loss": 0.6924,
+      "step": 10205
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1979390198807835e-05,
+      "loss": 0.7354,
+      "step": 10206
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1977979470467785e-05,
+      "loss": 0.7939,
+      "step": 10207
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.197656870115888e-05,
+      "loss": 0.7659,
+      "step": 10208
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1975157890910336e-05,
+      "loss": 0.8708,
+      "step": 10209
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.197374703975138e-05,
+      "loss": 0.8555,
+      "step": 10210
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.197233614771123e-05,
+      "loss": 0.7891,
+      "step": 10211
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1970925214819112e-05,
+      "loss": 0.7271,
+      "step": 10212
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1969514241104247e-05,
+      "loss": 0.7627,
+      "step": 10213
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1968103226595863e-05,
+      "loss": 0.822,
+      "step": 10214
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1966692171323184e-05,
+      "loss": 0.7529,
+      "step": 10215
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1965281075315436e-05,
+      "loss": 0.8218,
+      "step": 10216
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1963869938601846e-05,
+      "loss": 0.7744,
+      "step": 10217
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1962458761211643e-05,
+      "loss": 0.7512,
+      "step": 10218
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1961047543174056e-05,
+      "loss": 0.7686,
+      "step": 10219
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1959636284518314e-05,
+      "loss": 0.729,
+      "step": 10220
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1958224985273648e-05,
+      "loss": 0.6577,
+      "step": 10221
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1956813645469292e-05,
+      "loss": 1.0059,
+      "step": 10222
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1955402265134475e-05,
+      "loss": 0.7881,
+      "step": 10223
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1953990844298434e-05,
+      "loss": 0.7007,
+      "step": 10224
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1952579382990396e-05,
+      "loss": 0.7361,
+      "step": 10225
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1951167881239602e-05,
+      "loss": 0.7324,
+      "step": 10226
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1949756339075286e-05,
+      "loss": 0.696,
+      "step": 10227
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1948344756526685e-05,
+      "loss": 0.7959,
+      "step": 10228
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1946933133623033e-05,
+      "loss": 0.6614,
+      "step": 10229
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1945521470393576e-05,
+      "loss": 0.7793,
+      "step": 10230
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1944109766867544e-05,
+      "loss": 0.7046,
+      "step": 10231
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1942698023074182e-05,
+      "loss": 0.7432,
+      "step": 10232
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1941286239042729e-05,
+      "loss": 0.8555,
+      "step": 10233
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1939874414802429e-05,
+      "loss": 0.7388,
+      "step": 10234
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.193846255038252e-05,
+      "loss": 0.7822,
+      "step": 10235
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1937050645812249e-05,
+      "loss": 0.8633,
+      "step": 10236
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.193563870112086e-05,
+      "loss": 0.7742,
+      "step": 10237
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1934226716337596e-05,
+      "loss": 0.7803,
+      "step": 10238
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1932814691491705e-05,
+      "loss": 0.6667,
+      "step": 10239
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1931402626612431e-05,
+      "loss": 0.7495,
+      "step": 10240
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1929990521729023e-05,
+      "loss": 0.7905,
+      "step": 10241
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1928578376870728e-05,
+      "loss": 0.7866,
+      "step": 10242
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1927166192066795e-05,
+      "loss": 0.7915,
+      "step": 10243
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1925753967346474e-05,
+      "loss": 0.7544,
+      "step": 10244
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1924341702739015e-05,
+      "loss": 0.709,
+      "step": 10245
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.192292939827367e-05,
+      "loss": 0.7837,
+      "step": 10246
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1921517053979694e-05,
+      "loss": 0.7437,
+      "step": 10247
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1920104669886339e-05,
+      "loss": 0.748,
+      "step": 10248
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1918692246022855e-05,
+      "loss": 0.7847,
+      "step": 10249
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1917279782418502e-05,
+      "loss": 0.7217,
+      "step": 10250
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1915867279102535e-05,
+      "loss": 0.7891,
+      "step": 10251
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1914454736104206e-05,
+      "loss": 0.6904,
+      "step": 10252
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1913042153452776e-05,
+      "loss": 0.8418,
+      "step": 10253
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1911629531177506e-05,
+      "loss": 0.6519,
+      "step": 10254
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1910216869307647e-05,
+      "loss": 0.7153,
+      "step": 10255
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1908804167872463e-05,
+      "loss": 0.7012,
+      "step": 10256
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1907391426901217e-05,
+      "loss": 0.7842,
+      "step": 10257
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1905978646423162e-05,
+      "loss": 0.7744,
+      "step": 10258
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1904565826467569e-05,
+      "loss": 0.7603,
+      "step": 10259
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1903152967063701e-05,
+      "loss": 0.7061,
+      "step": 10260
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1901740068240814e-05,
+      "loss": 0.9004,
+      "step": 10261
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1900327130028183e-05,
+      "loss": 0.739,
+      "step": 10262
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1898914152455064e-05,
+      "loss": 0.7334,
+      "step": 10263
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1897501135550727e-05,
+      "loss": 0.7539,
+      "step": 10264
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1896088079344439e-05,
+      "loss": 0.7764,
+      "step": 10265
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1894674983865469e-05,
+      "loss": 0.7656,
+      "step": 10266
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1893261849143084e-05,
+      "loss": 0.7056,
+      "step": 10267
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1891848675206559e-05,
+      "loss": 0.658,
+      "step": 10268
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1890435462085156e-05,
+      "loss": 0.7395,
+      "step": 10269
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1889022209808148e-05,
+      "loss": 0.7671,
+      "step": 10270
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1887608918404809e-05,
+      "loss": 0.6865,
+      "step": 10271
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1886195587904414e-05,
+      "loss": 0.7515,
+      "step": 10272
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1884782218336235e-05,
+      "loss": 0.729,
+      "step": 10273
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1883368809729543e-05,
+      "loss": 0.7778,
+      "step": 10274
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1881955362113615e-05,
+      "loss": 0.769,
+      "step": 10275
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.188054187551773e-05,
+      "loss": 0.8501,
+      "step": 10276
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1879128349971164e-05,
+      "loss": 0.7405,
+      "step": 10277
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.187771478550319e-05,
+      "loss": 0.7598,
+      "step": 10278
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1876301182143091e-05,
+      "loss": 0.6943,
+      "step": 10279
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1874887539920144e-05,
+      "loss": 0.7563,
+      "step": 10280
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.187347385886363e-05,
+      "loss": 0.7478,
+      "step": 10281
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.187206013900283e-05,
+      "loss": 0.7881,
+      "step": 10282
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1870646380367027e-05,
+      "loss": 0.7944,
+      "step": 10283
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1869232582985497e-05,
+      "loss": 0.8149,
+      "step": 10284
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1867818746887531e-05,
+      "loss": 0.7542,
+      "step": 10285
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.186640487210241e-05,
+      "loss": 0.7646,
+      "step": 10286
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.186499095865942e-05,
+      "loss": 0.7397,
+      "step": 10287
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1863577006587843e-05,
+      "loss": 0.752,
+      "step": 10288
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.186216301591697e-05,
+      "loss": 0.8081,
+      "step": 10289
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1860748986676084e-05,
+      "loss": 0.7749,
+      "step": 10290
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1859334918894479e-05,
+      "loss": 0.8301,
+      "step": 10291
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1857920812601438e-05,
+      "loss": 0.8457,
+      "step": 10292
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1856506667826253e-05,
+      "loss": 0.7666,
+      "step": 10293
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1855092484598214e-05,
+      "loss": 0.708,
+      "step": 10294
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1853678262946613e-05,
+      "loss": 0.7383,
+      "step": 10295
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.185226400290074e-05,
+      "loss": 0.7593,
+      "step": 10296
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.185084970448989e-05,
+      "loss": 0.7466,
+      "step": 10297
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1849435367743358e-05,
+      "loss": 0.9009,
+      "step": 10298
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1848020992690434e-05,
+      "loss": 0.6663,
+      "step": 10299
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1846606579360418e-05,
+      "loss": 0.6963,
+      "step": 10300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.18451921277826e-05,
+      "loss": 0.7034,
+      "step": 10301
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1843777637986282e-05,
+      "loss": 0.6763,
+      "step": 10302
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1842363110000763e-05,
+      "loss": 0.8271,
+      "step": 10303
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1840948543855336e-05,
+      "loss": 0.7188,
+      "step": 10304
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1839533939579302e-05,
+      "loss": 0.7192,
+      "step": 10305
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.183811929720196e-05,
+      "loss": 0.7285,
+      "step": 10306
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1836704616752615e-05,
+      "loss": 0.7676,
+      "step": 10307
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1835289898260565e-05,
+      "loss": 0.8164,
+      "step": 10308
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1833875141755113e-05,
+      "loss": 0.6948,
+      "step": 10309
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.183246034726556e-05,
+      "loss": 0.7822,
+      "step": 10310
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1831045514821215e-05,
+      "loss": 0.8511,
+      "step": 10311
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1829630644451377e-05,
+      "loss": 0.7034,
+      "step": 10312
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.182821573618536e-05,
+      "loss": 0.8164,
+      "step": 10313
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1826800790052462e-05,
+      "loss": 0.6982,
+      "step": 10314
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.182538580608199e-05,
+      "loss": 0.7739,
+      "step": 10315
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1823970784303258e-05,
+      "loss": 0.8096,
+      "step": 10316
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1822555724745569e-05,
+      "loss": 0.8506,
+      "step": 10317
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1821140627438235e-05,
+      "loss": 0.7812,
+      "step": 10318
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1819725492410564e-05,
+      "loss": 0.8086,
+      "step": 10319
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1818310319691868e-05,
+      "loss": 0.6423,
+      "step": 10320
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1816895109311462e-05,
+      "loss": 0.8242,
+      "step": 10321
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1815479861298656e-05,
+      "loss": 0.7192,
+      "step": 10322
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1814064575682757e-05,
+      "loss": 0.8613,
+      "step": 10323
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1812649252493091e-05,
+      "loss": 0.8071,
+      "step": 10324
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1811233891758967e-05,
+      "loss": 0.7383,
+      "step": 10325
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1809818493509697e-05,
+      "loss": 0.7744,
+      "step": 10326
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1808403057774605e-05,
+      "loss": 0.7629,
+      "step": 10327
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1806987584583006e-05,
+      "loss": 0.8042,
+      "step": 10328
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.180557207396421e-05,
+      "loss": 0.8896,
+      "step": 10329
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1804156525947545e-05,
+      "loss": 0.7744,
+      "step": 10330
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1802740940562328e-05,
+      "loss": 0.7166,
+      "step": 10331
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1801325317837877e-05,
+      "loss": 0.8433,
+      "step": 10332
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1799909657803517e-05,
+      "loss": 0.7827,
+      "step": 10333
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1798493960488567e-05,
+      "loss": 0.7412,
+      "step": 10334
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1797078225922345e-05,
+      "loss": 0.7944,
+      "step": 10335
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1795662454134184e-05,
+      "loss": 0.6458,
+      "step": 10336
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1794246645153404e-05,
+      "loss": 0.6533,
+      "step": 10337
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1792830799009329e-05,
+      "loss": 0.8101,
+      "step": 10338
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1791414915731285e-05,
+      "loss": 0.7676,
+      "step": 10339
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.17899989953486e-05,
+      "loss": 0.7207,
+      "step": 10340
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.17885830378906e-05,
+      "loss": 0.7825,
+      "step": 10341
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.178716704338661e-05,
+      "loss": 0.769,
+      "step": 10342
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1785751011865967e-05,
+      "loss": 0.791,
+      "step": 10343
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1784334943357993e-05,
+      "loss": 0.7671,
+      "step": 10344
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.178291883789202e-05,
+      "loss": 0.811,
+      "step": 10345
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1781502695497379e-05,
+      "loss": 0.7092,
+      "step": 10346
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1780086516203399e-05,
+      "loss": 0.7749,
+      "step": 10347
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1778670300039422e-05,
+      "loss": 0.7749,
+      "step": 10348
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.177725404703477e-05,
+      "loss": 0.7173,
+      "step": 10349
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1775837757218786e-05,
+      "loss": 0.8032,
+      "step": 10350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.17744214306208e-05,
+      "loss": 0.8203,
+      "step": 10351
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.177300506727015e-05,
+      "loss": 0.75,
+      "step": 10352
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.177158866719617e-05,
+      "loss": 0.8442,
+      "step": 10353
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1770172230428198e-05,
+      "loss": 0.7666,
+      "step": 10354
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1768755756995573e-05,
+      "loss": 0.7964,
+      "step": 10355
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1767339246927636e-05,
+      "loss": 0.7349,
+      "step": 10356
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.176592270025372e-05,
+      "loss": 0.7661,
+      "step": 10357
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1764506117003172e-05,
+      "loss": 0.7344,
+      "step": 10358
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1763089497205327e-05,
+      "loss": 0.8691,
+      "step": 10359
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1761672840889529e-05,
+      "loss": 0.7227,
+      "step": 10360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1760256148085121e-05,
+      "loss": 0.876,
+      "step": 10361
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1758839418821446e-05,
+      "loss": 0.8042,
+      "step": 10362
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1757422653127848e-05,
+      "loss": 0.7915,
+      "step": 10363
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1756005851033675e-05,
+      "loss": 0.8071,
+      "step": 10364
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1754589012568266e-05,
+      "loss": 0.7607,
+      "step": 10365
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1753172137760969e-05,
+      "loss": 0.7627,
+      "step": 10366
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1751755226641134e-05,
+      "loss": 0.6836,
+      "step": 10367
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.175033827923811e-05,
+      "loss": 0.8896,
+      "step": 10368
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.174892129558124e-05,
+      "loss": 0.7393,
+      "step": 10369
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1747504275699875e-05,
+      "loss": 0.7925,
+      "step": 10370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1746087219623368e-05,
+      "loss": 0.7427,
+      "step": 10371
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1744670127381066e-05,
+      "loss": 0.8462,
+      "step": 10372
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1743252999002324e-05,
+      "loss": 0.7983,
+      "step": 10373
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1741835834516491e-05,
+      "loss": 0.7522,
+      "step": 10374
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1740418633952922e-05,
+      "loss": 0.7373,
+      "step": 10375
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1739001397340969e-05,
+      "loss": 0.8008,
+      "step": 10376
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1737584124709986e-05,
+      "loss": 0.8115,
+      "step": 10377
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1736166816089334e-05,
+      "loss": 0.8315,
+      "step": 10378
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1734749471508361e-05,
+      "loss": 0.6858,
+      "step": 10379
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.173333209099643e-05,
+      "loss": 0.7361,
+      "step": 10380
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1731914674582896e-05,
+      "loss": 0.7954,
+      "step": 10381
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1730497222297116e-05,
+      "loss": 0.812,
+      "step": 10382
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1729079734168452e-05,
+      "loss": 0.7402,
+      "step": 10383
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1727662210226258e-05,
+      "loss": 0.7749,
+      "step": 10384
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1726244650499902e-05,
+      "loss": 0.7356,
+      "step": 10385
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.172482705501874e-05,
+      "loss": 0.7822,
+      "step": 10386
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1723409423812135e-05,
+      "loss": 0.7285,
+      "step": 10387
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1721991756909452e-05,
+      "loss": 0.791,
+      "step": 10388
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.172057405434005e-05,
+      "loss": 0.792,
+      "step": 10389
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1719156316133295e-05,
+      "loss": 0.7886,
+      "step": 10390
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1717738542318555e-05,
+      "loss": 0.751,
+      "step": 10391
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1716320732925193e-05,
+      "loss": 0.8267,
+      "step": 10392
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1714902887982573e-05,
+      "loss": 0.7737,
+      "step": 10393
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1713485007520066e-05,
+      "loss": 0.7539,
+      "step": 10394
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1712067091567039e-05,
+      "loss": 0.769,
+      "step": 10395
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1710649140152857e-05,
+      "loss": 0.8062,
+      "step": 10396
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1709231153306895e-05,
+      "loss": 0.7886,
+      "step": 10397
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1707813131058518e-05,
+      "loss": 0.8286,
+      "step": 10398
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.17063950734371e-05,
+      "loss": 0.7759,
+      "step": 10399
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1704976980472009e-05,
+      "loss": 0.7803,
+      "step": 10400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.170355885219262e-05,
+      "loss": 0.6799,
+      "step": 10401
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1702140688628312e-05,
+      "loss": 0.7661,
+      "step": 10402
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1700722489808447e-05,
+      "loss": 0.7783,
+      "step": 10403
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1699304255762404e-05,
+      "loss": 0.7283,
+      "step": 10404
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1697885986519559e-05,
+      "loss": 0.8228,
+      "step": 10405
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1696467682109288e-05,
+      "loss": 0.7197,
+      "step": 10406
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1695049342560969e-05,
+      "loss": 0.6826,
+      "step": 10407
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1693630967903974e-05,
+      "loss": 0.6174,
+      "step": 10408
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1692212558167688e-05,
+      "loss": 0.832,
+      "step": 10409
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1690794113381484e-05,
+      "loss": 0.7681,
+      "step": 10410
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1689375633574746e-05,
+      "loss": 0.6406,
+      "step": 10411
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.168795711877685e-05,
+      "loss": 0.8271,
+      "step": 10412
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.168653856901718e-05,
+      "loss": 0.7854,
+      "step": 10413
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1685119984325117e-05,
+      "loss": 0.7563,
+      "step": 10414
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1683701364730045e-05,
+      "loss": 0.7637,
+      "step": 10415
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1682282710261342e-05,
+      "loss": 0.8408,
+      "step": 10416
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1680864020948399e-05,
+      "loss": 0.7983,
+      "step": 10417
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1679445296820596e-05,
+      "loss": 0.6895,
+      "step": 10418
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1678026537907316e-05,
+      "loss": 0.7715,
+      "step": 10419
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1676607744237949e-05,
+      "loss": 0.7173,
+      "step": 10420
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1675188915841882e-05,
+      "loss": 0.8071,
+      "step": 10421
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.16737700527485e-05,
+      "loss": 0.8208,
+      "step": 10422
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1672351154987193e-05,
+      "loss": 0.9292,
+      "step": 10423
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1670932222587348e-05,
+      "loss": 0.6582,
+      "step": 10424
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1669513255578356e-05,
+      "loss": 0.8262,
+      "step": 10425
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1668094253989608e-05,
+      "loss": 0.7341,
+      "step": 10426
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1666675217850494e-05,
+      "loss": 0.7021,
+      "step": 10427
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1665256147190407e-05,
+      "loss": 0.8442,
+      "step": 10428
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1663837042038738e-05,
+      "loss": 0.6865,
+      "step": 10429
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.166241790242488e-05,
+      "loss": 0.749,
+      "step": 10430
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.166099872837823e-05,
+      "loss": 0.814,
+      "step": 10431
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.165957951992818e-05,
+      "loss": 0.7939,
+      "step": 10432
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1658160277104122e-05,
+      "loss": 0.8296,
+      "step": 10433
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1656740999935458e-05,
+      "loss": 0.7642,
+      "step": 10434
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1655321688451582e-05,
+      "loss": 0.8503,
+      "step": 10435
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1653902342681891e-05,
+      "loss": 0.8569,
+      "step": 10436
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1652482962655782e-05,
+      "loss": 0.7954,
+      "step": 10437
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1651063548402659e-05,
+      "loss": 0.7856,
+      "step": 10438
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1649644099951915e-05,
+      "loss": 0.7434,
+      "step": 10439
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1648224617332957e-05,
+      "loss": 0.7651,
+      "step": 10440
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1646805100575183e-05,
+      "loss": 0.7866,
+      "step": 10441
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1645385549707992e-05,
+      "loss": 0.7515,
+      "step": 10442
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1643965964760791e-05,
+      "loss": 0.7925,
+      "step": 10443
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1642546345762979e-05,
+      "loss": 0.8164,
+      "step": 10444
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1641126692743962e-05,
+      "loss": 0.7378,
+      "step": 10445
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1639707005733145e-05,
+      "loss": 0.7471,
+      "step": 10446
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1638287284759935e-05,
+      "loss": 0.7178,
+      "step": 10447
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.163686752985373e-05,
+      "loss": 0.7153,
+      "step": 10448
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1635447741043944e-05,
+      "loss": 0.688,
+      "step": 10449
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1634027918359982e-05,
+      "loss": 0.8086,
+      "step": 10450
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1632608061831255e-05,
+      "loss": 0.8042,
+      "step": 10451
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1631188171487167e-05,
+      "loss": 0.8457,
+      "step": 10452
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1629768247357132e-05,
+      "loss": 0.9468,
+      "step": 10453
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1628348289470555e-05,
+      "loss": 0.7236,
+      "step": 10454
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.162692829785685e-05,
+      "loss": 0.7681,
+      "step": 10455
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.162550827254543e-05,
+      "loss": 0.7083,
+      "step": 10456
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1624088213565705e-05,
+      "loss": 0.7522,
+      "step": 10457
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1622668120947086e-05,
+      "loss": 0.752,
+      "step": 10458
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1621247994718993e-05,
+      "loss": 0.7432,
+      "step": 10459
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1619827834910836e-05,
+      "loss": 0.6704,
+      "step": 10460
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1618407641552027e-05,
+      "loss": 0.7385,
+      "step": 10461
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1616987414671988e-05,
+      "loss": 0.7617,
+      "step": 10462
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1615567154300132e-05,
+      "loss": 0.7603,
+      "step": 10463
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1614146860465874e-05,
+      "loss": 0.6404,
+      "step": 10464
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1612726533198637e-05,
+      "loss": 0.7173,
+      "step": 10465
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1611306172527836e-05,
+      "loss": 0.9009,
+      "step": 10466
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1609885778482892e-05,
+      "loss": 0.8008,
+      "step": 10467
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1608465351093224e-05,
+      "loss": 0.8105,
+      "step": 10468
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1607044890388252e-05,
+      "loss": 0.7437,
+      "step": 10469
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.16056243963974e-05,
+      "loss": 0.8066,
+      "step": 10470
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1604203869150083e-05,
+      "loss": 0.8071,
+      "step": 10471
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1602783308675732e-05,
+      "loss": 0.7388,
+      "step": 10472
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1601362715003765e-05,
+      "loss": 0.8438,
+      "step": 10473
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.159994208816361e-05,
+      "loss": 0.7075,
+      "step": 10474
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1598521428184687e-05,
+      "loss": 0.7676,
+      "step": 10475
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1597100735096426e-05,
+      "loss": 0.7939,
+      "step": 10476
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1595680008928253e-05,
+      "loss": 0.7148,
+      "step": 10477
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1594259249709587e-05,
+      "loss": 0.7817,
+      "step": 10478
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1592838457469866e-05,
+      "loss": 0.7285,
+      "step": 10479
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1591417632238511e-05,
+      "loss": 0.6306,
+      "step": 10480
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1589996774044953e-05,
+      "loss": 0.6704,
+      "step": 10481
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1588575882918622e-05,
+      "loss": 0.6931,
+      "step": 10482
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1587154958888946e-05,
+      "loss": 0.8232,
+      "step": 10483
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.158573400198536e-05,
+      "loss": 0.8394,
+      "step": 10484
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1584313012237294e-05,
+      "loss": 0.7349,
+      "step": 10485
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1582891989674176e-05,
+      "loss": 0.7222,
+      "step": 10486
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1581470934325444e-05,
+      "loss": 0.8052,
+      "step": 10487
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.158004984622053e-05,
+      "loss": 0.7124,
+      "step": 10488
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1578628725388868e-05,
+      "loss": 0.8103,
+      "step": 10489
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1577207571859892e-05,
+      "loss": 0.7949,
+      "step": 10490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1575786385663044e-05,
+      "loss": 0.7949,
+      "step": 10491
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.157436516682775e-05,
+      "loss": 0.6892,
+      "step": 10492
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1572943915383453e-05,
+      "loss": 0.7554,
+      "step": 10493
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1571522631359588e-05,
+      "loss": 0.6914,
+      "step": 10494
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1570101314785598e-05,
+      "loss": 0.8408,
+      "step": 10495
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1568679965690917e-05,
+      "loss": 0.7349,
+      "step": 10496
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1567258584104986e-05,
+      "loss": 0.7207,
+      "step": 10497
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1565837170057246e-05,
+      "loss": 0.7188,
+      "step": 10498
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1564415723577138e-05,
+      "loss": 0.8535,
+      "step": 10499
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1562994244694103e-05,
+      "loss": 0.8252,
+      "step": 10500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1561572733437582e-05,
+      "loss": 0.7439,
+      "step": 10501
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1560151189837023e-05,
+      "loss": 0.8057,
+      "step": 10502
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1558729613921866e-05,
+      "loss": 0.6978,
+      "step": 10503
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1557308005721557e-05,
+      "loss": 0.843,
+      "step": 10504
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1555886365265539e-05,
+      "loss": 0.6846,
+      "step": 10505
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.155446469258326e-05,
+      "loss": 0.7744,
+      "step": 10506
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1553042987704163e-05,
+      "loss": 0.7764,
+      "step": 10507
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.15516212506577e-05,
+      "loss": 0.8164,
+      "step": 10508
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1550199481473314e-05,
+      "loss": 0.7949,
+      "step": 10509
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1548777680180454e-05,
+      "loss": 0.7234,
+      "step": 10510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1547355846808572e-05,
+      "loss": 0.7778,
+      "step": 10511
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1545933981387114e-05,
+      "loss": 0.8784,
+      "step": 10512
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1544512083945534e-05,
+      "loss": 0.7478,
+      "step": 10513
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1543090154513278e-05,
+      "loss": 0.7151,
+      "step": 10514
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1541668193119801e-05,
+      "loss": 0.8066,
+      "step": 10515
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1540246199794558e-05,
+      "loss": 0.728,
+      "step": 10516
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1538824174567e-05,
+      "loss": 0.8779,
+      "step": 10517
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1537402117466578e-05,
+      "loss": 0.8618,
+      "step": 10518
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1535980028522748e-05,
+      "loss": 0.7036,
+      "step": 10519
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1534557907764968e-05,
+      "loss": 0.7661,
+      "step": 10520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.153313575522269e-05,
+      "loss": 0.832,
+      "step": 10521
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.153171357092537e-05,
+      "loss": 0.8579,
+      "step": 10522
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1530291354902467e-05,
+      "loss": 0.7661,
+      "step": 10523
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1528869107183438e-05,
+      "loss": 0.8115,
+      "step": 10524
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1527446827797739e-05,
+      "loss": 0.7266,
+      "step": 10525
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1526024516774832e-05,
+      "loss": 0.7183,
+      "step": 10526
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1524602174144175e-05,
+      "loss": 0.7598,
+      "step": 10527
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.152317979993523e-05,
+      "loss": 0.791,
+      "step": 10528
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1521757394177456e-05,
+      "loss": 0.7544,
+      "step": 10529
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1520334956900317e-05,
+      "loss": 0.7136,
+      "step": 10530
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1518912488133272e-05,
+      "loss": 0.8018,
+      "step": 10531
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1517489987905783e-05,
+      "loss": 0.7373,
+      "step": 10532
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.151606745624732e-05,
+      "loss": 0.8442,
+      "step": 10533
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1514644893187339e-05,
+      "loss": 0.8032,
+      "step": 10534
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.151322229875531e-05,
+      "loss": 0.7271,
+      "step": 10535
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1511799672980702e-05,
+      "loss": 0.8677,
+      "step": 10536
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1510377015892974e-05,
+      "loss": 0.8188,
+      "step": 10537
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1508954327521591e-05,
+      "loss": 0.7358,
+      "step": 10538
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1507531607896025e-05,
+      "loss": 0.7773,
+      "step": 10539
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1506108857045746e-05,
+      "loss": 0.7285,
+      "step": 10540
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1504686075000219e-05,
+      "loss": 0.8052,
+      "step": 10541
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1503263261788918e-05,
+      "loss": 0.8452,
+      "step": 10542
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1501840417441306e-05,
+      "loss": 0.7891,
+      "step": 10543
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1500417541986856e-05,
+      "loss": 0.8267,
+      "step": 10544
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1498994635455045e-05,
+      "loss": 0.7979,
+      "step": 10545
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1497571697875338e-05,
+      "loss": 0.8516,
+      "step": 10546
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.149614872927721e-05,
+      "loss": 0.7686,
+      "step": 10547
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1494725729690133e-05,
+      "loss": 0.7539,
+      "step": 10548
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1493302699143584e-05,
+      "loss": 0.7185,
+      "step": 10549
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1491879637667033e-05,
+      "loss": 0.8848,
+      "step": 10550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1490456545289961e-05,
+      "loss": 0.7305,
+      "step": 10551
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1489033422041838e-05,
+      "loss": 0.8916,
+      "step": 10552
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1487610267952143e-05,
+      "loss": 0.7104,
+      "step": 10553
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1486187083050355e-05,
+      "loss": 0.7788,
+      "step": 10554
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1484763867365947e-05,
+      "loss": 0.6953,
+      "step": 10555
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.14833406209284e-05,
+      "loss": 0.7197,
+      "step": 10556
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1481917343767194e-05,
+      "loss": 0.8105,
+      "step": 10557
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1480494035911806e-05,
+      "loss": 0.6973,
+      "step": 10558
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1479070697391719e-05,
+      "loss": 0.7686,
+      "step": 10559
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1477647328236413e-05,
+      "loss": 0.7305,
+      "step": 10560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1476223928475367e-05,
+      "loss": 0.7832,
+      "step": 10561
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1474800498138068e-05,
+      "loss": 0.8164,
+      "step": 10562
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1473377037253994e-05,
+      "loss": 0.6982,
+      "step": 10563
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.147195354585263e-05,
+      "loss": 0.708,
+      "step": 10564
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1470530023963462e-05,
+      "loss": 0.6389,
+      "step": 10565
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1469106471615971e-05,
+      "loss": 0.7905,
+      "step": 10566
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1467682888839647e-05,
+      "loss": 0.7959,
+      "step": 10567
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1466259275663971e-05,
+      "loss": 0.8271,
+      "step": 10568
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1464835632118432e-05,
+      "loss": 0.6951,
+      "step": 10569
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1463411958232518e-05,
+      "loss": 0.6733,
+      "step": 10570
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1461988254035715e-05,
+      "loss": 0.7637,
+      "step": 10571
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1460564519557511e-05,
+      "loss": 0.7725,
+      "step": 10572
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1459140754827398e-05,
+      "loss": 0.6978,
+      "step": 10573
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1457716959874865e-05,
+      "loss": 0.7212,
+      "step": 10574
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.14562931347294e-05,
+      "loss": 0.7441,
+      "step": 10575
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1454869279420495e-05,
+      "loss": 0.7458,
+      "step": 10576
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1453445393977642e-05,
+      "loss": 0.7822,
+      "step": 10577
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1452021478430333e-05,
+      "loss": 0.7783,
+      "step": 10578
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1450597532808061e-05,
+      "loss": 0.6333,
+      "step": 10579
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1449173557140319e-05,
+      "loss": 0.7498,
+      "step": 10580
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1447749551456604e-05,
+      "loss": 0.7344,
+      "step": 10581
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1446325515786408e-05,
+      "loss": 0.73,
+      "step": 10582
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1444901450159223e-05,
+      "loss": 0.7783,
+      "step": 10583
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1443477354604548e-05,
+      "loss": 0.7561,
+      "step": 10584
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1442053229151884e-05,
+      "loss": 0.7998,
+      "step": 10585
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.144062907383072e-05,
+      "loss": 0.7065,
+      "step": 10586
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1439204888670561e-05,
+      "loss": 0.7378,
+      "step": 10587
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.14377806737009e-05,
+      "loss": 0.8027,
+      "step": 10588
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1436356428951238e-05,
+      "loss": 0.7466,
+      "step": 10589
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1434932154451078e-05,
+      "loss": 0.6479,
+      "step": 10590
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.143350785022991e-05,
+      "loss": 0.7634,
+      "step": 10591
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1432083516317248e-05,
+      "loss": 0.6926,
+      "step": 10592
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1430659152742587e-05,
+      "loss": 0.6753,
+      "step": 10593
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1429234759535428e-05,
+      "loss": 0.8086,
+      "step": 10594
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1427810336725278e-05,
+      "loss": 0.749,
+      "step": 10595
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1426385884341637e-05,
+      "loss": 0.7554,
+      "step": 10596
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1424961402414008e-05,
+      "loss": 0.79,
+      "step": 10597
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1423536890971897e-05,
+      "loss": 0.7761,
+      "step": 10598
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1422112350044812e-05,
+      "loss": 0.7832,
+      "step": 10599
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1420687779662254e-05,
+      "loss": 0.793,
+      "step": 10600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1419263179853731e-05,
+      "loss": 0.7175,
+      "step": 10601
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.141783855064875e-05,
+      "loss": 0.7214,
+      "step": 10602
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1416413892076818e-05,
+      "loss": 0.7344,
+      "step": 10603
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1414989204167446e-05,
+      "loss": 0.7632,
+      "step": 10604
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1413564486950141e-05,
+      "loss": 0.6855,
+      "step": 10605
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1412139740454414e-05,
+      "loss": 0.7622,
+      "step": 10606
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.141071496470977e-05,
+      "loss": 0.7231,
+      "step": 10607
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1409290159745725e-05,
+      "loss": 0.7681,
+      "step": 10608
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.140786532559179e-05,
+      "loss": 0.8262,
+      "step": 10609
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1406440462277475e-05,
+      "loss": 0.6733,
+      "step": 10610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1405015569832292e-05,
+      "loss": 0.8257,
+      "step": 10611
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1403590648285755e-05,
+      "loss": 0.6997,
+      "step": 10612
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1402165697667376e-05,
+      "loss": 0.8228,
+      "step": 10613
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1400740718006672e-05,
+      "loss": 0.8027,
+      "step": 10614
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1399315709333151e-05,
+      "loss": 0.7422,
+      "step": 10615
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1397890671676342e-05,
+      "loss": 0.7378,
+      "step": 10616
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1396465605065747e-05,
+      "loss": 0.8896,
+      "step": 10617
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1395040509530892e-05,
+      "loss": 0.7798,
+      "step": 10618
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1393615385101291e-05,
+      "loss": 0.811,
+      "step": 10619
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1392190231806462e-05,
+      "loss": 0.6934,
+      "step": 10620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1390765049675923e-05,
+      "loss": 0.6624,
+      "step": 10621
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1389339838739192e-05,
+      "loss": 0.7266,
+      "step": 10622
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1387914599025791e-05,
+      "loss": 0.7944,
+      "step": 10623
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1386489330565238e-05,
+      "loss": 0.6992,
+      "step": 10624
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.138506403338706e-05,
+      "loss": 0.8047,
+      "step": 10625
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1383638707520767e-05,
+      "loss": 0.7861,
+      "step": 10626
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.138221335299589e-05,
+      "loss": 0.7627,
+      "step": 10627
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.138078796984195e-05,
+      "loss": 0.7754,
+      "step": 10628
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1379362558088466e-05,
+      "loss": 0.6965,
+      "step": 10629
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.137793711776497e-05,
+      "loss": 0.7095,
+      "step": 10630
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1376511648900979e-05,
+      "loss": 0.7466,
+      "step": 10631
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1375086151526021e-05,
+      "loss": 0.7041,
+      "step": 10632
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1373660625669621e-05,
+      "loss": 0.731,
+      "step": 10633
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1372235071361306e-05,
+      "loss": 0.8804,
+      "step": 10634
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1370809488630601e-05,
+      "loss": 0.6794,
+      "step": 10635
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1369383877507035e-05,
+      "loss": 0.7515,
+      "step": 10636
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1367958238020136e-05,
+      "loss": 0.6959,
+      "step": 10637
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1366532570199432e-05,
+      "loss": 0.7507,
+      "step": 10638
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1365106874074449e-05,
+      "loss": 0.7437,
+      "step": 10639
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1363681149674724e-05,
+      "loss": 0.7925,
+      "step": 10640
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.136225539702978e-05,
+      "loss": 0.8169,
+      "step": 10641
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1360829616169151e-05,
+      "loss": 0.8198,
+      "step": 10642
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.135940380712237e-05,
+      "loss": 0.7881,
+      "step": 10643
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1357977969918967e-05,
+      "loss": 0.7476,
+      "step": 10644
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1356552104588473e-05,
+      "loss": 0.7788,
+      "step": 10645
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1355126211160425e-05,
+      "loss": 0.7209,
+      "step": 10646
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1353700289664356e-05,
+      "loss": 0.7529,
+      "step": 10647
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1352274340129799e-05,
+      "loss": 0.8564,
+      "step": 10648
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1350848362586286e-05,
+      "loss": 0.699,
+      "step": 10649
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.134942235706336e-05,
+      "loss": 0.7449,
+      "step": 10650
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1347996323590551e-05,
+      "loss": 0.7603,
+      "step": 10651
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.13465702621974e-05,
+      "loss": 0.8335,
+      "step": 10652
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1345144172913438e-05,
+      "loss": 0.7808,
+      "step": 10653
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1343718055768208e-05,
+      "loss": 0.7656,
+      "step": 10654
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1342291910791249e-05,
+      "loss": 0.792,
+      "step": 10655
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1340865738012097e-05,
+      "loss": 0.6663,
+      "step": 10656
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1339439537460291e-05,
+      "loss": 0.7505,
+      "step": 10657
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1338013309165375e-05,
+      "loss": 0.8389,
+      "step": 10658
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1336587053156886e-05,
+      "loss": 0.7773,
+      "step": 10659
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1335160769464369e-05,
+      "loss": 0.7603,
+      "step": 10660
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1333734458117362e-05,
+      "loss": 0.7896,
+      "step": 10661
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.133230811914541e-05,
+      "loss": 0.7974,
+      "step": 10662
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1330881752578055e-05,
+      "loss": 0.813,
+      "step": 10663
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.132945535844484e-05,
+      "loss": 0.7163,
+      "step": 10664
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1328028936775311e-05,
+      "loss": 0.7429,
+      "step": 10665
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1326602487599012e-05,
+      "loss": 0.8452,
+      "step": 10666
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1325176010945487e-05,
+      "loss": 0.7598,
+      "step": 10667
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1323749506844282e-05,
+      "loss": 0.7852,
+      "step": 10668
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1322322975324944e-05,
+      "loss": 0.8428,
+      "step": 10669
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1320896416417026e-05,
+      "loss": 0.7504,
+      "step": 10670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1319469830150065e-05,
+      "loss": 0.7319,
+      "step": 10671
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1318043216553616e-05,
+      "loss": 0.8018,
+      "step": 10672
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1316616575657223e-05,
+      "loss": 0.7837,
+      "step": 10673
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.131518990749044e-05,
+      "loss": 0.8452,
+      "step": 10674
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1313763212082814e-05,
+      "loss": 0.7842,
+      "step": 10675
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1312336489463896e-05,
+      "loss": 0.833,
+      "step": 10676
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1310909739663239e-05,
+      "loss": 0.8354,
+      "step": 10677
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.130948296271039e-05,
+      "loss": 0.8628,
+      "step": 10678
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1308056158634906e-05,
+      "loss": 0.7542,
+      "step": 10679
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1306629327466334e-05,
+      "loss": 0.7488,
+      "step": 10680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1305202469234234e-05,
+      "loss": 0.8164,
+      "step": 10681
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1303775583968155e-05,
+      "loss": 0.7725,
+      "step": 10682
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1302348671697652e-05,
+      "loss": 0.7769,
+      "step": 10683
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1300921732452283e-05,
+      "loss": 0.7117,
+      "step": 10684
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1299494766261601e-05,
+      "loss": 0.7314,
+      "step": 10685
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1298067773155161e-05,
+      "loss": 0.8223,
+      "step": 10686
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.129664075316252e-05,
+      "loss": 0.7031,
+      "step": 10687
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1295213706313236e-05,
+      "loss": 0.8169,
+      "step": 10688
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1293786632636867e-05,
+      "loss": 0.6733,
+      "step": 10689
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.129235953216297e-05,
+      "loss": 0.7837,
+      "step": 10690
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1290932404921107e-05,
+      "loss": 0.8633,
+      "step": 10691
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.128950525094083e-05,
+      "loss": 0.7749,
+      "step": 10692
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1288078070251708e-05,
+      "loss": 0.7993,
+      "step": 10693
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1286650862883293e-05,
+      "loss": 0.6895,
+      "step": 10694
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1285223628865154e-05,
+      "loss": 0.8276,
+      "step": 10695
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1283796368226846e-05,
+      "loss": 0.7021,
+      "step": 10696
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1282369080997936e-05,
+      "loss": 0.7891,
+      "step": 10697
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1280941767207984e-05,
+      "loss": 0.7966,
+      "step": 10698
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1279514426886556e-05,
+      "loss": 0.6108,
+      "step": 10699
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.127808706006321e-05,
+      "loss": 0.7969,
+      "step": 10700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1276659666767515e-05,
+      "loss": 0.7954,
+      "step": 10701
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1275232247029034e-05,
+      "loss": 0.8169,
+      "step": 10702
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1273804800877332e-05,
+      "loss": 0.7683,
+      "step": 10703
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1272377328341974e-05,
+      "loss": 0.748,
+      "step": 10704
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1270949829452533e-05,
+      "loss": 0.7373,
+      "step": 10705
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.126952230423857e-05,
+      "loss": 0.6831,
+      "step": 10706
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1268094752729652e-05,
+      "loss": 0.7378,
+      "step": 10707
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1266667174955353e-05,
+      "loss": 0.7061,
+      "step": 10708
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1265239570945235e-05,
+      "loss": 0.7583,
+      "step": 10709
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1263811940728871e-05,
+      "loss": 0.7441,
+      "step": 10710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1262384284335828e-05,
+      "loss": 0.7349,
+      "step": 10711
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1260956601795681e-05,
+      "loss": 0.804,
+      "step": 10712
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1259528893137997e-05,
+      "loss": 0.7744,
+      "step": 10713
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.125810115839235e-05,
+      "loss": 0.844,
+      "step": 10714
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1256673397588307e-05,
+      "loss": 0.7861,
+      "step": 10715
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1255245610755446e-05,
+      "loss": 0.7319,
+      "step": 10716
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1253817797923334e-05,
+      "loss": 0.6711,
+      "step": 10717
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1252389959121553e-05,
+      "loss": 0.8062,
+      "step": 10718
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.125096209437967e-05,
+      "loss": 0.752,
+      "step": 10719
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1249534203727261e-05,
+      "loss": 0.7485,
+      "step": 10720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1248106287193903e-05,
+      "loss": 0.8491,
+      "step": 10721
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1246678344809173e-05,
+      "loss": 0.7244,
+      "step": 10722
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1245250376602643e-05,
+      "loss": 0.77,
+      "step": 10723
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.124382238260389e-05,
+      "loss": 0.7485,
+      "step": 10724
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1242394362842497e-05,
+      "loss": 0.7476,
+      "step": 10725
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1240966317348036e-05,
+      "loss": 0.7661,
+      "step": 10726
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1239538246150087e-05,
+      "loss": 0.8257,
+      "step": 10727
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.123811014927823e-05,
+      "loss": 0.686,
+      "step": 10728
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1236682026762044e-05,
+      "loss": 0.8379,
+      "step": 10729
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1235253878631107e-05,
+      "loss": 0.7749,
+      "step": 10730
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1233825704915002e-05,
+      "loss": 0.7524,
+      "step": 10731
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1232397505643309e-05,
+      "loss": 0.689,
+      "step": 10732
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.123096928084561e-05,
+      "loss": 0.8367,
+      "step": 10733
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1229541030551488e-05,
+      "loss": 0.7725,
+      "step": 10734
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1228112754790522e-05,
+      "loss": 0.6709,
+      "step": 10735
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1226684453592298e-05,
+      "loss": 0.7563,
+      "step": 10736
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1225256126986399e-05,
+      "loss": 0.7307,
+      "step": 10737
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1223827775002409e-05,
+      "loss": 0.7949,
+      "step": 10738
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1222399397669912e-05,
+      "loss": 0.8047,
+      "step": 10739
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1220970995018496e-05,
+      "loss": 0.7568,
+      "step": 10740
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1219542567077746e-05,
+      "loss": 0.6956,
+      "step": 10741
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1218114113877246e-05,
+      "loss": 0.791,
+      "step": 10742
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1216685635446583e-05,
+      "loss": 0.6306,
+      "step": 10743
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1215257131815345e-05,
+      "loss": 0.7424,
+      "step": 10744
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1213828603013122e-05,
+      "loss": 0.7861,
+      "step": 10745
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.12124000490695e-05,
+      "loss": 0.6492,
+      "step": 10746
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.121097147001407e-05,
+      "loss": 0.8071,
+      "step": 10747
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1209542865876416e-05,
+      "loss": 0.8818,
+      "step": 10748
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1208114236686136e-05,
+      "loss": 0.7576,
+      "step": 10749
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1206685582472814e-05,
+      "loss": 0.7749,
+      "step": 10750
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1205256903266044e-05,
+      "loss": 0.6858,
+      "step": 10751
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1203828199095417e-05,
+      "loss": 0.7732,
+      "step": 10752
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1202399469990524e-05,
+      "loss": 0.8179,
+      "step": 10753
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.120097071598096e-05,
+      "loss": 0.7722,
+      "step": 10754
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1199541937096313e-05,
+      "loss": 0.7866,
+      "step": 10755
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1198113133366182e-05,
+      "loss": 0.7781,
+      "step": 10756
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1196684304820156e-05,
+      "loss": 0.8755,
+      "step": 10757
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1195255451487837e-05,
+      "loss": 0.6934,
+      "step": 10758
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1193826573398814e-05,
+      "loss": 0.7075,
+      "step": 10759
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1192397670582684e-05,
+      "loss": 0.7148,
+      "step": 10760
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1190968743069044e-05,
+      "loss": 0.7205,
+      "step": 10761
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1189539790887489e-05,
+      "loss": 0.6516,
+      "step": 10762
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1188110814067617e-05,
+      "loss": 0.7046,
+      "step": 10763
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1186681812639026e-05,
+      "loss": 0.7104,
+      "step": 10764
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1185252786631316e-05,
+      "loss": 0.8086,
+      "step": 10765
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.118382373607408e-05,
+      "loss": 0.7197,
+      "step": 10766
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1182394660996923e-05,
+      "loss": 0.8394,
+      "step": 10767
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1180965561429442e-05,
+      "loss": 0.814,
+      "step": 10768
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1179536437401235e-05,
+      "loss": 0.812,
+      "step": 10769
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1178107288941906e-05,
+      "loss": 0.769,
+      "step": 10770
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1176678116081058e-05,
+      "loss": 0.7446,
+      "step": 10771
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1175248918848289e-05,
+      "loss": 0.7036,
+      "step": 10772
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1173819697273206e-05,
+      "loss": 0.772,
+      "step": 10773
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1172390451385407e-05,
+      "loss": 0.7651,
+      "step": 10774
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1170961181214493e-05,
+      "loss": 0.8071,
+      "step": 10775
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1169531886790072e-05,
+      "loss": 0.7241,
+      "step": 10776
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.116810256814175e-05,
+      "loss": 0.7031,
+      "step": 10777
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1166673225299128e-05,
+      "loss": 0.665,
+      "step": 10778
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.116524385829181e-05,
+      "loss": 0.7969,
+      "step": 10779
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1163814467149405e-05,
+      "loss": 0.7415,
+      "step": 10780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1162385051901516e-05,
+      "loss": 0.8362,
+      "step": 10781
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1160955612577755e-05,
+      "loss": 0.7217,
+      "step": 10782
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1159526149207727e-05,
+      "loss": 0.8135,
+      "step": 10783
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1158096661821038e-05,
+      "loss": 0.7861,
+      "step": 10784
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1156667150447298e-05,
+      "loss": 0.7959,
+      "step": 10785
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1155237615116115e-05,
+      "loss": 0.7949,
+      "step": 10786
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1153808055857096e-05,
+      "loss": 0.7349,
+      "step": 10787
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1152378472699853e-05,
+      "loss": 0.7876,
+      "step": 10788
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1150948865674003e-05,
+      "loss": 0.689,
+      "step": 10789
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1149519234809143e-05,
+      "loss": 0.8018,
+      "step": 10790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1148089580134892e-05,
+      "loss": 0.7832,
+      "step": 10791
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.114665990168086e-05,
+      "loss": 0.7639,
+      "step": 10792
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1145230199476662e-05,
+      "loss": 0.8257,
+      "step": 10793
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1143800473551904e-05,
+      "loss": 0.8301,
+      "step": 10794
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1142370723936209e-05,
+      "loss": 0.7578,
+      "step": 10795
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1140940950659183e-05,
+      "loss": 0.9248,
+      "step": 10796
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1139511153750441e-05,
+      "loss": 0.7642,
+      "step": 10797
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1138081333239603e-05,
+      "loss": 0.7568,
+      "step": 10798
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1136651489156278e-05,
+      "loss": 0.6819,
+      "step": 10799
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1135221621530083e-05,
+      "loss": 0.8096,
+      "step": 10800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1133791730390638e-05,
+      "loss": 0.7568,
+      "step": 10801
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1132361815767554e-05,
+      "loss": 0.8628,
+      "step": 10802
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1130931877690455e-05,
+      "loss": 0.6973,
+      "step": 10803
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1129501916188949e-05,
+      "loss": 0.5806,
+      "step": 10804
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1128071931292661e-05,
+      "loss": 0.7607,
+      "step": 10805
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1126641923031205e-05,
+      "loss": 0.6724,
+      "step": 10806
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1125211891434206e-05,
+      "loss": 0.8271,
+      "step": 10807
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.112378183653128e-05,
+      "loss": 0.686,
+      "step": 10808
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1122351758352048e-05,
+      "loss": 0.7256,
+      "step": 10809
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.112092165692613e-05,
+      "loss": 0.7749,
+      "step": 10810
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1119491532283144e-05,
+      "loss": 0.7583,
+      "step": 10811
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1118061384452716e-05,
+      "loss": 0.8735,
+      "step": 10812
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1116631213464465e-05,
+      "loss": 0.7539,
+      "step": 10813
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1115201019348014e-05,
+      "loss": 0.6929,
+      "step": 10814
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1113770802132988e-05,
+      "loss": 0.7524,
+      "step": 10815
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1112340561849009e-05,
+      "loss": 0.8208,
+      "step": 10816
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.11109102985257e-05,
+      "loss": 0.7278,
+      "step": 10817
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1109480012192683e-05,
+      "loss": 0.7815,
+      "step": 10818
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1108049702879588e-05,
+      "loss": 0.8071,
+      "step": 10819
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1106619370616039e-05,
+      "loss": 0.7964,
+      "step": 10820
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.110518901543166e-05,
+      "loss": 0.738,
+      "step": 10821
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1103758637356074e-05,
+      "loss": 0.6296,
+      "step": 10822
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1102328236418915e-05,
+      "loss": 0.9487,
+      "step": 10823
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1100897812649804e-05,
+      "loss": 0.6855,
+      "step": 10824
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1099467366078374e-05,
+      "loss": 0.8066,
+      "step": 10825
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1098036896734248e-05,
+      "loss": 0.7686,
+      "step": 10826
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.109660640464706e-05,
+      "loss": 0.7471,
+      "step": 10827
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1095175889846433e-05,
+      "loss": 0.7346,
+      "step": 10828
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1093745352361998e-05,
+      "loss": 0.6858,
+      "step": 10829
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1092314792223389e-05,
+      "loss": 0.8149,
+      "step": 10830
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1090884209460233e-05,
+      "loss": 0.7505,
+      "step": 10831
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.108945360410216e-05,
+      "loss": 0.6958,
+      "step": 10832
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1088022976178805e-05,
+      "loss": 0.8418,
+      "step": 10833
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1086592325719797e-05,
+      "loss": 0.8135,
+      "step": 10834
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.108516165275477e-05,
+      "loss": 0.6775,
+      "step": 10835
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1083730957313356e-05,
+      "loss": 0.7773,
+      "step": 10836
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1082300239425188e-05,
+      "loss": 0.8164,
+      "step": 10837
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.10808694991199e-05,
+      "loss": 0.7588,
+      "step": 10838
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1079438736427126e-05,
+      "loss": 0.7134,
+      "step": 10839
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1078007951376502e-05,
+      "loss": 0.8242,
+      "step": 10840
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.107657714399766e-05,
+      "loss": 0.7666,
+      "step": 10841
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1075146314320238e-05,
+      "loss": 0.7095,
+      "step": 10842
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1073715462373871e-05,
+      "loss": 0.7505,
+      "step": 10843
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1072284588188195e-05,
+      "loss": 0.7729,
+      "step": 10844
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.107085369179285e-05,
+      "loss": 0.79,
+      "step": 10845
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1069422773217468e-05,
+      "loss": 0.6492,
+      "step": 10846
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1067991832491694e-05,
+      "loss": 0.7769,
+      "step": 10847
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1066560869645165e-05,
+      "loss": 0.7061,
+      "step": 10848
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1065129884707514e-05,
+      "loss": 0.8418,
+      "step": 10849
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1063698877708383e-05,
+      "loss": 0.7241,
+      "step": 10850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1062267848677413e-05,
+      "loss": 0.7676,
+      "step": 10851
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1060836797644241e-05,
+      "loss": 0.7432,
+      "step": 10852
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1059405724638512e-05,
+      "loss": 0.895,
+      "step": 10853
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1057974629689867e-05,
+      "loss": 0.7959,
+      "step": 10854
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1056543512827942e-05,
+      "loss": 0.6902,
+      "step": 10855
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1055112374082385e-05,
+      "loss": 0.7583,
+      "step": 10856
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1053681213482837e-05,
+      "loss": 0.6968,
+      "step": 10857
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1052250031058935e-05,
+      "loss": 0.8047,
+      "step": 10858
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.105081882684033e-05,
+      "loss": 0.7075,
+      "step": 10859
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1049387600856664e-05,
+      "loss": 0.6968,
+      "step": 10860
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.104795635313758e-05,
+      "loss": 0.8096,
+      "step": 10861
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1046525083712722e-05,
+      "loss": 0.6736,
+      "step": 10862
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1045093792611739e-05,
+      "loss": 0.644,
+      "step": 10863
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1043662479864272e-05,
+      "loss": 0.7849,
+      "step": 10864
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.104223114549997e-05,
+      "loss": 0.7651,
+      "step": 10865
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1040799789548475e-05,
+      "loss": 0.6914,
+      "step": 10866
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.103936841203944e-05,
+      "loss": 0.6934,
+      "step": 10867
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1037937013002507e-05,
+      "loss": 0.7778,
+      "step": 10868
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1036505592467326e-05,
+      "loss": 0.7375,
+      "step": 10869
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1035074150463547e-05,
+      "loss": 0.7368,
+      "step": 10870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1033642687020816e-05,
+      "loss": 0.7871,
+      "step": 10871
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1032211202168785e-05,
+      "loss": 0.6868,
+      "step": 10872
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1030779695937101e-05,
+      "loss": 0.707,
+      "step": 10873
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1029348168355416e-05,
+      "loss": 0.7959,
+      "step": 10874
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.102791661945338e-05,
+      "loss": 0.7524,
+      "step": 10875
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1026485049260642e-05,
+      "loss": 0.8555,
+      "step": 10876
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1025053457806856e-05,
+      "loss": 0.7725,
+      "step": 10877
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1023621845121675e-05,
+      "loss": 0.7554,
+      "step": 10878
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1022190211234747e-05,
+      "loss": 0.7129,
+      "step": 10879
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1020758556175726e-05,
+      "loss": 0.7397,
+      "step": 10880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1019326879974266e-05,
+      "loss": 0.7866,
+      "step": 10881
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1017895182660019e-05,
+      "loss": 0.8105,
+      "step": 10882
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1016463464262639e-05,
+      "loss": 0.7637,
+      "step": 10883
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1015031724811785e-05,
+      "loss": 0.7607,
+      "step": 10884
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1013599964337107e-05,
+      "loss": 0.832,
+      "step": 10885
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.101216818286826e-05,
+      "loss": 0.8286,
+      "step": 10886
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1010736380434905e-05,
+      "loss": 0.7622,
+      "step": 10887
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1009304557066695e-05,
+      "loss": 0.8179,
+      "step": 10888
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1007872712793286e-05,
+      "loss": 0.6931,
+      "step": 10889
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1006440847644335e-05,
+      "loss": 0.6919,
+      "step": 10890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.10050089616495e-05,
+      "loss": 0.7725,
+      "step": 10891
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1003577054838439e-05,
+      "loss": 0.7588,
+      "step": 10892
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.100214512724081e-05,
+      "loss": 0.7202,
+      "step": 10893
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1000713178886272e-05,
+      "loss": 0.7295,
+      "step": 10894
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0999281209804482e-05,
+      "loss": 0.7959,
+      "step": 10895
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0997849220025105e-05,
+      "loss": 0.8525,
+      "step": 10896
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0996417209577796e-05,
+      "loss": 0.9155,
+      "step": 10897
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0994985178492218e-05,
+      "loss": 0.7148,
+      "step": 10898
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0993553126798031e-05,
+      "loss": 0.708,
+      "step": 10899
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0992121054524898e-05,
+      "loss": 0.8828,
+      "step": 10900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0990688961702479e-05,
+      "loss": 0.7798,
+      "step": 10901
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0989256848360434e-05,
+      "loss": 0.6843,
+      "step": 10902
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.098782471452843e-05,
+      "loss": 0.7781,
+      "step": 10903
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0986392560236128e-05,
+      "loss": 0.813,
+      "step": 10904
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0984960385513193e-05,
+      "loss": 0.7615,
+      "step": 10905
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0983528190389285e-05,
+      "loss": 0.7729,
+      "step": 10906
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0982095974894074e-05,
+      "loss": 0.7798,
+      "step": 10907
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0980663739057218e-05,
+      "loss": 0.7827,
+      "step": 10908
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0979231482908387e-05,
+      "loss": 0.8511,
+      "step": 10909
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0977799206477246e-05,
+      "loss": 0.7983,
+      "step": 10910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0976366909793461e-05,
+      "loss": 0.7031,
+      "step": 10911
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0974934592886697e-05,
+      "loss": 0.8672,
+      "step": 10912
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0973502255786621e-05,
+      "loss": 0.7988,
+      "step": 10913
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.09720698985229e-05,
+      "loss": 0.7725,
+      "step": 10914
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0970637521125206e-05,
+      "loss": 0.7227,
+      "step": 10915
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0969205123623202e-05,
+      "loss": 0.8843,
+      "step": 10916
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0967772706046555e-05,
+      "loss": 0.7498,
+      "step": 10917
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.096634026842494e-05,
+      "loss": 0.7568,
+      "step": 10918
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0964907810788024e-05,
+      "loss": 0.7725,
+      "step": 10919
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0963475333165474e-05,
+      "loss": 0.873,
+      "step": 10920
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0962042835586963e-05,
+      "loss": 0.749,
+      "step": 10921
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0960610318082161e-05,
+      "loss": 0.8506,
+      "step": 10922
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0959177780680739e-05,
+      "loss": 0.7749,
+      "step": 10923
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0957745223412368e-05,
+      "loss": 0.7583,
+      "step": 10924
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.095631264630672e-05,
+      "loss": 0.7798,
+      "step": 10925
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0954880049393465e-05,
+      "loss": 0.7705,
+      "step": 10926
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.095344743270228e-05,
+      "loss": 0.7632,
+      "step": 10927
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0952014796262837e-05,
+      "loss": 0.7871,
+      "step": 10928
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0950582140104807e-05,
+      "loss": 0.8267,
+      "step": 10929
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0949149464257865e-05,
+      "loss": 0.7896,
+      "step": 10930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0947716768751686e-05,
+      "loss": 0.7139,
+      "step": 10931
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0946284053615944e-05,
+      "loss": 0.7158,
+      "step": 10932
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0944851318880314e-05,
+      "loss": 0.7886,
+      "step": 10933
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0943418564574473e-05,
+      "loss": 0.7039,
+      "step": 10934
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.094198579072809e-05,
+      "loss": 0.8062,
+      "step": 10935
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0940552997370852e-05,
+      "loss": 0.7476,
+      "step": 10936
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0939120184532434e-05,
+      "loss": 0.7905,
+      "step": 10937
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0937687352242505e-05,
+      "loss": 0.793,
+      "step": 10938
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0936254500530747e-05,
+      "loss": 0.8003,
+      "step": 10939
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0934821629426839e-05,
+      "loss": 0.7017,
+      "step": 10940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.093338873896046e-05,
+      "loss": 0.7388,
+      "step": 10941
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0931955829161286e-05,
+      "loss": 0.7427,
+      "step": 10942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0930522900058997e-05,
+      "loss": 0.7793,
+      "step": 10943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0929089951683272e-05,
+      "loss": 0.8999,
+      "step": 10944
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0927656984063793e-05,
+      "loss": 0.7856,
+      "step": 10945
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0926223997230239e-05,
+      "loss": 0.7039,
+      "step": 10946
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.092479099121229e-05,
+      "loss": 0.7495,
+      "step": 10947
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0923357966039625e-05,
+      "loss": 0.7446,
+      "step": 10948
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0921924921741931e-05,
+      "loss": 0.7119,
+      "step": 10949
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0920491858348888e-05,
+      "loss": 0.7471,
+      "step": 10950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0919058775890176e-05,
+      "loss": 0.7603,
+      "step": 10951
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0917625674395485e-05,
+      "loss": 0.7925,
+      "step": 10952
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0916192553894487e-05,
+      "loss": 0.6738,
+      "step": 10953
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0914759414416872e-05,
+      "loss": 0.7092,
+      "step": 10954
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.091332625599232e-05,
+      "loss": 0.7495,
+      "step": 10955
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0911893078650521e-05,
+      "loss": 0.6689,
+      "step": 10956
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0910459882421157e-05,
+      "loss": 0.7559,
+      "step": 10957
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.090902666733391e-05,
+      "loss": 0.7656,
+      "step": 10958
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0907593433418468e-05,
+      "loss": 0.7161,
+      "step": 10959
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0906160180704514e-05,
+      "loss": 0.7095,
+      "step": 10960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0904726909221744e-05,
+      "loss": 0.8496,
+      "step": 10961
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0903293618999833e-05,
+      "loss": 0.7112,
+      "step": 10962
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0901860310068476e-05,
+      "loss": 0.7295,
+      "step": 10963
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0900426982457355e-05,
+      "loss": 0.7734,
+      "step": 10964
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0898993636196158e-05,
+      "loss": 0.7876,
+      "step": 10965
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0897560271314578e-05,
+      "loss": 0.698,
+      "step": 10966
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0896126887842302e-05,
+      "loss": 0.8105,
+      "step": 10967
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0894693485809016e-05,
+      "loss": 0.7461,
+      "step": 10968
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0893260065244409e-05,
+      "loss": 0.7356,
+      "step": 10969
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0891826626178172e-05,
+      "loss": 0.7192,
+      "step": 10970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0890393168639998e-05,
+      "loss": 0.7056,
+      "step": 10971
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0888959692659571e-05,
+      "loss": 0.7788,
+      "step": 10972
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0887526198266589e-05,
+      "loss": 0.8027,
+      "step": 10973
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0886092685490739e-05,
+      "loss": 0.8464,
+      "step": 10974
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0884659154361714e-05,
+      "loss": 0.7788,
+      "step": 10975
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0883225604909207e-05,
+      "loss": 0.7026,
+      "step": 10976
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0881792037162908e-05,
+      "loss": 0.7646,
+      "step": 10977
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0880358451152511e-05,
+      "loss": 0.7212,
+      "step": 10978
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0878924846907707e-05,
+      "loss": 0.7905,
+      "step": 10979
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0877491224458192e-05,
+      "loss": 0.7632,
+      "step": 10980
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.087605758383366e-05,
+      "loss": 0.6865,
+      "step": 10981
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0874623925063808e-05,
+      "loss": 0.678,
+      "step": 10982
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0873190248178323e-05,
+      "loss": 0.8174,
+      "step": 10983
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0871756553206902e-05,
+      "loss": 0.7446,
+      "step": 10984
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0870322840179243e-05,
+      "loss": 0.7422,
+      "step": 10985
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0868889109125045e-05,
+      "loss": 0.769,
+      "step": 10986
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0867455360073998e-05,
+      "loss": 0.8091,
+      "step": 10987
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0866021593055801e-05,
+      "loss": 0.7617,
+      "step": 10988
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.086458780810015e-05,
+      "loss": 0.7708,
+      "step": 10989
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0863154005236744e-05,
+      "loss": 0.8354,
+      "step": 10990
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0861720184495281e-05,
+      "loss": 0.748,
+      "step": 10991
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0860286345905455e-05,
+      "loss": 0.77,
+      "step": 10992
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.085885248949697e-05,
+      "loss": 0.811,
+      "step": 10993
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0857418615299518e-05,
+      "loss": 0.7766,
+      "step": 10994
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0855984723342803e-05,
+      "loss": 0.7891,
+      "step": 10995
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0854550813656523e-05,
+      "loss": 0.7461,
+      "step": 10996
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0853116886270378e-05,
+      "loss": 0.8008,
+      "step": 10997
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0851682941214069e-05,
+      "loss": 0.7402,
+      "step": 10998
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0850248978517295e-05,
+      "loss": 0.8018,
+      "step": 10999
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0848814998209758e-05,
+      "loss": 0.7969,
+      "step": 11000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0847381000321158e-05,
+      "loss": 0.7866,
+      "step": 11001
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.08459469848812e-05,
+      "loss": 0.7026,
+      "step": 11002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0844512951919579e-05,
+      "loss": 0.7954,
+      "step": 11003
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0843078901466006e-05,
+      "loss": 0.7134,
+      "step": 11004
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0841644833550178e-05,
+      "loss": 0.7036,
+      "step": 11005
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0840210748201801e-05,
+      "loss": 0.8081,
+      "step": 11006
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0838776645450574e-05,
+      "loss": 0.7363,
+      "step": 11007
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0837342525326209e-05,
+      "loss": 0.7437,
+      "step": 11008
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0835908387858403e-05,
+      "loss": 0.6306,
+      "step": 11009
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0834474233076862e-05,
+      "loss": 0.7258,
+      "step": 11010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.083304006101129e-05,
+      "loss": 0.7705,
+      "step": 11011
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0831605871691398e-05,
+      "loss": 0.749,
+      "step": 11012
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0830171665146885e-05,
+      "loss": 0.7964,
+      "step": 11013
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.082873744140746e-05,
+      "loss": 0.7346,
+      "step": 11014
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.082730320050283e-05,
+      "loss": 0.7656,
+      "step": 11015
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0825868942462701e-05,
+      "loss": 0.8018,
+      "step": 11016
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0824434667316776e-05,
+      "loss": 0.8267,
+      "step": 11017
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.082300037509477e-05,
+      "loss": 0.6135,
+      "step": 11018
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0821566065826383e-05,
+      "loss": 0.8799,
+      "step": 11019
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.082013173954133e-05,
+      "loss": 0.7612,
+      "step": 11020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0818697396269315e-05,
+      "loss": 0.7734,
+      "step": 11021
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0817263036040048e-05,
+      "loss": 0.7036,
+      "step": 11022
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0815828658883239e-05,
+      "loss": 0.7925,
+      "step": 11023
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0814394264828596e-05,
+      "loss": 0.7534,
+      "step": 11024
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0812959853905826e-05,
+      "loss": 0.7632,
+      "step": 11025
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0811525426144649e-05,
+      "loss": 0.7007,
+      "step": 11026
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0810090981574767e-05,
+      "loss": 0.8525,
+      "step": 11027
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0808656520225895e-05,
+      "loss": 0.7388,
+      "step": 11028
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.080722204212774e-05,
+      "loss": 0.6748,
+      "step": 11029
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0805787547310016e-05,
+      "loss": 0.7324,
+      "step": 11030
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0804353035802437e-05,
+      "loss": 0.6812,
+      "step": 11031
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0802918507634711e-05,
+      "loss": 0.8237,
+      "step": 11032
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0801483962836554e-05,
+      "loss": 0.8145,
+      "step": 11033
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.080004940143768e-05,
+      "loss": 0.7769,
+      "step": 11034
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0798614823467796e-05,
+      "loss": 0.8062,
+      "step": 11035
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0797180228956623e-05,
+      "loss": 0.6819,
+      "step": 11036
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0795745617933871e-05,
+      "loss": 0.9053,
+      "step": 11037
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0794310990429256e-05,
+      "loss": 0.7271,
+      "step": 11038
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.079287634647249e-05,
+      "loss": 0.7549,
+      "step": 11039
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0791441686093291e-05,
+      "loss": 0.7544,
+      "step": 11040
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0790007009321377e-05,
+      "loss": 0.8599,
+      "step": 11041
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0788572316186456e-05,
+      "loss": 0.7925,
+      "step": 11042
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.078713760671825e-05,
+      "loss": 0.7529,
+      "step": 11043
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0785702880946472e-05,
+      "loss": 0.7466,
+      "step": 11044
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0784268138900842e-05,
+      "loss": 0.8047,
+      "step": 11045
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0782833380611073e-05,
+      "loss": 0.7578,
+      "step": 11046
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0781398606106888e-05,
+      "loss": 0.6812,
+      "step": 11047
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0779963815417997e-05,
+      "loss": 0.8159,
+      "step": 11048
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0778529008574123e-05,
+      "loss": 0.9131,
+      "step": 11049
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0777094185604986e-05,
+      "loss": 0.8521,
+      "step": 11050
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0775659346540303e-05,
+      "loss": 0.8228,
+      "step": 11051
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0774224491409793e-05,
+      "loss": 0.751,
+      "step": 11052
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0772789620243173e-05,
+      "loss": 0.7373,
+      "step": 11053
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0771354733070167e-05,
+      "loss": 0.6113,
+      "step": 11054
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0769919829920492e-05,
+      "loss": 0.7847,
+      "step": 11055
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0768484910823873e-05,
+      "loss": 0.791,
+      "step": 11056
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0767049975810024e-05,
+      "loss": 0.8374,
+      "step": 11057
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076561502490867e-05,
+      "loss": 0.7993,
+      "step": 11058
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076418005814953e-05,
+      "loss": 0.8057,
+      "step": 11059
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0762745075562326e-05,
+      "loss": 0.8223,
+      "step": 11060
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076131007717678e-05,
+      "loss": 0.7009,
+      "step": 11061
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.075987506302262e-05,
+      "loss": 0.7949,
+      "step": 11062
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0758440033129562e-05,
+      "loss": 0.7974,
+      "step": 11063
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0757004987527332e-05,
+      "loss": 0.7397,
+      "step": 11064
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0755569926245655e-05,
+      "loss": 0.8027,
+      "step": 11065
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.075413484931425e-05,
+      "loss": 0.7302,
+      "step": 11066
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0752699756762845e-05,
+      "loss": 0.8154,
+      "step": 11067
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0751264648621161e-05,
+      "loss": 0.7212,
+      "step": 11068
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0749829524918927e-05,
+      "loss": 0.7332,
+      "step": 11069
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0748394385685863e-05,
+      "loss": 0.7192,
+      "step": 11070
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0746959230951699e-05,
+      "loss": 0.6606,
+      "step": 11071
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0745524060746155e-05,
+      "loss": 0.7493,
+      "step": 11072
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0744088875098964e-05,
+      "loss": 0.7881,
+      "step": 11073
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0742653674039842e-05,
+      "loss": 0.6736,
+      "step": 11074
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0741218457598527e-05,
+      "loss": 0.7566,
+      "step": 11075
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073978322580474e-05,
+      "loss": 0.8213,
+      "step": 11076
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0738347978688209e-05,
+      "loss": 0.7693,
+      "step": 11077
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0736912716278662e-05,
+      "loss": 0.8052,
+      "step": 11078
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0735477438605826e-05,
+      "loss": 0.7295,
+      "step": 11079
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073404214569943e-05,
+      "loss": 0.7964,
+      "step": 11080
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0732606837589202e-05,
+      "loss": 0.8311,
+      "step": 11081
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0731171514304872e-05,
+      "loss": 0.7344,
+      "step": 11082
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0729736175876166e-05,
+      "loss": 0.7637,
+      "step": 11083
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0728300822332816e-05,
+      "loss": 0.7676,
+      "step": 11084
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0726865453704552e-05,
+      "loss": 0.7703,
+      "step": 11085
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0725430070021103e-05,
+      "loss": 0.7876,
+      "step": 11086
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.07239946713122e-05,
+      "loss": 0.7251,
+      "step": 11087
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0722559257607572e-05,
+      "loss": 0.8013,
+      "step": 11088
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.072112382893695e-05,
+      "loss": 0.7585,
+      "step": 11089
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0719688385330069e-05,
+      "loss": 0.8164,
+      "step": 11090
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0718252926816657e-05,
+      "loss": 0.7334,
+      "step": 11091
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0716817453426447e-05,
+      "loss": 0.7568,
+      "step": 11092
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0715381965189172e-05,
+      "loss": 0.8447,
+      "step": 11093
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0713946462134564e-05,
+      "loss": 0.7539,
+      "step": 11094
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0712510944292354e-05,
+      "loss": 0.7168,
+      "step": 11095
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0711075411692276e-05,
+      "loss": 0.8521,
+      "step": 11096
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0709639864364065e-05,
+      "loss": 0.6997,
+      "step": 11097
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0708204302337454e-05,
+      "loss": 0.7837,
+      "step": 11098
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0706768725642175e-05,
+      "loss": 0.7275,
+      "step": 11099
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0705333134307968e-05,
+      "loss": 0.7419,
+      "step": 11100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0703897528364563e-05,
+      "loss": 0.8252,
+      "step": 11101
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0702461907841695e-05,
+      "loss": 0.771,
+      "step": 11102
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.07010262727691e-05,
+      "loss": 0.6514,
+      "step": 11103
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0699590623176513e-05,
+      "loss": 0.7832,
+      "step": 11104
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.069815495909367e-05,
+      "loss": 0.7949,
+      "step": 11105
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.069671928055031e-05,
+      "loss": 0.6372,
+      "step": 11106
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0695283587576167e-05,
+      "loss": 0.791,
+      "step": 11107
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0693847880200976e-05,
+      "loss": 0.7783,
+      "step": 11108
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0692412158454474e-05,
+      "loss": 0.8555,
+      "step": 11109
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0690976422366405e-05,
+      "loss": 0.7283,
+      "step": 11110
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.06895406719665e-05,
+      "loss": 0.7217,
+      "step": 11111
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0688104907284498e-05,
+      "loss": 0.8477,
+      "step": 11112
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.068666912835014e-05,
+      "loss": 0.7217,
+      "step": 11113
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.068523333519316e-05,
+      "loss": 0.7642,
+      "step": 11114
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0683797527843302e-05,
+      "loss": 0.8447,
+      "step": 11115
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0682361706330302e-05,
+      "loss": 0.7842,
+      "step": 11116
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0680925870683901e-05,
+      "loss": 0.8164,
+      "step": 11117
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0679490020933836e-05,
+      "loss": 0.7788,
+      "step": 11118
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0678054157109852e-05,
+      "loss": 0.7322,
+      "step": 11119
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0676618279241682e-05,
+      "loss": 0.7358,
+      "step": 11120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0675182387359073e-05,
+      "loss": 0.7056,
+      "step": 11121
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0673746481491762e-05,
+      "loss": 0.7329,
+      "step": 11122
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0672310561669494e-05,
+      "loss": 0.7153,
+      "step": 11123
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0670874627922006e-05,
+      "loss": 0.7317,
+      "step": 11124
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0669438680279044e-05,
+      "loss": 0.7627,
+      "step": 11125
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0668002718770345e-05,
+      "loss": 0.7842,
+      "step": 11126
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0666566743425655e-05,
+      "loss": 0.8062,
+      "step": 11127
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0665130754274717e-05,
+      "loss": 0.7754,
+      "step": 11128
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0663694751347274e-05,
+      "loss": 0.7585,
+      "step": 11129
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0662258734673068e-05,
+      "loss": 0.6885,
+      "step": 11130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.066082270428184e-05,
+      "loss": 0.665,
+      "step": 11131
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0659386660203339e-05,
+      "loss": 0.8145,
+      "step": 11132
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0657950602467303e-05,
+      "loss": 0.772,
+      "step": 11133
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0656514531103484e-05,
+      "loss": 0.7217,
+      "step": 11134
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0655078446141619e-05,
+      "loss": 0.7759,
+      "step": 11135
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0653642347611455e-05,
+      "loss": 0.8003,
+      "step": 11136
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.065220623554274e-05,
+      "loss": 0.7583,
+      "step": 11137
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0650770109965213e-05,
+      "loss": 0.8301,
+      "step": 11138
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0649333970908631e-05,
+      "loss": 0.7886,
+      "step": 11139
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.064789781840273e-05,
+      "loss": 0.7734,
+      "step": 11140
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.064646165247726e-05,
+      "loss": 0.8032,
+      "step": 11141
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0645025473161968e-05,
+      "loss": 0.8145,
+      "step": 11142
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.06435892804866e-05,
+      "loss": 0.7856,
+      "step": 11143
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.06421530744809e-05,
+      "loss": 0.7505,
+      "step": 11144
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0640716855174626e-05,
+      "loss": 0.7703,
+      "step": 11145
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0639280622597512e-05,
+      "loss": 0.812,
+      "step": 11146
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0637844376779314e-05,
+      "loss": 0.7351,
+      "step": 11147
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0636408117749778e-05,
+      "loss": 0.7446,
+      "step": 11148
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0634971845538654e-05,
+      "loss": 0.7373,
+      "step": 11149
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0633535560175685e-05,
+      "loss": 0.6765,
+      "step": 11150
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0632099261690627e-05,
+      "loss": 0.6445,
+      "step": 11151
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0630662950113226e-05,
+      "loss": 0.7954,
+      "step": 11152
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0629226625473234e-05,
+      "loss": 0.73,
+      "step": 11153
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0627790287800399e-05,
+      "loss": 0.7688,
+      "step": 11154
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0626353937124471e-05,
+      "loss": 0.7297,
+      "step": 11155
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0624917573475202e-05,
+      "loss": 0.7803,
+      "step": 11156
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0623481196882342e-05,
+      "loss": 0.7676,
+      "step": 11157
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.062204480737564e-05,
+      "loss": 0.7092,
+      "step": 11158
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0620608404984846e-05,
+      "loss": 0.7905,
+      "step": 11159
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.061917198973972e-05,
+      "loss": 0.8638,
+      "step": 11160
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0617735561670004e-05,
+      "loss": 0.7043,
+      "step": 11161
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0616299120805456e-05,
+      "loss": 0.7407,
+      "step": 11162
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0614862667175821e-05,
+      "loss": 0.7869,
+      "step": 11163
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.061342620081086e-05,
+      "loss": 0.793,
+      "step": 11164
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0611989721740321e-05,
+      "loss": 0.7563,
+      "step": 11165
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0610553229993958e-05,
+      "loss": 0.811,
+      "step": 11166
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0609116725601525e-05,
+      "loss": 0.7817,
+      "step": 11167
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0607680208592777e-05,
+      "loss": 0.7061,
+      "step": 11168
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0606243678997463e-05,
+      "loss": 0.7837,
+      "step": 11169
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0604807136845343e-05,
+      "loss": 0.7361,
+      "step": 11170
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0603370582166165e-05,
+      "loss": 0.7666,
+      "step": 11171
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.060193401498969e-05,
+      "loss": 0.73,
+      "step": 11172
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.060049743534567e-05,
+      "loss": 0.7273,
+      "step": 11173
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0599060843263858e-05,
+      "loss": 0.7651,
+      "step": 11174
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0597624238774011e-05,
+      "loss": 0.7075,
+      "step": 11175
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0596187621905888e-05,
+      "loss": 0.8228,
+      "step": 11176
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.059475099268924e-05,
+      "loss": 0.8232,
+      "step": 11177
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0593314351153825e-05,
+      "loss": 0.7651,
+      "step": 11178
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0591877697329401e-05,
+      "loss": 0.7891,
+      "step": 11179
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0590441031245719e-05,
+      "loss": 0.7627,
+      "step": 11180
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0589004352932543e-05,
+      "loss": 0.6885,
+      "step": 11181
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.058756766241963e-05,
+      "loss": 0.7974,
+      "step": 11182
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0586130959736731e-05,
+      "loss": 0.7827,
+      "step": 11183
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0584694244913608e-05,
+      "loss": 0.7886,
+      "step": 11184
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0583257517980019e-05,
+      "loss": 0.8223,
+      "step": 11185
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0581820778965719e-05,
+      "loss": 0.7495,
+      "step": 11186
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0580384027900472e-05,
+      "loss": 0.8369,
+      "step": 11187
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0578947264814033e-05,
+      "loss": 0.7,
+      "step": 11188
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0577510489736158e-05,
+      "loss": 0.8003,
+      "step": 11189
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0576073702696612e-05,
+      "loss": 0.751,
+      "step": 11190
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0574636903725153e-05,
+      "loss": 0.7773,
+      "step": 11191
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0573200092851539e-05,
+      "loss": 0.8081,
+      "step": 11192
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.057176327010553e-05,
+      "loss": 0.7935,
+      "step": 11193
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.057032643551689e-05,
+      "loss": 0.749,
+      "step": 11194
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0568889589115372e-05,
+      "loss": 0.7319,
+      "step": 11195
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0567452730930743e-05,
+      "loss": 0.7603,
+      "step": 11196
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.056601586099276e-05,
+      "loss": 0.6841,
+      "step": 11197
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0564578979331186e-05,
+      "loss": 0.8469,
+      "step": 11198
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0563142085975784e-05,
+      "loss": 0.6213,
+      "step": 11199
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0561705180956312e-05,
+      "loss": 0.7754,
+      "step": 11200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0560268264302535e-05,
+      "loss": 0.8062,
+      "step": 11201
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0558831336044212e-05,
+      "loss": 0.7842,
+      "step": 11202
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.055739439621111e-05,
+      "loss": 0.7817,
+      "step": 11203
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0555957444832985e-05,
+      "loss": 0.7581,
+      "step": 11204
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0554520481939609e-05,
+      "loss": 0.8118,
+      "step": 11205
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0553083507560737e-05,
+      "loss": 0.6511,
+      "step": 11206
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0551646521726134e-05,
+      "loss": 0.8335,
+      "step": 11207
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0550209524465565e-05,
+      "loss": 0.8071,
+      "step": 11208
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0548772515808794e-05,
+      "loss": 0.7319,
+      "step": 11209
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0547335495785583e-05,
+      "loss": 0.8101,
+      "step": 11210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.05458984644257e-05,
+      "loss": 0.7314,
+      "step": 11211
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0544461421758903e-05,
+      "loss": 0.7729,
+      "step": 11212
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0543024367814963e-05,
+      "loss": 0.7681,
+      "step": 11213
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0541587302623644e-05,
+      "loss": 0.7607,
+      "step": 11214
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0540150226214705e-05,
+      "loss": 0.79,
+      "step": 11215
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.053871313861792e-05,
+      "loss": 0.7168,
+      "step": 11216
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0537276039863049e-05,
+      "loss": 0.7861,
+      "step": 11217
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0535838929979862e-05,
+      "loss": 0.7397,
+      "step": 11218
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0534401808998123e-05,
+      "loss": 0.7549,
+      "step": 11219
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0532964676947599e-05,
+      "loss": 0.8418,
+      "step": 11220
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0531527533858054e-05,
+      "loss": 0.7786,
+      "step": 11221
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0530090379759257e-05,
+      "loss": 0.79,
+      "step": 11222
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0528653214680973e-05,
+      "loss": 0.7002,
+      "step": 11223
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0527216038652971e-05,
+      "loss": 0.7788,
+      "step": 11224
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0525778851705022e-05,
+      "loss": 0.7441,
+      "step": 11225
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0524341653866888e-05,
+      "loss": 0.771,
+      "step": 11226
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0522904445168338e-05,
+      "loss": 0.7861,
+      "step": 11227
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0521467225639139e-05,
+      "loss": 0.8013,
+      "step": 11228
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0520029995309065e-05,
+      "loss": 0.7856,
+      "step": 11229
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.051859275420788e-05,
+      "loss": 0.7598,
+      "step": 11230
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0517155502365353e-05,
+      "loss": 0.7847,
+      "step": 11231
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0515718239811254e-05,
+      "loss": 0.77,
+      "step": 11232
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0514280966575355e-05,
+      "loss": 0.7559,
+      "step": 11233
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0512843682687423e-05,
+      "loss": 0.7979,
+      "step": 11234
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0511406388177226e-05,
+      "loss": 0.8184,
+      "step": 11235
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0509969083074535e-05,
+      "loss": 0.7117,
+      "step": 11236
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.050853176740912e-05,
+      "loss": 0.7866,
+      "step": 11237
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0507094441210753e-05,
+      "loss": 0.7554,
+      "step": 11238
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0505657104509201e-05,
+      "loss": 0.8384,
+      "step": 11239
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0504219757334237e-05,
+      "loss": 0.7378,
+      "step": 11240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0502782399715633e-05,
+      "loss": 0.6956,
+      "step": 11241
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0501345031683161e-05,
+      "loss": 0.6982,
+      "step": 11242
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0499907653266592e-05,
+      "loss": 0.7559,
+      "step": 11243
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0498470264495695e-05,
+      "loss": 0.7407,
+      "step": 11244
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0497032865400246e-05,
+      "loss": 0.675,
+      "step": 11245
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.049559545601001e-05,
+      "loss": 0.7725,
+      "step": 11246
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0494158036354766e-05,
+      "loss": 0.7988,
+      "step": 11247
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0492720606464286e-05,
+      "loss": 0.8271,
+      "step": 11248
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0491283166368342e-05,
+      "loss": 0.6812,
+      "step": 11249
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0489845716096704e-05,
+      "loss": 0.8047,
+      "step": 11250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048840825567915e-05,
+      "loss": 0.7539,
+      "step": 11251
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0486970785145445e-05,
+      "loss": 0.7456,
+      "step": 11252
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0485533304525373e-05,
+      "loss": 0.7251,
+      "step": 11253
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.04840958138487e-05,
+      "loss": 0.7815,
+      "step": 11254
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0482658313145206e-05,
+      "loss": 0.8271,
+      "step": 11255
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048122080244466e-05,
+      "loss": 0.7827,
+      "step": 11256
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0479783281776841e-05,
+      "loss": 0.7886,
+      "step": 11257
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0478345751171519e-05,
+      "loss": 0.7671,
+      "step": 11258
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0476908210658474e-05,
+      "loss": 0.7415,
+      "step": 11259
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0475470660267474e-05,
+      "loss": 0.7373,
+      "step": 11260
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0474033100028303e-05,
+      "loss": 0.856,
+      "step": 11261
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0472595529970729e-05,
+      "loss": 0.6997,
+      "step": 11262
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.047115795012453e-05,
+      "loss": 0.7539,
+      "step": 11263
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0469720360519485e-05,
+      "loss": 0.7495,
+      "step": 11264
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0468282761185364e-05,
+      "loss": 0.7397,
+      "step": 11265
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0466845152151948e-05,
+      "loss": 0.7344,
+      "step": 11266
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.046540753344901e-05,
+      "loss": 0.7766,
+      "step": 11267
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0463969905106333e-05,
+      "loss": 0.7842,
+      "step": 11268
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0462532267153684e-05,
+      "loss": 0.7144,
+      "step": 11269
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0461094619620851e-05,
+      "loss": 0.752,
+      "step": 11270
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0459656962537603e-05,
+      "loss": 0.7471,
+      "step": 11271
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0458219295933721e-05,
+      "loss": 0.832,
+      "step": 11272
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0456781619838981e-05,
+      "loss": 0.8345,
+      "step": 11273
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0455343934283162e-05,
+      "loss": 0.6985,
+      "step": 11274
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0453906239296042e-05,
+      "loss": 0.6907,
+      "step": 11275
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.04524685349074e-05,
+      "loss": 0.7773,
+      "step": 11276
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.045103082114701e-05,
+      "loss": 0.8076,
+      "step": 11277
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0449593098044656e-05,
+      "loss": 0.7925,
+      "step": 11278
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0448155365630115e-05,
+      "loss": 0.7573,
+      "step": 11279
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0446717623933164e-05,
+      "loss": 0.8193,
+      "step": 11280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0445279872983586e-05,
+      "loss": 0.6982,
+      "step": 11281
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.044384211281116e-05,
+      "loss": 0.7925,
+      "step": 11282
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.044240434344566e-05,
+      "loss": 0.7559,
+      "step": 11283
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0440966564916872e-05,
+      "loss": 0.7168,
+      "step": 11284
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0439528777254573e-05,
+      "loss": 0.7285,
+      "step": 11285
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0438090980488545e-05,
+      "loss": 0.6514,
+      "step": 11286
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0436653174648566e-05,
+      "loss": 0.699,
+      "step": 11287
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.043521535976442e-05,
+      "loss": 0.7632,
+      "step": 11288
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.043377753586588e-05,
+      "loss": 0.7549,
+      "step": 11289
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0432339702982736e-05,
+      "loss": 0.8057,
+      "step": 11290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0430901861144767e-05,
+      "loss": 0.6616,
+      "step": 11291
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0429464010381746e-05,
+      "loss": 0.7122,
+      "step": 11292
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0428026150723467e-05,
+      "loss": 0.7334,
+      "step": 11293
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0426588282199707e-05,
+      "loss": 0.7461,
+      "step": 11294
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0425150404840244e-05,
+      "loss": 0.7896,
+      "step": 11295
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0423712518674863e-05,
+      "loss": 0.6616,
+      "step": 11296
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0422274623733343e-05,
+      "loss": 0.7612,
+      "step": 11297
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0420836720045473e-05,
+      "loss": 0.7524,
+      "step": 11298
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.041939880764103e-05,
+      "loss": 0.7881,
+      "step": 11299
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0417960886549799e-05,
+      "loss": 0.7891,
+      "step": 11300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.041652295680156e-05,
+      "loss": 0.7676,
+      "step": 11301
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0415085018426101e-05,
+      "loss": 0.7422,
+      "step": 11302
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.04136470714532e-05,
+      "loss": 0.9131,
+      "step": 11303
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0412209115912645e-05,
+      "loss": 0.8315,
+      "step": 11304
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0410771151834216e-05,
+      "loss": 0.801,
+      "step": 11305
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0409333179247698e-05,
+      "loss": 0.9287,
+      "step": 11306
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0407895198182877e-05,
+      "loss": 0.748,
+      "step": 11307
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0406457208669538e-05,
+      "loss": 0.7212,
+      "step": 11308
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0405019210737462e-05,
+      "loss": 0.6858,
+      "step": 11309
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0403581204416433e-05,
+      "loss": 0.8271,
+      "step": 11310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0402143189736238e-05,
+      "loss": 0.7568,
+      "step": 11311
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.040070516672666e-05,
+      "loss": 0.8135,
+      "step": 11312
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0399267135417485e-05,
+      "loss": 0.7756,
+      "step": 11313
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0397829095838498e-05,
+      "loss": 0.7021,
+      "step": 11314
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0396391048019487e-05,
+      "loss": 0.7754,
+      "step": 11315
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0394952991990232e-05,
+      "loss": 0.7651,
+      "step": 11316
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0393514927780522e-05,
+      "loss": 0.8125,
+      "step": 11317
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0392076855420143e-05,
+      "loss": 0.793,
+      "step": 11318
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0390638774938882e-05,
+      "loss": 0.6628,
+      "step": 11319
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0389200686366524e-05,
+      "loss": 0.75,
+      "step": 11320
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0387762589732854e-05,
+      "loss": 0.7866,
+      "step": 11321
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.038632448506766e-05,
+      "loss": 0.7891,
+      "step": 11322
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0384886372400728e-05,
+      "loss": 0.7632,
+      "step": 11323
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0383448251761849e-05,
+      "loss": 0.8335,
+      "step": 11324
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0382010123180805e-05,
+      "loss": 0.7432,
+      "step": 11325
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0380571986687385e-05,
+      "loss": 0.7651,
+      "step": 11326
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0379133842311373e-05,
+      "loss": 0.7402,
+      "step": 11327
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.037769569008256e-05,
+      "loss": 0.8257,
+      "step": 11328
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0376257530030733e-05,
+      "loss": 0.7295,
+      "step": 11329
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0374819362185682e-05,
+      "loss": 0.7573,
+      "step": 11330
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0373381186577193e-05,
+      "loss": 0.6541,
+      "step": 11331
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0371943003235055e-05,
+      "loss": 0.7646,
+      "step": 11332
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0370504812189052e-05,
+      "loss": 0.7896,
+      "step": 11333
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0369066613468981e-05,
+      "loss": 0.7263,
+      "step": 11334
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0367628407104625e-05,
+      "loss": 0.6843,
+      "step": 11335
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0366190193125772e-05,
+      "loss": 0.7959,
+      "step": 11336
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0364751971562212e-05,
+      "loss": 0.8032,
+      "step": 11337
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.036331374244374e-05,
+      "loss": 0.7935,
+      "step": 11338
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0361875505800134e-05,
+      "loss": 0.8179,
+      "step": 11339
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0360437261661191e-05,
+      "loss": 0.7749,
+      "step": 11340
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0358999010056701e-05,
+      "loss": 0.8306,
+      "step": 11341
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0357560751016446e-05,
+      "loss": 0.7434,
+      "step": 11342
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0356122484570228e-05,
+      "loss": 0.6499,
+      "step": 11343
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.035468421074783e-05,
+      "loss": 0.6248,
+      "step": 11344
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0353245929579041e-05,
+      "loss": 0.7915,
+      "step": 11345
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0351807641093655e-05,
+      "loss": 0.7817,
+      "step": 11346
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.035036934532146e-05,
+      "loss": 0.7598,
+      "step": 11347
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0348931042292247e-05,
+      "loss": 0.7393,
+      "step": 11348
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0347492732035809e-05,
+      "loss": 0.7593,
+      "step": 11349
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0346054414581934e-05,
+      "loss": 0.8137,
+      "step": 11350
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0344616089960414e-05,
+      "loss": 0.635,
+      "step": 11351
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0343177758201044e-05,
+      "loss": 0.7583,
+      "step": 11352
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0341739419333611e-05,
+      "loss": 0.7632,
+      "step": 11353
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0340301073387906e-05,
+      "loss": 0.7817,
+      "step": 11354
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0338862720393723e-05,
+      "loss": 0.7661,
+      "step": 11355
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0337424360380855e-05,
+      "loss": 0.8359,
+      "step": 11356
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.033598599337909e-05,
+      "loss": 0.6992,
+      "step": 11357
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0334547619418224e-05,
+      "loss": 0.6707,
+      "step": 11358
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0333109238528048e-05,
+      "loss": 0.7495,
+      "step": 11359
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0331670850738353e-05,
+      "loss": 0.707,
+      "step": 11360
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0330232456078933e-05,
+      "loss": 0.9067,
+      "step": 11361
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0328794054579581e-05,
+      "loss": 0.7686,
+      "step": 11362
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0327355646270088e-05,
+      "loss": 0.8213,
+      "step": 11363
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0325917231180249e-05,
+      "loss": 0.8301,
+      "step": 11364
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0324478809339858e-05,
+      "loss": 0.7651,
+      "step": 11365
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0323040380778707e-05,
+      "loss": 0.8804,
+      "step": 11366
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0321601945526588e-05,
+      "loss": 0.7832,
+      "step": 11367
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0320163503613294e-05,
+      "loss": 0.7236,
+      "step": 11368
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.031872505506862e-05,
+      "loss": 0.7485,
+      "step": 11369
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0317286599922363e-05,
+      "loss": 0.697,
+      "step": 11370
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0315848138204315e-05,
+      "loss": 0.6926,
+      "step": 11371
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0314409669944267e-05,
+      "loss": 0.7373,
+      "step": 11372
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0312971195172017e-05,
+      "loss": 0.7769,
+      "step": 11373
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0311532713917356e-05,
+      "loss": 0.8604,
+      "step": 11374
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0310094226210081e-05,
+      "loss": 0.7231,
+      "step": 11375
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0308655732079989e-05,
+      "loss": 0.7109,
+      "step": 11376
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.030721723155687e-05,
+      "loss": 0.7192,
+      "step": 11377
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0305778724670519e-05,
+      "loss": 0.7935,
+      "step": 11378
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0304340211450733e-05,
+      "loss": 0.6887,
+      "step": 11379
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.030290169192731e-05,
+      "loss": 0.6615,
+      "step": 11380
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.030146316613004e-05,
+      "loss": 0.8486,
+      "step": 11381
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.030002463408872e-05,
+      "loss": 0.7029,
+      "step": 11382
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0298586095833152e-05,
+      "loss": 0.7339,
+      "step": 11383
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0297147551393121e-05,
+      "loss": 0.8354,
+      "step": 11384
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0295709000798428e-05,
+      "loss": 0.7363,
+      "step": 11385
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0294270444078871e-05,
+      "loss": 0.8003,
+      "step": 11386
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0292831881264243e-05,
+      "loss": 0.7837,
+      "step": 11387
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0291393312384338e-05,
+      "loss": 0.7554,
+      "step": 11388
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0289954737468958e-05,
+      "loss": 0.7559,
+      "step": 11389
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0288516156547898e-05,
+      "loss": 0.7705,
+      "step": 11390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0287077569650952e-05,
+      "loss": 0.7427,
+      "step": 11391
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.028563897680792e-05,
+      "loss": 0.7969,
+      "step": 11392
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0284200378048594e-05,
+      "loss": 0.6797,
+      "step": 11393
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0282761773402771e-05,
+      "loss": 0.6689,
+      "step": 11394
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0281323162900256e-05,
+      "loss": 0.8311,
+      "step": 11395
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.027988454657084e-05,
+      "loss": 0.6863,
+      "step": 11396
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0278445924444323e-05,
+      "loss": 0.7886,
+      "step": 11397
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0277007296550501e-05,
+      "loss": 0.7148,
+      "step": 11398
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.027556866291917e-05,
+      "loss": 0.7358,
+      "step": 11399
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.027413002358013e-05,
+      "loss": 0.8237,
+      "step": 11400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0272691378563176e-05,
+      "loss": 0.8022,
+      "step": 11401
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0271252727898109e-05,
+      "loss": 0.7842,
+      "step": 11402
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0269814071614726e-05,
+      "loss": 0.7705,
+      "step": 11403
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0268375409742823e-05,
+      "loss": 0.7876,
+      "step": 11404
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.02669367423122e-05,
+      "loss": 0.772,
+      "step": 11405
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0265498069352657e-05,
+      "loss": 0.7256,
+      "step": 11406
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0264059390893992e-05,
+      "loss": 0.7676,
+      "step": 11407
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0262620706966002e-05,
+      "loss": 0.7629,
+      "step": 11408
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0261182017598485e-05,
+      "loss": 0.8896,
+      "step": 11409
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0259743322821243e-05,
+      "loss": 0.6763,
+      "step": 11410
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0258304622664073e-05,
+      "loss": 0.7917,
+      "step": 11411
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0256865917156773e-05,
+      "loss": 0.8069,
+      "step": 11412
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0255427206329148e-05,
+      "loss": 0.7732,
+      "step": 11413
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0253988490210987e-05,
+      "loss": 0.7383,
+      "step": 11414
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0252549768832097e-05,
+      "loss": 0.8262,
+      "step": 11415
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0251111042222274e-05,
+      "loss": 0.7759,
+      "step": 11416
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.024967231041132e-05,
+      "loss": 0.7822,
+      "step": 11417
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0248233573429033e-05,
+      "loss": 0.7529,
+      "step": 11418
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0246794831305214e-05,
+      "loss": 0.8123,
+      "step": 11419
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0245356084069661e-05,
+      "loss": 0.7915,
+      "step": 11420
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0243917331752179e-05,
+      "loss": 0.7395,
+      "step": 11421
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0242478574382561e-05,
+      "loss": 0.7363,
+      "step": 11422
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0241039811990614e-05,
+      "loss": 0.7881,
+      "step": 11423
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0239601044606133e-05,
+      "loss": 0.6973,
+      "step": 11424
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0238162272258923e-05,
+      "loss": 0.7388,
+      "step": 11425
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.023672349497878e-05,
+      "loss": 0.7588,
+      "step": 11426
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0235284712795506e-05,
+      "loss": 0.7886,
+      "step": 11427
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0233845925738905e-05,
+      "loss": 0.7793,
+      "step": 11428
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0232407133838776e-05,
+      "loss": 0.7866,
+      "step": 11429
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0230968337124918e-05,
+      "loss": 0.7773,
+      "step": 11430
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0229529535627129e-05,
+      "loss": 0.8232,
+      "step": 11431
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.022809072937522e-05,
+      "loss": 0.8486,
+      "step": 11432
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0226651918398984e-05,
+      "loss": 0.7476,
+      "step": 11433
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0225213102728226e-05,
+      "loss": 0.7432,
+      "step": 11434
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0223774282392747e-05,
+      "loss": 0.7788,
+      "step": 11435
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0222335457422348e-05,
+      "loss": 0.833,
+      "step": 11436
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0220896627846831e-05,
+      "loss": 0.7109,
+      "step": 11437
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0219457793695997e-05,
+      "loss": 0.8281,
+      "step": 11438
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0218018954999648e-05,
+      "loss": 0.6982,
+      "step": 11439
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0216580111787587e-05,
+      "loss": 0.834,
+      "step": 11440
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0215141264089615e-05,
+      "loss": 0.6514,
+      "step": 11441
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0213702411935535e-05,
+      "loss": 0.7034,
+      "step": 11442
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0212263555355144e-05,
+      "loss": 0.8047,
+      "step": 11443
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0210824694378253e-05,
+      "loss": 0.7109,
+      "step": 11444
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0209385829034659e-05,
+      "loss": 0.835,
+      "step": 11445
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0207946959354166e-05,
+      "loss": 0.8623,
+      "step": 11446
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0206508085366576e-05,
+      "loss": 0.7778,
+      "step": 11447
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0205069207101692e-05,
+      "loss": 0.8242,
+      "step": 11448
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0203630324589316e-05,
+      "loss": 0.7681,
+      "step": 11449
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0202191437859253e-05,
+      "loss": 0.7725,
+      "step": 11450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0200752546941301e-05,
+      "loss": 0.813,
+      "step": 11451
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0199313651865268e-05,
+      "loss": 0.7158,
+      "step": 11452
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0197874752660956e-05,
+      "loss": 0.7837,
+      "step": 11453
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0196435849358166e-05,
+      "loss": 0.7793,
+      "step": 11454
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0194996941986705e-05,
+      "loss": 0.6797,
+      "step": 11455
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0193558030576373e-05,
+      "loss": 0.7605,
+      "step": 11456
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0192119115156972e-05,
+      "loss": 0.7305,
+      "step": 11457
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0190680195758314e-05,
+      "loss": 0.8286,
+      "step": 11458
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0189241272410191e-05,
+      "loss": 0.7646,
+      "step": 11459
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0187802345142414e-05,
+      "loss": 0.8262,
+      "step": 11460
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0186363413984785e-05,
+      "loss": 0.8398,
+      "step": 11461
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0184924478967108e-05,
+      "loss": 0.8486,
+      "step": 11462
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0183485540119189e-05,
+      "loss": 0.6606,
+      "step": 11463
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0182046597470828e-05,
+      "loss": 0.7751,
+      "step": 11464
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.018060765105183e-05,
+      "loss": 0.896,
+      "step": 11465
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0179168700892001e-05,
+      "loss": 0.6423,
+      "step": 11466
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0177729747021143e-05,
+      "loss": 0.6885,
+      "step": 11467
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0176290789469063e-05,
+      "loss": 0.7112,
+      "step": 11468
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0174851828265563e-05,
+      "loss": 0.7783,
+      "step": 11469
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0173412863440449e-05,
+      "loss": 0.6812,
+      "step": 11470
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0171973895023521e-05,
+      "loss": 0.7427,
+      "step": 11471
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0170534923044593e-05,
+      "loss": 0.7798,
+      "step": 11472
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0169095947533463e-05,
+      "loss": 0.6929,
+      "step": 11473
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0167656968519934e-05,
+      "loss": 0.7769,
+      "step": 11474
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0166217986033818e-05,
+      "loss": 0.7605,
+      "step": 11475
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0164779000104912e-05,
+      "loss": 0.8584,
+      "step": 11476
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0163340010763025e-05,
+      "loss": 0.7812,
+      "step": 11477
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0161901018037962e-05,
+      "loss": 0.7827,
+      "step": 11478
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0160462021959529e-05,
+      "loss": 0.8218,
+      "step": 11479
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0159023022557528e-05,
+      "loss": 0.7925,
+      "step": 11480
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0157584019861768e-05,
+      "loss": 0.7988,
+      "step": 11481
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0156145013902051e-05,
+      "loss": 0.7505,
+      "step": 11482
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0154706004708181e-05,
+      "loss": 0.6685,
+      "step": 11483
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0153266992309972e-05,
+      "loss": 0.8555,
+      "step": 11484
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0151827976737219e-05,
+      "loss": 0.7778,
+      "step": 11485
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0150388958019734e-05,
+      "loss": 0.7695,
+      "step": 11486
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0148949936187325e-05,
+      "loss": 0.7236,
+      "step": 11487
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0147510911269791e-05,
+      "loss": 0.7322,
+      "step": 11488
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.014607188329694e-05,
+      "loss": 0.7881,
+      "step": 11489
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0144632852298577e-05,
+      "loss": 0.7256,
+      "step": 11490
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.014319381830451e-05,
+      "loss": 0.7183,
+      "step": 11491
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0141754781344545e-05,
+      "loss": 0.895,
+      "step": 11492
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0140315741448487e-05,
+      "loss": 0.7832,
+      "step": 11493
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0138876698646141e-05,
+      "loss": 0.6323,
+      "step": 11494
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0137437652967316e-05,
+      "loss": 0.7559,
+      "step": 11495
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0135998604441815e-05,
+      "loss": 0.7075,
+      "step": 11496
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0134559553099446e-05,
+      "loss": 0.6299,
+      "step": 11497
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0133120498970015e-05,
+      "loss": 0.7324,
+      "step": 11498
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.013168144208333e-05,
+      "loss": 0.8462,
+      "step": 11499
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0130242382469196e-05,
+      "loss": 0.752,
+      "step": 11500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0128803320157419e-05,
+      "loss": 0.8257,
+      "step": 11501
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.012736425517781e-05,
+      "loss": 0.8164,
+      "step": 11502
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0125925187560166e-05,
+      "loss": 0.7139,
+      "step": 11503
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0124486117334302e-05,
+      "loss": 0.8828,
+      "step": 11504
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0123047044530021e-05,
+      "loss": 0.7285,
+      "step": 11505
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.012160796917713e-05,
+      "loss": 0.8701,
+      "step": 11506
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0120168891305438e-05,
+      "loss": 0.8164,
+      "step": 11507
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0118729810944747e-05,
+      "loss": 0.7288,
+      "step": 11508
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0117290728124871e-05,
+      "loss": 0.7368,
+      "step": 11509
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0115851642875613e-05,
+      "loss": 0.73,
+      "step": 11510
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.011441255522678e-05,
+      "loss": 0.7397,
+      "step": 11511
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.011297346520818e-05,
+      "loss": 0.8721,
+      "step": 11512
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0111534372849618e-05,
+      "loss": 0.8828,
+      "step": 11513
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0110095278180905e-05,
+      "loss": 0.7388,
+      "step": 11514
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0108656181231844e-05,
+      "loss": 0.7524,
+      "step": 11515
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0107217082032246e-05,
+      "loss": 0.7065,
+      "step": 11516
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0105777980611918e-05,
+      "loss": 0.7734,
+      "step": 11517
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0104338877000663e-05,
+      "loss": 0.73,
+      "step": 11518
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0102899771228291e-05,
+      "loss": 0.7988,
+      "step": 11519
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.010146066332461e-05,
+      "loss": 0.7681,
+      "step": 11520
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0100021553319428e-05,
+      "loss": 0.7747,
+      "step": 11521
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.009858244124255e-05,
+      "loss": 0.7632,
+      "step": 11522
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0097143327123788e-05,
+      "loss": 0.7288,
+      "step": 11523
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0095704210992944e-05,
+      "loss": 0.6841,
+      "step": 11524
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0094265092879831e-05,
+      "loss": 0.7524,
+      "step": 11525
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0092825972814254e-05,
+      "loss": 0.6758,
+      "step": 11526
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0091386850826021e-05,
+      "loss": 0.8569,
+      "step": 11527
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.008994772694494e-05,
+      "loss": 0.8232,
+      "step": 11528
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.008850860120082e-05,
+      "loss": 0.8691,
+      "step": 11529
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0087069473623465e-05,
+      "loss": 0.8345,
+      "step": 11530
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0085630344242685e-05,
+      "loss": 0.8252,
+      "step": 11531
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.008419121308829e-05,
+      "loss": 0.8623,
+      "step": 11532
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0082752080190086e-05,
+      "loss": 0.6765,
+      "step": 11533
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.008131294557788e-05,
+      "loss": 0.749,
+      "step": 11534
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0079873809281484e-05,
+      "loss": 0.8301,
+      "step": 11535
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0078434671330704e-05,
+      "loss": 0.7412,
+      "step": 11536
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0076995531755344e-05,
+      "loss": 0.719,
+      "step": 11537
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0075556390585219e-05,
+      "loss": 0.595,
+      "step": 11538
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0074117247850133e-05,
+      "loss": 0.834,
+      "step": 11539
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0072678103579894e-05,
+      "loss": 0.7734,
+      "step": 11540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0071238957804313e-05,
+      "loss": 0.7656,
+      "step": 11541
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0069799810553193e-05,
+      "loss": 0.7139,
+      "step": 11542
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0068360661856348e-05,
+      "loss": 0.8228,
+      "step": 11543
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0066921511743586e-05,
+      "loss": 0.7715,
+      "step": 11544
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0065482360244715e-05,
+      "loss": 0.7368,
+      "step": 11545
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0064043207389537e-05,
+      "loss": 0.8037,
+      "step": 11546
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0062604053207869e-05,
+      "loss": 0.8042,
+      "step": 11547
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0061164897729513e-05,
+      "loss": 0.7266,
+      "step": 11548
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0059725740984285e-05,
+      "loss": 0.6816,
+      "step": 11549
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0058286583001986e-05,
+      "loss": 0.6443,
+      "step": 11550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0056847423812429e-05,
+      "loss": 0.7422,
+      "step": 11551
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.005540826344542e-05,
+      "loss": 0.751,
+      "step": 11552
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0053969101930765e-05,
+      "loss": 0.6636,
+      "step": 11553
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0052529939298281e-05,
+      "loss": 0.8469,
+      "step": 11554
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.005109077557777e-05,
+      "loss": 0.6799,
+      "step": 11555
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0049651610799044e-05,
+      "loss": 0.8047,
+      "step": 11556
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0048212444991907e-05,
+      "loss": 0.875,
+      "step": 11557
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0046773278186172e-05,
+      "loss": 0.7856,
+      "step": 11558
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0045334110411646e-05,
+      "loss": 0.6492,
+      "step": 11559
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0043894941698138e-05,
+      "loss": 0.7715,
+      "step": 11560
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.004245577207546e-05,
+      "loss": 0.832,
+      "step": 11561
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0041016601573417e-05,
+      "loss": 0.8574,
+      "step": 11562
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0039577430221817e-05,
+      "loss": 0.7354,
+      "step": 11563
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0038138258050471e-05,
+      "loss": 0.792,
+      "step": 11564
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0036699085089188e-05,
+      "loss": 0.7559,
+      "step": 11565
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0035259911367774e-05,
+      "loss": 0.7197,
+      "step": 11566
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.003382073691604e-05,
+      "loss": 0.7979,
+      "step": 11567
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0032381561763796e-05,
+      "loss": 0.7275,
+      "step": 11568
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.003094238594085e-05,
+      "loss": 0.8135,
+      "step": 11569
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0029503209477009e-05,
+      "loss": 0.6978,
+      "step": 11570
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0028064032402083e-05,
+      "loss": 0.8159,
+      "step": 11571
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0026624854745884e-05,
+      "loss": 0.8623,
+      "step": 11572
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0025185676538214e-05,
+      "loss": 0.8096,
+      "step": 11573
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0023746497808888e-05,
+      "loss": 0.7876,
+      "step": 11574
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0022307318587714e-05,
+      "loss": 0.8467,
+      "step": 11575
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0020868138904503e-05,
+      "loss": 0.8755,
+      "step": 11576
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0019428958789059e-05,
+      "loss": 0.7578,
+      "step": 11577
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0017989778271192e-05,
+      "loss": 0.7573,
+      "step": 11578
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0016550597380713e-05,
+      "loss": 0.728,
+      "step": 11579
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.001511141614743e-05,
+      "loss": 0.7087,
+      "step": 11580
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.001367223460115e-05,
+      "loss": 0.8042,
+      "step": 11581
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0012233052771686e-05,
+      "loss": 0.6973,
+      "step": 11582
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0010793870688846e-05,
+      "loss": 0.7188,
+      "step": 11583
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0009354688382438e-05,
+      "loss": 0.8137,
+      "step": 11584
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0007915505882268e-05,
+      "loss": 0.6946,
+      "step": 11585
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0006476323218151e-05,
+      "loss": 0.8711,
+      "step": 11586
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0005037140419893e-05,
+      "loss": 0.6921,
+      "step": 11587
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0003597957517303e-05,
+      "loss": 0.6333,
+      "step": 11588
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0002158774540192e-05,
+      "loss": 0.7844,
+      "step": 11589
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0000719591518367e-05,
+      "loss": 0.751,
+      "step": 11590
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.999280408481637e-06,
+      "loss": 0.6997,
+      "step": 11591
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.99784122545981e-06,
+      "loss": 0.8545,
+      "step": 11592
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.996402042482699e-06,
+      "loss": 0.8057,
+      "step": 11593
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.994962859580107e-06,
+      "loss": 0.646,
+      "step": 11594
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.993523676781852e-06,
+      "loss": 0.825,
+      "step": 11595
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.992084494117734e-06,
+      "loss": 0.7285,
+      "step": 11596
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.990645311617565e-06,
+      "loss": 0.6667,
+      "step": 11597
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.989206129311159e-06,
+      "loss": 0.7935,
+      "step": 11598
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.987766947228315e-06,
+      "loss": 0.709,
+      "step": 11599
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.986327765398855e-06,
+      "loss": 0.8403,
+      "step": 11600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.984888583852574e-06,
+      "loss": 0.7964,
+      "step": 11601
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.983449402619292e-06,
+      "loss": 0.7544,
+      "step": 11602
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.98201022172881e-06,
+      "loss": 0.7129,
+      "step": 11603
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.980571041210948e-06,
+      "loss": 0.7979,
+      "step": 11604
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.9791318610955e-06,
+      "loss": 0.71,
+      "step": 11605
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.97769268141229e-06,
+      "loss": 0.8149,
+      "step": 11606
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.976253502191112e-06,
+      "loss": 0.8311,
+      "step": 11607
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.974814323461789e-06,
+      "loss": 0.7092,
+      "step": 11608
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.97337514525412e-06,
+      "loss": 0.6504,
+      "step": 11609
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.971935967597919e-06,
+      "loss": 0.7354,
+      "step": 11610
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.970496790522995e-06,
+      "loss": 0.7153,
+      "step": 11611
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.969057614059154e-06,
+      "loss": 0.7554,
+      "step": 11612
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.967618438236209e-06,
+      "loss": 0.7822,
+      "step": 11613
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.966179263083963e-06,
+      "loss": 0.7539,
+      "step": 11614
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.964740088632232e-06,
+      "loss": 0.8105,
+      "step": 11615
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.963300914910817e-06,
+      "loss": 0.7019,
+      "step": 11616
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.961861741949535e-06,
+      "loss": 0.6558,
+      "step": 11617
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.960422569778187e-06,
+      "loss": 0.7974,
+      "step": 11618
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.958983398426588e-06,
+      "loss": 0.7231,
+      "step": 11619
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.957544227924543e-06,
+      "loss": 0.7676,
+      "step": 11620
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.956105058301865e-06,
+      "loss": 0.615,
+      "step": 11621
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.954665889588356e-06,
+      "loss": 0.7769,
+      "step": 11622
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95322672181383e-06,
+      "loss": 0.7729,
+      "step": 11623
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.951787555008096e-06,
+      "loss": 0.6948,
+      "step": 11624
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.950348389200961e-06,
+      "loss": 0.7705,
+      "step": 11625
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.94890922442223e-06,
+      "loss": 0.8247,
+      "step": 11626
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.947470060701722e-06,
+      "loss": 0.7061,
+      "step": 11627
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.946030898069238e-06,
+      "loss": 0.7305,
+      "step": 11628
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.944591736554586e-06,
+      "loss": 0.7007,
+      "step": 11629
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.943152576187578e-06,
+      "loss": 0.8203,
+      "step": 11630
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.941713416998019e-06,
+      "loss": 0.7124,
+      "step": 11631
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.94027425901572e-06,
+      "loss": 0.8003,
+      "step": 11632
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.938835102270488e-06,
+      "loss": 0.7283,
+      "step": 11633
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.937395946792134e-06,
+      "loss": 0.6787,
+      "step": 11634
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.935956792610464e-06,
+      "loss": 0.6943,
+      "step": 11635
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.934517639755289e-06,
+      "loss": 0.7932,
+      "step": 11636
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.933078488256413e-06,
+      "loss": 0.7959,
+      "step": 11637
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.931639338143653e-06,
+      "loss": 0.771,
+      "step": 11638
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.930200189446806e-06,
+      "loss": 0.8164,
+      "step": 11639
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.928761042195692e-06,
+      "loss": 0.8442,
+      "step": 11640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.927321896420108e-06,
+      "loss": 0.8025,
+      "step": 11641
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.92588275214987e-06,
+      "loss": 0.7671,
+      "step": 11642
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.924443609414786e-06,
+      "loss": 0.7637,
+      "step": 11643
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.923004468244657e-06,
+      "loss": 0.7573,
+      "step": 11644
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.921565328669301e-06,
+      "loss": 0.739,
+      "step": 11645
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.920126190718517e-06,
+      "loss": 0.6934,
+      "step": 11646
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.91868705442212e-06,
+      "loss": 0.8257,
+      "step": 11647
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.917247919809916e-06,
+      "loss": 0.7241,
+      "step": 11648
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.915808786911712e-06,
+      "loss": 0.8179,
+      "step": 11649
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.914369655757314e-06,
+      "loss": 0.7891,
+      "step": 11650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.912930526376538e-06,
+      "loss": 0.7131,
+      "step": 11651
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.911491398799182e-06,
+      "loss": 0.7732,
+      "step": 11652
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.910052273055062e-06,
+      "loss": 0.748,
+      "step": 11653
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.908613149173978e-06,
+      "loss": 0.7266,
+      "step": 11654
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.907174027185747e-06,
+      "loss": 0.813,
+      "step": 11655
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.905734907120167e-06,
+      "loss": 0.7319,
+      "step": 11656
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.904295789007058e-06,
+      "loss": 0.7817,
+      "step": 11657
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.902856672876218e-06,
+      "loss": 0.6973,
+      "step": 11658
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.901417558757452e-06,
+      "loss": 0.7598,
+      "step": 11659
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.899978446680577e-06,
+      "loss": 0.624,
+      "step": 11660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.898539336675393e-06,
+      "loss": 0.7397,
+      "step": 11661
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.897100228771714e-06,
+      "loss": 0.8262,
+      "step": 11662
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.895661122999339e-06,
+      "loss": 0.814,
+      "step": 11663
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.894222019388087e-06,
+      "loss": 0.8306,
+      "step": 11664
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.892782917967756e-06,
+      "loss": 0.8438,
+      "step": 11665
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.891343818768158e-06,
+      "loss": 0.7583,
+      "step": 11666
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.889904721819097e-06,
+      "loss": 0.7534,
+      "step": 11667
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.888465627150383e-06,
+      "loss": 0.7104,
+      "step": 11668
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.88702653479182e-06,
+      "loss": 0.7964,
+      "step": 11669
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.885587444773223e-06,
+      "loss": 0.7632,
+      "step": 11670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.884148357124388e-06,
+      "loss": 0.8496,
+      "step": 11671
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.882709271875132e-06,
+      "loss": 0.7441,
+      "step": 11672
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.881270189055254e-06,
+      "loss": 0.7673,
+      "step": 11673
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.879831108694564e-06,
+      "loss": 0.7607,
+      "step": 11674
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.878392030822875e-06,
+      "loss": 0.8057,
+      "step": 11675
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.876952955469982e-06,
+      "loss": 0.6729,
+      "step": 11676
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.875513882665703e-06,
+      "loss": 0.7527,
+      "step": 11677
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.874074812439837e-06,
+      "loss": 0.8389,
+      "step": 11678
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.872635744822196e-06,
+      "loss": 0.7195,
+      "step": 11679
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.871196679842583e-06,
+      "loss": 0.7178,
+      "step": 11680
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.869757617530806e-06,
+      "loss": 0.8311,
+      "step": 11681
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.868318557916671e-06,
+      "loss": 0.7617,
+      "step": 11682
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.866879501029986e-06,
+      "loss": 0.7227,
+      "step": 11683
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.865440446900554e-06,
+      "loss": 0.7532,
+      "step": 11684
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.86400139555819e-06,
+      "loss": 0.7876,
+      "step": 11685
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.862562347032687e-06,
+      "loss": 0.8081,
+      "step": 11686
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.86112330135386e-06,
+      "loss": 0.8208,
+      "step": 11687
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.859684258551518e-06,
+      "loss": 0.7817,
+      "step": 11688
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.858245218655458e-06,
+      "loss": 0.8057,
+      "step": 11689
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.856806181695494e-06,
+      "loss": 0.7031,
+      "step": 11690
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.855367147701426e-06,
+      "loss": 0.7812,
+      "step": 11691
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.853928116703067e-06,
+      "loss": 0.8262,
+      "step": 11692
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.852489088730212e-06,
+      "loss": 0.752,
+      "step": 11693
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.85105006381268e-06,
+      "loss": 0.769,
+      "step": 11694
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.849611041980266e-06,
+      "loss": 0.7051,
+      "step": 11695
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.848172023262782e-06,
+      "loss": 0.7009,
+      "step": 11696
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.84673300769003e-06,
+      "loss": 0.6375,
+      "step": 11697
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.84529399529182e-06,
+      "loss": 0.7827,
+      "step": 11698
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.84385498609795e-06,
+      "loss": 0.7886,
+      "step": 11699
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.842415980138236e-06,
+      "loss": 0.7129,
+      "step": 11700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.840976977442474e-06,
+      "loss": 0.7109,
+      "step": 11701
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.839537978040475e-06,
+      "loss": 0.8223,
+      "step": 11702
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.838098981962043e-06,
+      "loss": 0.7064,
+      "step": 11703
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.836659989236977e-06,
+      "loss": 0.7207,
+      "step": 11704
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.835220999895093e-06,
+      "loss": 0.7129,
+      "step": 11705
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.833782013966187e-06,
+      "loss": 0.7861,
+      "step": 11706
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.832343031480071e-06,
+      "loss": 0.7319,
+      "step": 11707
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.83090405246654e-06,
+      "loss": 0.7827,
+      "step": 11708
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.829465076955412e-06,
+      "loss": 0.7344,
+      "step": 11709
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.828026104976479e-06,
+      "loss": 0.7505,
+      "step": 11710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.826587136559554e-06,
+      "loss": 0.8066,
+      "step": 11711
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.825148171734439e-06,
+      "loss": 0.7363,
+      "step": 11712
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.82370921053094e-06,
+      "loss": 0.7183,
+      "step": 11713
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.822270252978857e-06,
+      "loss": 0.7104,
+      "step": 11714
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.820831299108002e-06,
+      "loss": 0.7285,
+      "step": 11715
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.819392348948171e-06,
+      "loss": 0.7583,
+      "step": 11716
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.817953402529175e-06,
+      "loss": 0.7158,
+      "step": 11717
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.816514459880818e-06,
+      "loss": 0.6648,
+      "step": 11718
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.815075521032894e-06,
+      "loss": 0.7866,
+      "step": 11719
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.813636586015219e-06,
+      "loss": 0.7488,
+      "step": 11720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.812197654857588e-06,
+      "loss": 0.7163,
+      "step": 11721
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.810758727589814e-06,
+      "loss": 0.7031,
+      "step": 11722
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.809319804241691e-06,
+      "loss": 0.7407,
+      "step": 11723
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.80788088484303e-06,
+      "loss": 0.7534,
+      "step": 11724
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.80644196942363e-06,
+      "loss": 0.7495,
+      "step": 11725
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.805003058013298e-06,
+      "loss": 0.6799,
+      "step": 11726
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.803564150641834e-06,
+      "loss": 0.8208,
+      "step": 11727
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.802125247339047e-06,
+      "loss": 0.6865,
+      "step": 11728
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.800686348134733e-06,
+      "loss": 0.7246,
+      "step": 11729
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.7992474530587e-06,
+      "loss": 0.729,
+      "step": 11730
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.797808562140749e-06,
+      "loss": 0.7085,
+      "step": 11731
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.796369675410687e-06,
+      "loss": 0.8086,
+      "step": 11732
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.794930792898315e-06,
+      "loss": 0.7202,
+      "step": 11733
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.793491914633428e-06,
+      "loss": 0.7598,
+      "step": 11734
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.792053040645839e-06,
+      "loss": 0.8257,
+      "step": 11735
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.790614170965344e-06,
+      "loss": 0.7214,
+      "step": 11736
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.78917530562175e-06,
+      "loss": 0.6211,
+      "step": 11737
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.787736444644858e-06,
+      "loss": 0.7136,
+      "step": 11738
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.78629758806447e-06,
+      "loss": 0.8184,
+      "step": 11739
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.784858735910387e-06,
+      "loss": 0.7043,
+      "step": 11740
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.783419888212417e-06,
+      "loss": 0.791,
+      "step": 11741
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.781981045000353e-06,
+      "loss": 0.6865,
+      "step": 11742
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.780542206304008e-06,
+      "loss": 0.8047,
+      "step": 11743
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.77910337215317e-06,
+      "loss": 0.8145,
+      "step": 11744
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.777664542577655e-06,
+      "loss": 0.7847,
+      "step": 11745
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.776225717607255e-06,
+      "loss": 0.7534,
+      "step": 11746
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.774786897271776e-06,
+      "loss": 0.729,
+      "step": 11747
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.773348081601018e-06,
+      "loss": 0.7637,
+      "step": 11748
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.771909270624784e-06,
+      "loss": 0.7886,
+      "step": 11749
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.770470464372873e-06,
+      "loss": 0.8804,
+      "step": 11750
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.769031662875087e-06,
+      "loss": 0.7827,
+      "step": 11751
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.767592866161229e-06,
+      "loss": 0.7646,
+      "step": 11752
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.766154074261097e-06,
+      "loss": 0.958,
+      "step": 11753
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.764715287204497e-06,
+      "loss": 0.8081,
+      "step": 11754
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.763276505021222e-06,
+      "loss": 0.7617,
+      "step": 11755
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.76183772774108e-06,
+      "loss": 0.7783,
+      "step": 11756
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.760398955393867e-06,
+      "loss": 0.7637,
+      "step": 11757
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.75896018800939e-06,
+      "loss": 0.7817,
+      "step": 11758
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.757521425617439e-06,
+      "loss": 0.7646,
+      "step": 11759
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.756082668247825e-06,
+      "loss": 0.7695,
+      "step": 11760
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.754643915930339e-06,
+      "loss": 0.833,
+      "step": 11761
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.75320516869479e-06,
+      "loss": 0.7378,
+      "step": 11762
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.751766426570972e-06,
+      "loss": 0.71,
+      "step": 11763
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.750327689588682e-06,
+      "loss": 0.7407,
+      "step": 11764
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.748888957777731e-06,
+      "loss": 0.7036,
+      "step": 11765
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.747450231167906e-06,
+      "loss": 0.8076,
+      "step": 11766
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.746011509789018e-06,
+      "loss": 0.8286,
+      "step": 11767
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.744572793670857e-06,
+      "loss": 0.7642,
+      "step": 11768
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.74313408284323e-06,
+      "loss": 0.7354,
+      "step": 11769
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.741695377335927e-06,
+      "loss": 0.73,
+      "step": 11770
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.74025667717876e-06,
+      "loss": 0.8521,
+      "step": 11771
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.738817982401515e-06,
+      "loss": 0.7925,
+      "step": 11772
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.737379293034001e-06,
+      "loss": 0.7554,
+      "step": 11773
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.73594060910601e-06,
+      "loss": 0.7754,
+      "step": 11774
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.734501930647345e-06,
+      "loss": 0.6912,
+      "step": 11775
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.733063257687801e-06,
+      "loss": 0.7012,
+      "step": 11776
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.73162459025718e-06,
+      "loss": 0.8022,
+      "step": 11777
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.730185928385279e-06,
+      "loss": 0.8169,
+      "step": 11778
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.728747272101893e-06,
+      "loss": 0.7764,
+      "step": 11779
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.727308621436829e-06,
+      "loss": 0.7827,
+      "step": 11780
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.725869976419874e-06,
+      "loss": 0.7852,
+      "step": 11781
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.724431337080835e-06,
+      "loss": 0.665,
+      "step": 11782
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.722992703449502e-06,
+      "loss": 0.79,
+      "step": 11783
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.72155407555568e-06,
+      "loss": 0.6919,
+      "step": 11784
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.720115453429159e-06,
+      "loss": 0.7808,
+      "step": 11785
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.718676837099746e-06,
+      "loss": 0.8154,
+      "step": 11786
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.717238226597229e-06,
+      "loss": 0.729,
+      "step": 11787
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.715799621951409e-06,
+      "loss": 0.8101,
+      "step": 11788
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.714361023192085e-06,
+      "loss": 0.7725,
+      "step": 11789
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.712922430349051e-06,
+      "loss": 0.7271,
+      "step": 11790
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.711483843452104e-06,
+      "loss": 0.8081,
+      "step": 11791
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.710045262531043e-06,
+      "loss": 0.7178,
+      "step": 11792
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.708606687615667e-06,
+      "loss": 0.793,
+      "step": 11793
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.707168118735762e-06,
+      "loss": 0.7261,
+      "step": 11794
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.705729555921135e-06,
+      "loss": 0.6978,
+      "step": 11795
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.704290999201575e-06,
+      "loss": 0.7026,
+      "step": 11796
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.702852448606884e-06,
+      "loss": 0.7046,
+      "step": 11797
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.701413904166852e-06,
+      "loss": 0.7883,
+      "step": 11798
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.699975365911281e-06,
+      "loss": 0.7173,
+      "step": 11799
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.698536833869961e-06,
+      "loss": 0.7153,
+      "step": 11800
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.697098308072692e-06,
+      "loss": 0.6047,
+      "step": 11801
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.695659788549268e-06,
+      "loss": 0.7593,
+      "step": 11802
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.694221275329483e-06,
+      "loss": 0.7974,
+      "step": 11803
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.692782768443132e-06,
+      "loss": 0.772,
+      "step": 11804
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.691344267920015e-06,
+      "loss": 0.772,
+      "step": 11805
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.689905773789922e-06,
+      "loss": 0.748,
+      "step": 11806
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.688467286082647e-06,
+      "loss": 0.7915,
+      "step": 11807
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.68702880482799e-06,
+      "loss": 0.8086,
+      "step": 11808
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.685590330055736e-06,
+      "loss": 0.689,
+      "step": 11809
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.68415186179569e-06,
+      "loss": 0.6885,
+      "step": 11810
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.682713400077639e-06,
+      "loss": 0.7788,
+      "step": 11811
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.681274944931384e-06,
+      "loss": 0.7832,
+      "step": 11812
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.67983649638671e-06,
+      "loss": 0.8521,
+      "step": 11813
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.678398054473417e-06,
+      "loss": 0.8423,
+      "step": 11814
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.676959619221298e-06,
+      "loss": 0.749,
+      "step": 11815
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.675521190660145e-06,
+      "loss": 0.71,
+      "step": 11816
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.674082768819751e-06,
+      "loss": 0.7334,
+      "step": 11817
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.672644353729913e-06,
+      "loss": 0.7017,
+      "step": 11818
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.67120594542042e-06,
+      "loss": 0.7573,
+      "step": 11819
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.66976754392107e-06,
+      "loss": 0.6565,
+      "step": 11820
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.668329149261652e-06,
+      "loss": 0.6646,
+      "step": 11821
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.666890761471955e-06,
+      "loss": 0.7534,
+      "step": 11822
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.665452380581781e-06,
+      "loss": 0.8062,
+      "step": 11823
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.664014006620914e-06,
+      "loss": 0.793,
+      "step": 11824
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.662575639619149e-06,
+      "loss": 0.7891,
+      "step": 11825
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.66113727960628e-06,
+      "loss": 0.8032,
+      "step": 11826
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.659698926612097e-06,
+      "loss": 0.7949,
+      "step": 11827
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.65826058066639e-06,
+      "loss": 0.8101,
+      "step": 11828
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.656822241798959e-06,
+      "loss": 0.7563,
+      "step": 11829
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.655383910039585e-06,
+      "loss": 0.7222,
+      "step": 11830
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.653945585418069e-06,
+      "loss": 0.8179,
+      "step": 11831
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.652507267964193e-06,
+      "loss": 0.7151,
+      "step": 11832
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.651068957707755e-06,
+      "loss": 0.8013,
+      "step": 11833
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.649630654678542e-06,
+      "loss": 0.7251,
+      "step": 11834
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.648192358906349e-06,
+      "loss": 0.748,
+      "step": 11835
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.646754070420965e-06,
+      "loss": 0.8413,
+      "step": 11836
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.645315789252175e-06,
+      "loss": 0.7769,
+      "step": 11837
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.643877515429775e-06,
+      "loss": 0.7524,
+      "step": 11838
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.642439248983555e-06,
+      "loss": 0.7229,
+      "step": 11839
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.641000989943304e-06,
+      "loss": 0.7048,
+      "step": 11840
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63956273833881e-06,
+      "loss": 0.7539,
+      "step": 11841
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.638124494199871e-06,
+      "loss": 0.7827,
+      "step": 11842
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.636686257556265e-06,
+      "loss": 0.7522,
+      "step": 11843
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63524802843779e-06,
+      "loss": 0.8105,
+      "step": 11844
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63380980687423e-06,
+      "loss": 0.8184,
+      "step": 11845
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.632371592895379e-06,
+      "loss": 0.7852,
+      "step": 11846
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.630933386531019e-06,
+      "loss": 0.8877,
+      "step": 11847
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.62949518781095e-06,
+      "loss": 0.8091,
+      "step": 11848
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.628056996764947e-06,
+      "loss": 0.6743,
+      "step": 11849
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.62661881342281e-06,
+      "loss": 0.7119,
+      "step": 11850
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.625180637814321e-06,
+      "loss": 0.6953,
+      "step": 11851
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.623742469969269e-06,
+      "loss": 0.8193,
+      "step": 11852
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.622304309917445e-06,
+      "loss": 0.7803,
+      "step": 11853
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.620866157688629e-06,
+      "loss": 0.8447,
+      "step": 11854
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.619428013312622e-06,
+      "loss": 0.9067,
+      "step": 11855
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.617989876819198e-06,
+      "loss": 0.7144,
+      "step": 11856
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.616551748238155e-06,
+      "loss": 0.7212,
+      "step": 11857
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.615113627599272e-06,
+      "loss": 0.7954,
+      "step": 11858
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.613675514932343e-06,
+      "loss": 0.8125,
+      "step": 11859
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.612237410267147e-06,
+      "loss": 0.8628,
+      "step": 11860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.61079931363348e-06,
+      "loss": 0.8247,
+      "step": 11861
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.60936122506112e-06,
+      "loss": 0.7651,
+      "step": 11862
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.607923144579858e-06,
+      "loss": 0.7446,
+      "step": 11863
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.606485072219478e-06,
+      "loss": 0.7371,
+      "step": 11864
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.60504700800977e-06,
+      "loss": 0.7739,
+      "step": 11865
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.603608951980518e-06,
+      "loss": 0.7297,
+      "step": 11866
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.602170904161503e-06,
+      "loss": 0.7437,
+      "step": 11867
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.60073286458252e-06,
+      "loss": 0.8232,
+      "step": 11868
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.599294833273343e-06,
+      "loss": 0.7529,
+      "step": 11869
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.597856810263767e-06,
+      "loss": 0.7588,
+      "step": 11870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.596418795583569e-06,
+      "loss": 0.7822,
+      "step": 11871
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.594980789262543e-06,
+      "loss": 0.7217,
+      "step": 11872
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.593542791330464e-06,
+      "loss": 0.8174,
+      "step": 11873
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.592104801817124e-06,
+      "loss": 0.8086,
+      "step": 11874
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.590666820752301e-06,
+      "loss": 0.8403,
+      "step": 11875
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.589228848165787e-06,
+      "loss": 0.7417,
+      "step": 11876
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.587790884087358e-06,
+      "loss": 0.8459,
+      "step": 11877
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.586352928546803e-06,
+      "loss": 0.6929,
+      "step": 11878
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.584914981573902e-06,
+      "loss": 0.8257,
+      "step": 11879
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.583477043198443e-06,
+      "loss": 0.8271,
+      "step": 11880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.582039113450208e-06,
+      "loss": 0.7734,
+      "step": 11881
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.580601192358974e-06,
+      "loss": 0.7036,
+      "step": 11882
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.579163279954533e-06,
+      "loss": 0.7363,
+      "step": 11883
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.577725376266658e-06,
+      "loss": 0.771,
+      "step": 11884
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.576287481325142e-06,
+      "loss": 0.7832,
+      "step": 11885
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.57484959515976e-06,
+      "loss": 0.7524,
+      "step": 11886
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.573411717800298e-06,
+      "loss": 0.7578,
+      "step": 11887
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.571973849276533e-06,
+      "loss": 0.6912,
+      "step": 11888
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.570535989618255e-06,
+      "loss": 0.812,
+      "step": 11889
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.569098138855237e-06,
+      "loss": 0.6943,
+      "step": 11890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.567660297017265e-06,
+      "loss": 0.749,
+      "step": 11891
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.566222464134121e-06,
+      "loss": 0.7622,
+      "step": 11892
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.564784640235584e-06,
+      "loss": 0.7671,
+      "step": 11893
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.563346825351434e-06,
+      "loss": 0.8979,
+      "step": 11894
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.561909019511458e-06,
+      "loss": 0.8657,
+      "step": 11895
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.560471222745432e-06,
+      "loss": 0.8628,
+      "step": 11896
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.55903343508313e-06,
+      "loss": 0.7314,
+      "step": 11897
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.557595656554344e-06,
+      "loss": 0.8025,
+      "step": 11898
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.556157887188844e-06,
+      "loss": 0.7729,
+      "step": 11899
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.554720127016417e-06,
+      "loss": 0.8169,
+      "step": 11900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.553282376066838e-06,
+      "loss": 0.8408,
+      "step": 11901
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.551844634369887e-06,
+      "loss": 0.7227,
+      "step": 11902
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.550406901955347e-06,
+      "loss": 0.7305,
+      "step": 11903
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.548969178852992e-06,
+      "loss": 0.7993,
+      "step": 11904
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.547531465092602e-06,
+      "loss": 0.7515,
+      "step": 11905
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.546093760703962e-06,
+      "loss": 0.8198,
+      "step": 11906
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.54465606571684e-06,
+      "loss": 0.6865,
+      "step": 11907
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.543218380161022e-06,
+      "loss": 0.7725,
+      "step": 11908
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.54178070406628e-06,
+      "loss": 0.7354,
+      "step": 11909
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.5403430374624e-06,
+      "loss": 0.7471,
+      "step": 11910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.538905380379156e-06,
+      "loss": 0.7366,
+      "step": 11911
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.537467732846317e-06,
+      "loss": 0.7075,
+      "step": 11912
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.536030094893674e-06,
+      "loss": 0.8076,
+      "step": 11913
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.534592466550992e-06,
+      "loss": 0.7769,
+      "step": 11914
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.533154847848057e-06,
+      "loss": 0.6396,
+      "step": 11915
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.53171723881464e-06,
+      "loss": 0.7935,
+      "step": 11916
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.53027963948052e-06,
+      "loss": 0.8145,
+      "step": 11917
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.528842049875471e-06,
+      "loss": 0.7856,
+      "step": 11918
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.527404470029274e-06,
+      "loss": 0.7393,
+      "step": 11919
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.525966899971699e-06,
+      "loss": 0.7876,
+      "step": 11920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.524529339732527e-06,
+      "loss": 0.7046,
+      "step": 11921
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.523091789341527e-06,
+      "loss": 0.7827,
+      "step": 11922
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.521654248828483e-06,
+      "loss": 0.8027,
+      "step": 11923
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.520216718223159e-06,
+      "loss": 0.7236,
+      "step": 11924
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.518779197555341e-06,
+      "loss": 0.7778,
+      "step": 11925
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.517341686854797e-06,
+      "loss": 0.6978,
+      "step": 11926
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.515904186151302e-06,
+      "loss": 0.7832,
+      "step": 11927
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.51446669547463e-06,
+      "loss": 0.7739,
+      "step": 11928
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.513029214854557e-06,
+      "loss": 0.7764,
+      "step": 11929
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.511591744320855e-06,
+      "loss": 0.7969,
+      "step": 11930
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.510154283903297e-06,
+      "loss": 0.8364,
+      "step": 11931
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.508716833631661e-06,
+      "loss": 0.791,
+      "step": 11932
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.507279393535712e-06,
+      "loss": 0.7119,
+      "step": 11933
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.505841963645235e-06,
+      "loss": 0.7524,
+      "step": 11934
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.50440454398999e-06,
+      "loss": 0.7212,
+      "step": 11935
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.50296713459976e-06,
+      "loss": 0.729,
+      "step": 11936
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.501529735504305e-06,
+      "loss": 0.7317,
+      "step": 11937
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.50009234673341e-06,
+      "loss": 0.7637,
+      "step": 11938
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.498654968316839e-06,
+      "loss": 0.6973,
+      "step": 11939
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.497217600284368e-06,
+      "loss": 0.7812,
+      "step": 11940
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.495780242665766e-06,
+      "loss": 0.749,
+      "step": 11941
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.4943428954908e-06,
+      "loss": 0.7368,
+      "step": 11942
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.492905558789254e-06,
+      "loss": 0.7114,
+      "step": 11943
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.491468232590883e-06,
+      "loss": 0.8015,
+      "step": 11944
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.49003091692547e-06,
+      "loss": 0.77,
+      "step": 11945
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.488593611822777e-06,
+      "loss": 0.7319,
+      "step": 11946
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.487156317312582e-06,
+      "loss": 0.8081,
+      "step": 11947
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.485719033424647e-06,
+      "loss": 0.8521,
+      "step": 11948
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.484281760188748e-06,
+      "loss": 0.7471,
+      "step": 11949
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.482844497634647e-06,
+      "loss": 0.7788,
+      "step": 11950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.481407245792124e-06,
+      "loss": 0.8257,
+      "step": 11951
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.479970004690937e-06,
+      "loss": 0.7451,
+      "step": 11952
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.478532774360864e-06,
+      "loss": 0.8701,
+      "step": 11953
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.477095554831665e-06,
+      "loss": 0.8079,
+      "step": 11954
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.475658346133117e-06,
+      "loss": 0.8071,
+      "step": 11955
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.474221148294983e-06,
+      "loss": 0.6707,
+      "step": 11956
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.47278396134703e-06,
+      "loss": 0.7534,
+      "step": 11957
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.47134678531903e-06,
+      "loss": 0.7466,
+      "step": 11958
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.469909620240747e-06,
+      "loss": 0.7078,
+      "step": 11959
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.46847246614195e-06,
+      "loss": 0.7588,
+      "step": 11960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467035323052403e-06,
+      "loss": 0.8613,
+      "step": 11961
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.465598191001882e-06,
+      "loss": 0.7144,
+      "step": 11962
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.46416107002014e-06,
+      "loss": 0.7871,
+      "step": 11963
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.462723960136953e-06,
+      "loss": 0.7568,
+      "step": 11964
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.461286861382082e-06,
+      "loss": 0.7539,
+      "step": 11965
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.459849773785298e-06,
+      "loss": 0.7251,
+      "step": 11966
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.45841269737636e-06,
+      "loss": 0.833,
+      "step": 11967
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.456975632185039e-06,
+      "loss": 0.7715,
+      "step": 11968
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.455538578241099e-06,
+      "loss": 0.78,
+      "step": 11969
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.454101535574305e-06,
+      "loss": 0.7734,
+      "step": 11970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.452664504214422e-06,
+      "loss": 0.7505,
+      "step": 11971
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.45122748419121e-06,
+      "loss": 0.6572,
+      "step": 11972
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.449790475534439e-06,
+      "loss": 0.8125,
+      "step": 11973
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.448353478273868e-06,
+      "loss": 0.7534,
+      "step": 11974
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.446916492439268e-06,
+      "loss": 0.729,
+      "step": 11975
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.445479518060394e-06,
+      "loss": 0.6497,
+      "step": 11976
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.444042555167016e-06,
+      "loss": 0.7917,
+      "step": 11977
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.442605603788892e-06,
+      "loss": 0.7505,
+      "step": 11978
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.44116866395579e-06,
+      "loss": 0.6655,
+      "step": 11979
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.439731735697466e-06,
+      "loss": 0.7639,
+      "step": 11980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.438294819043692e-06,
+      "loss": 0.7051,
+      "step": 11981
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.436857914024217e-06,
+      "loss": 0.7485,
+      "step": 11982
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.435421020668815e-06,
+      "loss": 0.8403,
+      "step": 11983
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.43398413900724e-06,
+      "loss": 0.7324,
+      "step": 11984
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.43254726906926e-06,
+      "loss": 0.7363,
+      "step": 11985
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.431110410884633e-06,
+      "loss": 0.7686,
+      "step": 11986
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.429673564483115e-06,
+      "loss": 0.7749,
+      "step": 11987
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.428236729894475e-06,
+      "loss": 0.8716,
+      "step": 11988
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.426799907148464e-06,
+      "loss": 0.7886,
+      "step": 11989
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.425363096274852e-06,
+      "loss": 0.707,
+      "step": 11990
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.42392629730339e-06,
+      "loss": 0.8438,
+      "step": 11991
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.422489510263843e-06,
+      "loss": 0.7764,
+      "step": 11992
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.42105273518597e-06,
+      "loss": 0.7939,
+      "step": 11993
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.419615972099532e-06,
+      "loss": 0.731,
+      "step": 11994
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.418179221034281e-06,
+      "loss": 0.6162,
+      "step": 11995
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.416742482019985e-06,
+      "loss": 0.7891,
+      "step": 11996
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.415305755086394e-06,
+      "loss": 0.6804,
+      "step": 11997
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.413869040263272e-06,
+      "loss": 0.6328,
+      "step": 11998
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.412432337580372e-06,
+      "loss": 0.7988,
+      "step": 11999
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.410995647067459e-06,
+      "loss": 0.7427,
+      "step": 12000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.409558968754284e-06,
+      "loss": 0.7959,
+      "step": 12001
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.408122302670604e-06,
+      "loss": 0.7095,
+      "step": 12002
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.406685648846177e-06,
+      "loss": 0.7573,
+      "step": 12003
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.405249007310763e-06,
+      "loss": 0.8799,
+      "step": 12004
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.403812378094114e-06,
+      "loss": 0.8281,
+      "step": 12005
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.40237576122599e-06,
+      "loss": 0.7485,
+      "step": 12006
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.400939156736146e-06,
+      "loss": 0.8442,
+      "step": 12007
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.399502564654332e-06,
+      "loss": 0.7544,
+      "step": 12008
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.398065985010313e-06,
+      "loss": 0.7893,
+      "step": 12009
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.396629417833835e-06,
+      "loss": 0.7495,
+      "step": 12010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.39519286315466e-06,
+      "loss": 0.7656,
+      "step": 12011
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.393756321002537e-06,
+      "loss": 0.8003,
+      "step": 12012
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.392319791407227e-06,
+      "loss": 0.8367,
+      "step": 12013
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.390883274398475e-06,
+      "loss": 0.6885,
+      "step": 12014
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.389446770006044e-06,
+      "loss": 0.7891,
+      "step": 12015
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.388010278259682e-06,
+      "loss": 0.7261,
+      "step": 12016
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.386573799189144e-06,
+      "loss": 0.7422,
+      "step": 12017
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.385137332824182e-06,
+      "loss": 0.7603,
+      "step": 12018
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.383700879194549e-06,
+      "loss": 0.811,
+      "step": 12019
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.38226443833e-06,
+      "loss": 0.8345,
+      "step": 12020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.380828010260283e-06,
+      "loss": 0.7607,
+      "step": 12021
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.379391595015156e-06,
+      "loss": 0.7817,
+      "step": 12022
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.377955192624361e-06,
+      "loss": 0.7715,
+      "step": 12023
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.376518803117662e-06,
+      "loss": 0.7334,
+      "step": 12024
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.3750824265248e-06,
+      "loss": 0.7744,
+      "step": 12025
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37364606287553e-06,
+      "loss": 0.7866,
+      "step": 12026
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.372209712199601e-06,
+      "loss": 0.707,
+      "step": 12027
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.370773374526768e-06,
+      "loss": 0.7122,
+      "step": 12028
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.369337049886776e-06,
+      "loss": 0.7236,
+      "step": 12029
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.367900738309378e-06,
+      "loss": 0.7886,
+      "step": 12030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.366464439824318e-06,
+      "loss": 0.7524,
+      "step": 12031
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.365028154461351e-06,
+      "loss": 0.7339,
+      "step": 12032
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.363591882250228e-06,
+      "loss": 0.6016,
+      "step": 12033
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.36215562322069e-06,
+      "loss": 0.6775,
+      "step": 12034
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.360719377402493e-06,
+      "loss": 0.7812,
+      "step": 12035
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.359283144825377e-06,
+      "loss": 0.7046,
+      "step": 12036
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.357846925519101e-06,
+      "loss": 0.7437,
+      "step": 12037
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.3564107195134e-06,
+      "loss": 0.7559,
+      "step": 12038
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.354974526838035e-06,
+      "loss": 0.7759,
+      "step": 12039
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.353538347522741e-06,
+      "loss": 0.8423,
+      "step": 12040
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.352102181597274e-06,
+      "loss": 0.8071,
+      "step": 12041
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.35066602909137e-06,
+      "loss": 0.7715,
+      "step": 12042
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.349229890034788e-06,
+      "loss": 0.7266,
+      "step": 12043
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.347793764457265e-06,
+      "loss": 0.7227,
+      "step": 12044
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.346357652388549e-06,
+      "loss": 0.7083,
+      "step": 12045
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.344921553858386e-06,
+      "loss": 0.7007,
+      "step": 12046
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.34348546889652e-06,
+      "loss": 0.7314,
+      "step": 12047
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.342049397532702e-06,
+      "loss": 0.7246,
+      "step": 12048
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.340613339796664e-06,
+      "loss": 0.7515,
+      "step": 12049
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.339177295718164e-06,
+      "loss": 0.7783,
+      "step": 12050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.337741265326936e-06,
+      "loss": 0.6865,
+      "step": 12051
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.33630524865273e-06,
+      "loss": 0.8672,
+      "step": 12052
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.334869245725282e-06,
+      "loss": 0.7471,
+      "step": 12053
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.333433256574348e-06,
+      "loss": 0.7754,
+      "step": 12054
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.331997281229656e-06,
+      "loss": 0.8462,
+      "step": 12055
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.33056131972096e-06,
+      "loss": 0.6655,
+      "step": 12056
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.329125372077996e-06,
+      "loss": 0.7769,
+      "step": 12057
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.32768943833051e-06,
+      "loss": 0.7588,
+      "step": 12058
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.326253518508243e-06,
+      "loss": 0.7905,
+      "step": 12059
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.32481761264093e-06,
+      "loss": 0.7627,
+      "step": 12060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.323381720758322e-06,
+      "loss": 0.8145,
+      "step": 12061
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.321945842890153e-06,
+      "loss": 0.7051,
+      "step": 12062
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.320509979066168e-06,
+      "loss": 0.7407,
+      "step": 12063
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.3190741293161e-06,
+      "loss": 0.8047,
+      "step": 12064
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.317638293669702e-06,
+      "loss": 0.749,
+      "step": 12065
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.3162024721567e-06,
+      "loss": 0.7095,
+      "step": 12066
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.314766664806844e-06,
+      "loss": 0.6375,
+      "step": 12067
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.313330871649862e-06,
+      "loss": 0.8032,
+      "step": 12068
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.311895092715504e-06,
+      "loss": 0.8145,
+      "step": 12069
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.310459328033503e-06,
+      "loss": 0.7227,
+      "step": 12070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.309023577633598e-06,
+      "loss": 0.8105,
+      "step": 12071
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.307587841545524e-06,
+      "loss": 0.7192,
+      "step": 12072
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.306152119799029e-06,
+      "loss": 0.6802,
+      "step": 12073
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.30471641242384e-06,
+      "loss": 0.7871,
+      "step": 12074
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.303280719449694e-06,
+      "loss": 0.7612,
+      "step": 12075
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.301845040906334e-06,
+      "loss": 0.9141,
+      "step": 12076
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.30040937682349e-06,
+      "loss": 0.7231,
+      "step": 12077
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.298973727230907e-06,
+      "loss": 0.6233,
+      "step": 12078
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.297538092158309e-06,
+      "loss": 0.6919,
+      "step": 12079
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.29610247163544e-06,
+      "loss": 0.7046,
+      "step": 12080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.294666865692034e-06,
+      "loss": 0.7686,
+      "step": 12081
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.293231274357826e-06,
+      "loss": 0.7832,
+      "step": 12082
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.291795697662549e-06,
+      "loss": 0.8418,
+      "step": 12083
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.290360135635937e-06,
+      "loss": 0.8076,
+      "step": 12084
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.288924588307725e-06,
+      "loss": 0.8335,
+      "step": 12085
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.28748905570765e-06,
+      "loss": 0.8745,
+      "step": 12086
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.286053537865438e-06,
+      "loss": 0.8433,
+      "step": 12087
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.284618034810832e-06,
+      "loss": 0.7969,
+      "step": 12088
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.283182546573558e-06,
+      "loss": 0.7661,
+      "step": 12089
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.281747073183347e-06,
+      "loss": 0.7207,
+      "step": 12090
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.280311614669936e-06,
+      "loss": 0.7222,
+      "step": 12091
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.278876171063052e-06,
+      "loss": 0.7622,
+      "step": 12092
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.277440742392431e-06,
+      "loss": 0.7593,
+      "step": 12093
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.276005328687804e-06,
+      "loss": 0.6772,
+      "step": 12094
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.2745699299789e-06,
+      "loss": 0.7275,
+      "step": 12095
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.273134546295448e-06,
+      "loss": 0.7549,
+      "step": 12096
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.271699177667185e-06,
+      "loss": 0.7783,
+      "step": 12097
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.270263824123834e-06,
+      "loss": 0.6238,
+      "step": 12098
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.268828485695131e-06,
+      "loss": 0.728,
+      "step": 12099
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.267393162410798e-06,
+      "loss": 0.7607,
+      "step": 12100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.265957854300573e-06,
+      "loss": 0.688,
+      "step": 12101
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.264522561394175e-06,
+      "loss": 0.7666,
+      "step": 12102
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.263087283721342e-06,
+      "loss": 0.7397,
+      "step": 12103
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.261652021311796e-06,
+      "loss": 0.728,
+      "step": 12104
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.260216774195263e-06,
+      "loss": 0.8291,
+      "step": 12105
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.258781542401476e-06,
+      "loss": 0.79,
+      "step": 12106
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.25734632596016e-06,
+      "loss": 0.7148,
+      "step": 12107
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.255911124901043e-06,
+      "loss": 0.7983,
+      "step": 12108
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.254475939253847e-06,
+      "loss": 0.7029,
+      "step": 12109
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.253040769048306e-06,
+      "loss": 0.8354,
+      "step": 12110
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.251605614314138e-06,
+      "loss": 0.7671,
+      "step": 12111
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.250170475081078e-06,
+      "loss": 0.877,
+      "step": 12112
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.248735351378839e-06,
+      "loss": 0.7612,
+      "step": 12113
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.247300243237158e-06,
+      "loss": 0.7415,
+      "step": 12114
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.24586515068575e-06,
+      "loss": 0.7881,
+      "step": 12115
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.244430073754348e-06,
+      "loss": 0.7058,
+      "step": 12116
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.242995012472666e-06,
+      "loss": 0.7673,
+      "step": 12117
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.24155996687044e-06,
+      "loss": 0.6855,
+      "step": 12118
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.240124936977383e-06,
+      "loss": 0.7778,
+      "step": 12119
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.238689922823223e-06,
+      "loss": 0.7407,
+      "step": 12120
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.237254924437679e-06,
+      "loss": 0.8398,
+      "step": 12121
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.235819941850475e-06,
+      "loss": 0.7881,
+      "step": 12122
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.234384975091338e-06,
+      "loss": 0.7305,
+      "step": 12123
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.23295002418998e-06,
+      "loss": 0.7388,
+      "step": 12124
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.231515089176132e-06,
+      "loss": 0.8291,
+      "step": 12125
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.230080170079508e-06,
+      "loss": 0.813,
+      "step": 12126
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.228645266929836e-06,
+      "loss": 0.8691,
+      "step": 12127
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.227210379756826e-06,
+      "loss": 0.6282,
+      "step": 12128
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.22577550859021e-06,
+      "loss": 0.8242,
+      "step": 12129
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.224340653459698e-06,
+      "loss": 0.7788,
+      "step": 12130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.222905814395016e-06,
+      "loss": 0.7358,
+      "step": 12131
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.221470991425877e-06,
+      "loss": 0.7397,
+      "step": 12132
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.220036184582005e-06,
+      "loss": 0.7373,
+      "step": 12133
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.218601393893119e-06,
+      "loss": 0.7002,
+      "step": 12134
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.21716661938893e-06,
+      "loss": 0.7617,
+      "step": 12135
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.215731861099165e-06,
+      "loss": 0.7212,
+      "step": 12136
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.21429711905353e-06,
+      "loss": 0.7656,
+      "step": 12137
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.212862393281755e-06,
+      "loss": 0.7908,
+      "step": 12138
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.211427683813546e-06,
+      "loss": 0.7637,
+      "step": 12139
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.209992990678629e-06,
+      "loss": 0.7283,
+      "step": 12140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.208558313906709e-06,
+      "loss": 0.8081,
+      "step": 12141
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.207123653527513e-06,
+      "loss": 0.7034,
+      "step": 12142
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.205689009570746e-06,
+      "loss": 0.7544,
+      "step": 12143
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.204254382066134e-06,
+      "loss": 0.7234,
+      "step": 12144
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.202819771043378e-06,
+      "loss": 0.6829,
+      "step": 12145
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.201385176532205e-06,
+      "loss": 0.71,
+      "step": 12146
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.199950598562324e-06,
+      "loss": 0.8394,
+      "step": 12147
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.198516037163447e-06,
+      "loss": 0.7534,
+      "step": 12148
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.197081492365294e-06,
+      "loss": 0.7373,
+      "step": 12149
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.195646964197566e-06,
+      "loss": 0.8394,
+      "step": 12150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.194212452689988e-06,
+      "loss": 0.8545,
+      "step": 12151
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.192777957872263e-06,
+      "loss": 0.7402,
+      "step": 12152
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19134347977411e-06,
+      "loss": 0.7354,
+      "step": 12153
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.189909018425234e-06,
+      "loss": 0.8608,
+      "step": 12154
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.188474573855354e-06,
+      "loss": 0.7954,
+      "step": 12155
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.187040146094172e-06,
+      "loss": 0.8184,
+      "step": 12156
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.185605735171407e-06,
+      "loss": 0.676,
+      "step": 12157
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.184171341116765e-06,
+      "loss": 0.6157,
+      "step": 12158
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.182736963959955e-06,
+      "loss": 0.7109,
+      "step": 12159
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.181302603730689e-06,
+      "loss": 0.8047,
+      "step": 12160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.179868260458673e-06,
+      "loss": 0.8418,
+      "step": 12161
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.178433934173617e-06,
+      "loss": 0.7031,
+      "step": 12162
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.176999624905234e-06,
+      "loss": 0.6724,
+      "step": 12163
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.175565332683229e-06,
+      "loss": 0.6719,
+      "step": 12164
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.174131057537304e-06,
+      "loss": 0.7935,
+      "step": 12165
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.172696799497176e-06,
+      "loss": 0.7344,
+      "step": 12166
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.171262558592542e-06,
+      "loss": 0.772,
+      "step": 12167
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.169828334853118e-06,
+      "loss": 0.8442,
+      "step": 12168
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.168394128308605e-06,
+      "loss": 0.7803,
+      "step": 12169
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.166959938988711e-06,
+      "loss": 0.8052,
+      "step": 12170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.165525766923142e-06,
+      "loss": 0.8008,
+      "step": 12171
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.164091612141602e-06,
+      "loss": 0.7407,
+      "step": 12172
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.162657474673793e-06,
+      "loss": 0.7344,
+      "step": 12173
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.161223354549427e-06,
+      "loss": 0.6987,
+      "step": 12174
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.1597892517982e-06,
+      "loss": 0.8315,
+      "step": 12175
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.158355166449825e-06,
+      "loss": 0.6978,
+      "step": 12176
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.156921098533995e-06,
+      "loss": 0.8354,
+      "step": 12177
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.155487048080423e-06,
+      "loss": 0.7627,
+      "step": 12178
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.154053015118807e-06,
+      "loss": 0.7451,
+      "step": 12179
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.152618999678845e-06,
+      "loss": 0.7295,
+      "step": 12180
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.151185001790249e-06,
+      "loss": 0.8311,
+      "step": 12181
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.149751021482708e-06,
+      "loss": 0.7759,
+      "step": 12182
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.148317058785933e-06,
+      "loss": 0.729,
+      "step": 12183
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.146883113729624e-06,
+      "loss": 0.7969,
+      "step": 12184
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.14544918634348e-06,
+      "loss": 0.7725,
+      "step": 12185
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.144015276657198e-06,
+      "loss": 0.6826,
+      "step": 12186
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.142581384700485e-06,
+      "loss": 0.7852,
+      "step": 12187
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.141147510503032e-06,
+      "loss": 0.7271,
+      "step": 12188
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.139713654094548e-06,
+      "loss": 0.6816,
+      "step": 12189
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.13827981550472e-06,
+      "loss": 0.7012,
+      "step": 12190
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.136845994763257e-06,
+      "loss": 0.7847,
+      "step": 12191
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.13541219189985e-06,
+      "loss": 0.6697,
+      "step": 12192
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.133978406944202e-06,
+      "loss": 0.7446,
+      "step": 12193
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.132544639926004e-06,
+      "loss": 0.873,
+      "step": 12194
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.131110890874958e-06,
+      "loss": 0.6948,
+      "step": 12195
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.129677159820759e-06,
+      "loss": 0.7432,
+      "step": 12196
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.128243446793101e-06,
+      "loss": 0.7798,
+      "step": 12197
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.126809751821682e-06,
+      "loss": 0.7173,
+      "step": 12198
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.125376074936197e-06,
+      "loss": 0.749,
+      "step": 12199
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.123942416166343e-06,
+      "loss": 0.6848,
+      "step": 12200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.122508775541808e-06,
+      "loss": 0.7534,
+      "step": 12201
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.121075153092296e-06,
+      "loss": 0.8291,
+      "step": 12202
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.11964154884749e-06,
+      "loss": 0.7485,
+      "step": 12203
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.118207962837096e-06,
+      "loss": 0.73,
+      "step": 12204
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.116774395090794e-06,
+      "loss": 0.7725,
+      "step": 12205
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.115340845638289e-06,
+      "loss": 0.6992,
+      "step": 12206
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.113907314509261e-06,
+      "loss": 0.8066,
+      "step": 12207
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.112473801733414e-06,
+      "loss": 0.791,
+      "step": 12208
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.11104030734043e-06,
+      "loss": 0.7881,
+      "step": 12209
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.109606831360005e-06,
+      "loss": 0.7607,
+      "step": 12210
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.108173373821831e-06,
+      "loss": 0.7192,
+      "step": 12211
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.106739934755595e-06,
+      "loss": 0.6655,
+      "step": 12212
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.10530651419099e-06,
+      "loss": 0.6553,
+      "step": 12213
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.1038731121577e-06,
+      "loss": 0.7231,
+      "step": 12214
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.102439728685425e-06,
+      "loss": 0.8003,
+      "step": 12215
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.101006363803842e-06,
+      "loss": 0.8315,
+      "step": 12216
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.099573017542648e-06,
+      "loss": 0.7415,
+      "step": 12217
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.098139689931526e-06,
+      "loss": 0.8574,
+      "step": 12218
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.09670638100017e-06,
+      "loss": 0.7529,
+      "step": 12219
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.095273090778258e-06,
+      "loss": 0.8296,
+      "step": 12220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.093839819295487e-06,
+      "loss": 0.6665,
+      "step": 12221
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.092406566581535e-06,
+      "loss": 0.7896,
+      "step": 12222
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.090973332666094e-06,
+      "loss": 0.7732,
+      "step": 12223
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.08954011757885e-06,
+      "loss": 0.8154,
+      "step": 12224
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.088106921349482e-06,
+      "loss": 0.7871,
+      "step": 12225
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.086673744007685e-06,
+      "loss": 0.6846,
+      "step": 12226
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.085240585583131e-06,
+      "loss": 0.7322,
+      "step": 12227
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.083807446105518e-06,
+      "loss": 0.7231,
+      "step": 12228
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.08237432560452e-06,
+      "loss": 0.7686,
+      "step": 12229
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.080941224109826e-06,
+      "loss": 0.7915,
+      "step": 12230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.079508141651112e-06,
+      "loss": 0.7178,
+      "step": 12231
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.07807507825807e-06,
+      "loss": 0.7471,
+      "step": 12232
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.076642033960375e-06,
+      "loss": 0.7173,
+      "step": 12233
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.075209008787713e-06,
+      "loss": 0.7278,
+      "step": 12234
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.073776002769764e-06,
+      "loss": 0.7698,
+      "step": 12235
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.07234301593621e-06,
+      "loss": 0.8354,
+      "step": 12236
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.070910048316732e-06,
+      "loss": 0.6416,
+      "step": 12237
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.069477099941006e-06,
+      "loss": 0.8457,
+      "step": 12238
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.06804417083872e-06,
+      "loss": 0.6355,
+      "step": 12239
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.066611261039543e-06,
+      "loss": 0.6831,
+      "step": 12240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.065178370573164e-06,
+      "loss": 0.8687,
+      "step": 12241
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.063745499469255e-06,
+      "loss": 0.792,
+      "step": 12242
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.0623126477575e-06,
+      "loss": 0.7471,
+      "step": 12243
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.060879815467569e-06,
+      "loss": 0.7788,
+      "step": 12244
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.05944700262915e-06,
+      "loss": 0.7793,
+      "step": 12245
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.058014209271909e-06,
+      "loss": 0.8276,
+      "step": 12246
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.05658143542553e-06,
+      "loss": 0.7646,
+      "step": 12247
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.055148681119688e-06,
+      "loss": 0.7402,
+      "step": 12248
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.053715946384058e-06,
+      "loss": 0.8174,
+      "step": 12249
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.052283231248314e-06,
+      "loss": 0.873,
+      "step": 12250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.050850535742137e-06,
+      "loss": 0.7793,
+      "step": 12251
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.049417859895198e-06,
+      "loss": 0.7959,
+      "step": 12252
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.047985203737166e-06,
+      "loss": 0.7954,
+      "step": 12253
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.046552567297724e-06,
+      "loss": 0.7188,
+      "step": 12254
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.045119950606536e-06,
+      "loss": 0.8926,
+      "step": 12255
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.043687353693285e-06,
+      "loss": 0.7544,
+      "step": 12256
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.042254776587635e-06,
+      "loss": 0.7368,
+      "step": 12257
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.040822219319266e-06,
+      "loss": 0.7178,
+      "step": 12258
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.039389681917842e-06,
+      "loss": 0.793,
+      "step": 12259
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.03795716441304e-06,
+      "loss": 0.7529,
+      "step": 12260
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.03652466683453e-06,
+      "loss": 0.7573,
+      "step": 12261
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.035092189211979e-06,
+      "loss": 0.7368,
+      "step": 12262
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.03365973157506e-06,
+      "loss": 0.7473,
+      "step": 12263
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.032227293953447e-06,
+      "loss": 0.7175,
+      "step": 12264
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.0307948763768e-06,
+      "loss": 0.7866,
+      "step": 12265
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.029362478874796e-06,
+      "loss": 0.7676,
+      "step": 12266
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.027930101477104e-06,
+      "loss": 0.7983,
+      "step": 12267
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.026497744213382e-06,
+      "loss": 0.8027,
+      "step": 12268
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.025065407113308e-06,
+      "loss": 0.7266,
+      "step": 12269
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.023633090206542e-06,
+      "loss": 0.7881,
+      "step": 12270
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.022200793522757e-06,
+      "loss": 0.8115,
+      "step": 12271
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.020768517091614e-06,
+      "loss": 0.8276,
+      "step": 12272
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.019336260942784e-06,
+      "loss": 0.7139,
+      "step": 12273
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.017904025105931e-06,
+      "loss": 0.8096,
+      "step": 12274
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.016471809610718e-06,
+      "loss": 0.8364,
+      "step": 12275
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.015039614486808e-06,
+      "loss": 0.7852,
+      "step": 12276
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.013607439763874e-06,
+      "loss": 0.751,
+      "step": 12277
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.01217528547157e-06,
+      "loss": 0.8115,
+      "step": 12278
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.010743151639568e-06,
+      "loss": 0.8164,
+      "step": 12279
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.009311038297523e-06,
+      "loss": 0.8057,
+      "step": 12280
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.007878945475107e-06,
+      "loss": 0.5842,
+      "step": 12281
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.006446873201974e-06,
+      "loss": 0.7881,
+      "step": 12282
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.005014821507785e-06,
+      "loss": 0.7476,
+      "step": 12283
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.003582790422205e-06,
+      "loss": 0.729,
+      "step": 12284
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.002150779974899e-06,
+      "loss": 0.6831,
+      "step": 12285
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.00071879019552e-06,
+      "loss": 0.7656,
+      "step": 12286
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.99928682111373e-06,
+      "loss": 0.7234,
+      "step": 12287
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.997854872759194e-06,
+      "loss": 0.7563,
+      "step": 12288
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.996422945161563e-06,
+      "loss": 0.6294,
+      "step": 12289
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.994991038350504e-06,
+      "loss": 0.7515,
+      "step": 12290
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.993559152355666e-06,
+      "loss": 0.6924,
+      "step": 12291
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.992127287206716e-06,
+      "loss": 0.8711,
+      "step": 12292
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.990695442933304e-06,
+      "loss": 0.7744,
+      "step": 12293
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.989263619565096e-06,
+      "loss": 0.7246,
+      "step": 12294
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.987831817131738e-06,
+      "loss": 0.769,
+      "step": 12295
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.986400035662897e-06,
+      "loss": 0.7778,
+      "step": 12296
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.984968275188218e-06,
+      "loss": 0.7485,
+      "step": 12297
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.983536535737363e-06,
+      "loss": 0.7373,
+      "step": 12298
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.982104817339986e-06,
+      "loss": 0.7744,
+      "step": 12299
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.980673120025737e-06,
+      "loss": 0.8245,
+      "step": 12300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.97924144382428e-06,
+      "loss": 0.6982,
+      "step": 12301
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.977809788765256e-06,
+      "loss": 0.7476,
+      "step": 12302
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.97637815487833e-06,
+      "loss": 0.7305,
+      "step": 12303
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.974946542193144e-06,
+      "loss": 0.7817,
+      "step": 12304
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.97351495073936e-06,
+      "loss": 0.6829,
+      "step": 12305
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.972083380546621e-06,
+      "loss": 0.6985,
+      "step": 12306
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.970651831644586e-06,
+      "loss": 0.7725,
+      "step": 12307
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.969220304062899e-06,
+      "loss": 0.7305,
+      "step": 12308
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.967788797831219e-06,
+      "loss": 0.7397,
+      "step": 12309
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.966357312979184e-06,
+      "loss": 0.7383,
+      "step": 12310
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.964925849536454e-06,
+      "loss": 0.7905,
+      "step": 12311
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.963494407532677e-06,
+      "loss": 0.7544,
+      "step": 12312
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.962062986997496e-06,
+      "loss": 0.6777,
+      "step": 12313
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.960631587960566e-06,
+      "loss": 0.7837,
+      "step": 12314
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.959200210451528e-06,
+      "loss": 0.8022,
+      "step": 12315
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.957768854500036e-06,
+      "loss": 0.8843,
+      "step": 12316
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.95633752013573e-06,
+      "loss": 0.6948,
+      "step": 12317
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.954906207388266e-06,
+      "loss": 0.8506,
+      "step": 12318
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.953474916287278e-06,
+      "loss": 0.7346,
+      "step": 12319
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.952043646862422e-06,
+      "loss": 0.7856,
+      "step": 12320
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.950612399143338e-06,
+      "loss": 0.7505,
+      "step": 12321
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.949181173159672e-06,
+      "loss": 0.7693,
+      "step": 12322
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.947749968941066e-06,
+      "loss": 0.7056,
+      "step": 12323
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.946318786517168e-06,
+      "loss": 0.7241,
+      "step": 12324
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.944887625917618e-06,
+      "loss": 0.8267,
+      "step": 12325
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.94345648717206e-06,
+      "loss": 0.7568,
+      "step": 12326
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.942025370310138e-06,
+      "loss": 0.7456,
+      "step": 12327
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.94059427536149e-06,
+      "loss": 0.8101,
+      "step": 12328
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.939163202355762e-06,
+      "loss": 0.7202,
+      "step": 12329
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.937732151322592e-06,
+      "loss": 0.7632,
+      "step": 12330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.936301122291622e-06,
+      "loss": 0.7471,
+      "step": 12331
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.93487011529249e-06,
+      "loss": 0.8423,
+      "step": 12332
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.93343913035484e-06,
+      "loss": 0.8096,
+      "step": 12333
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.932008167508306e-06,
+      "loss": 0.7568,
+      "step": 12334
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.930577226782534e-06,
+      "loss": 0.7798,
+      "step": 12335
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.92914630820715e-06,
+      "loss": 0.7036,
+      "step": 12336
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.927715411811807e-06,
+      "loss": 0.8433,
+      "step": 12337
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.926284537626132e-06,
+      "loss": 0.8491,
+      "step": 12338
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.924853685679765e-06,
+      "loss": 0.7607,
+      "step": 12339
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.923422856002342e-06,
+      "loss": 0.7803,
+      "step": 12340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.921992048623503e-06,
+      "loss": 0.8325,
+      "step": 12341
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.92056126357288e-06,
+      "loss": 0.8145,
+      "step": 12342
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.919130500880104e-06,
+      "loss": 0.7832,
+      "step": 12343
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.917699760574817e-06,
+      "loss": 0.7979,
+      "step": 12344
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.916269042686647e-06,
+      "loss": 0.7344,
+      "step": 12345
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.914838347245235e-06,
+      "loss": 0.7373,
+      "step": 12346
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.913407674280206e-06,
+      "loss": 0.8364,
+      "step": 12347
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.911977023821198e-06,
+      "loss": 0.7363,
+      "step": 12348
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.910546395897843e-06,
+      "loss": 0.7295,
+      "step": 12349
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.90911579053977e-06,
+      "loss": 0.751,
+      "step": 12350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.907685207776615e-06,
+      "loss": 0.8091,
+      "step": 12351
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.906254647638004e-06,
+      "loss": 0.7251,
+      "step": 12352
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.904824110153569e-06,
+      "loss": 0.7046,
+      "step": 12353
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.903393595352944e-06,
+      "loss": 0.8237,
+      "step": 12354
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.901963103265752e-06,
+      "loss": 0.7246,
+      "step": 12355
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.90053263392163e-06,
+      "loss": 0.7075,
+      "step": 12356
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.899102187350201e-06,
+      "loss": 0.8091,
+      "step": 12357
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.897671763581088e-06,
+      "loss": 0.7144,
+      "step": 12358
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.896241362643931e-06,
+      "loss": 0.6865,
+      "step": 12359
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.894810984568346e-06,
+      "loss": 0.709,
+      "step": 12360
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.893380629383964e-06,
+      "loss": 0.8042,
+      "step": 12361
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.891950297120414e-06,
+      "loss": 0.7363,
+      "step": 12362
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.890519987807319e-06,
+      "loss": 0.7803,
+      "step": 12363
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.889089701474302e-06,
+      "loss": 0.8306,
+      "step": 12364
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.887659438150995e-06,
+      "loss": 0.8457,
+      "step": 12365
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.886229197867012e-06,
+      "loss": 0.728,
+      "step": 12366
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.884798980651988e-06,
+      "loss": 0.7729,
+      "step": 12367
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.883368786535536e-06,
+      "loss": 0.7915,
+      "step": 12368
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.881938615547287e-06,
+      "loss": 0.8013,
+      "step": 12369
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.880508467716858e-06,
+      "loss": 0.7324,
+      "step": 12370
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.879078343073874e-06,
+      "loss": 0.7671,
+      "step": 12371
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.877648241647959e-06,
+      "loss": 0.9199,
+      "step": 12372
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.876218163468723e-06,
+      "loss": 0.8643,
+      "step": 12373
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.874788108565796e-06,
+      "loss": 0.6307,
+      "step": 12374
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.873358076968797e-06,
+      "loss": 0.7803,
+      "step": 12375
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.871928068707344e-06,
+      "loss": 0.7573,
+      "step": 12376
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.870498083811053e-06,
+      "loss": 0.8232,
+      "step": 12377
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.86906812230955e-06,
+      "loss": 0.7646,
+      "step": 12378
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.867638184232446e-06,
+      "loss": 0.7734,
+      "step": 12379
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.866208269609366e-06,
+      "loss": 0.7803,
+      "step": 12380
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.864778378469916e-06,
+      "loss": 0.6743,
+      "step": 12381
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.863348510843724e-06,
+      "loss": 0.657,
+      "step": 12382
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.861918666760399e-06,
+      "loss": 0.7114,
+      "step": 12383
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.86048884624956e-06,
+      "loss": 0.709,
+      "step": 12384
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.859059049340819e-06,
+      "loss": 0.7524,
+      "step": 12385
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.857629276063795e-06,
+      "loss": 0.7522,
+      "step": 12386
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.856199526448098e-06,
+      "loss": 0.873,
+      "step": 12387
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.854769800523341e-06,
+      "loss": 0.7158,
+      "step": 12388
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.853340098319144e-06,
+      "loss": 0.7661,
+      "step": 12389
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.851910419865111e-06,
+      "loss": 0.7798,
+      "step": 12390
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.850480765190862e-06,
+      "loss": 0.7363,
+      "step": 12391
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.849051134326e-06,
+      "loss": 0.7507,
+      "step": 12392
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.847621527300148e-06,
+      "loss": 0.6826,
+      "step": 12393
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.846191944142903e-06,
+      "loss": 0.7378,
+      "step": 12394
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.844762384883889e-06,
+      "loss": 0.7266,
+      "step": 12395
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.843332849552702e-06,
+      "loss": 0.7505,
+      "step": 12396
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.841903338178964e-06,
+      "loss": 0.8064,
+      "step": 12397
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.840473850792273e-06,
+      "loss": 0.7783,
+      "step": 12398
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.839044387422247e-06,
+      "loss": 0.7319,
+      "step": 12399
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.837614948098484e-06,
+      "loss": 0.7192,
+      "step": 12400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.836185532850598e-06,
+      "loss": 0.8159,
+      "step": 12401
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.834756141708195e-06,
+      "loss": 0.7568,
+      "step": 12402
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.833326774700877e-06,
+      "loss": 0.7915,
+      "step": 12403
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.831897431858257e-06,
+      "loss": 0.7937,
+      "step": 12404
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.83046811320993e-06,
+      "loss": 0.752,
+      "step": 12405
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.829038818785512e-06,
+      "loss": 0.7605,
+      "step": 12406
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.827609548614598e-06,
+      "loss": 0.7212,
+      "step": 12407
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.8261803027268e-06,
+      "loss": 0.7092,
+      "step": 12408
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.824751081151711e-06,
+      "loss": 0.7754,
+      "step": 12409
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.823321883918945e-06,
+      "loss": 0.7534,
+      "step": 12410
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.821892711058093e-06,
+      "loss": 0.7925,
+      "step": 12411
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.820463562598769e-06,
+      "loss": 0.7251,
+      "step": 12412
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.819034438570561e-06,
+      "loss": 0.7295,
+      "step": 12413
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.81760533900308e-06,
+      "loss": 0.7026,
+      "step": 12414
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.816176263925923e-06,
+      "loss": 0.7344,
+      "step": 12415
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.814747213368688e-06,
+      "loss": 0.7168,
+      "step": 12416
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.81331818736098e-06,
+      "loss": 0.835,
+      "step": 12417
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.811889185932385e-06,
+      "loss": 0.7988,
+      "step": 12418
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.810460209112516e-06,
+      "loss": 0.6917,
+      "step": 12419
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.80903125693096e-06,
+      "loss": 0.8311,
+      "step": 12420
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.807602329417321e-06,
+      "loss": 0.834,
+      "step": 12421
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.80617342660119e-06,
+      "loss": 0.7078,
+      "step": 12422
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.804744548512168e-06,
+      "loss": 0.7646,
+      "step": 12423
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.803315695179845e-06,
+      "loss": 0.7065,
+      "step": 12424
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.801886866633821e-06,
+      "loss": 0.7737,
+      "step": 12425
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.800458062903689e-06,
+      "loss": 0.7705,
+      "step": 12426
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.799029284019045e-06,
+      "loss": 0.7305,
+      "step": 12427
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.79760053000948e-06,
+      "loss": 0.6367,
+      "step": 12428
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.796171800904588e-06,
+      "loss": 0.8013,
+      "step": 12429
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.794743096733958e-06,
+      "loss": 0.7554,
+      "step": 12430
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.79331441752719e-06,
+      "loss": 0.7837,
+      "step": 12431
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.79188576331387e-06,
+      "loss": 0.7788,
+      "step": 12432
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.790457134123585e-06,
+      "loss": 0.7241,
+      "step": 12433
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.789028529985937e-06,
+      "loss": 0.8096,
+      "step": 12434
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.787599950930503e-06,
+      "loss": 0.6892,
+      "step": 12435
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.786171396986883e-06,
+      "loss": 0.7764,
+      "step": 12436
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.784742868184657e-06,
+      "loss": 0.6936,
+      "step": 12437
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.78331436455342e-06,
+      "loss": 0.6733,
+      "step": 12438
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.78188588612276e-06,
+      "loss": 0.7954,
+      "step": 12439
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.780457432922259e-06,
+      "loss": 0.791,
+      "step": 12440
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.779029004981504e-06,
+      "loss": 0.7366,
+      "step": 12441
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.77760060233009e-06,
+      "loss": 0.7573,
+      "step": 12442
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.776172224997593e-06,
+      "loss": 0.7383,
+      "step": 12443
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.774743873013604e-06,
+      "loss": 0.7832,
+      "step": 12444
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.773315546407707e-06,
+      "loss": 0.7285,
+      "step": 12445
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.771887245209481e-06,
+      "loss": 0.8066,
+      "step": 12446
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.770458969448519e-06,
+      "loss": 0.8726,
+      "step": 12447
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.769030719154392e-06,
+      "loss": 0.6843,
+      "step": 12448
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.767602494356695e-06,
+      "loss": 0.6851,
+      "step": 12449
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.766174295085001e-06,
+      "loss": 0.6707,
+      "step": 12450
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.764746121368895e-06,
+      "loss": 0.7769,
+      "step": 12451
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.763317973237959e-06,
+      "loss": 0.7871,
+      "step": 12452
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.761889850721772e-06,
+      "loss": 0.7817,
+      "step": 12453
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.760461753849913e-06,
+      "loss": 0.7783,
+      "step": 12454
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.759033682651966e-06,
+      "loss": 0.7598,
+      "step": 12455
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.757605637157503e-06,
+      "loss": 0.8008,
+      "step": 12456
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.756177617396111e-06,
+      "loss": 0.6899,
+      "step": 12457
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.754749623397359e-06,
+      "loss": 0.6978,
+      "step": 12458
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.75332165519083e-06,
+      "loss": 0.853,
+      "step": 12459
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.751893712806102e-06,
+      "loss": 0.6641,
+      "step": 12460
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.75046579627274e-06,
+      "loss": 0.7734,
+      "step": 12461
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.749037905620334e-06,
+      "loss": 0.8535,
+      "step": 12462
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.747610040878452e-06,
+      "loss": 0.7588,
+      "step": 12463
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.746182202076669e-06,
+      "loss": 0.7588,
+      "step": 12464
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.744754389244557e-06,
+      "loss": 0.7876,
+      "step": 12465
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.743326602411696e-06,
+      "loss": 0.6785,
+      "step": 12466
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.741898841607653e-06,
+      "loss": 0.6895,
+      "step": 12467
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.740471106862006e-06,
+      "loss": 0.8174,
+      "step": 12468
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.73904339820432e-06,
+      "loss": 0.8496,
+      "step": 12469
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.737615715664173e-06,
+      "loss": 0.7236,
+      "step": 12470
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.73618805927113e-06,
+      "loss": 0.6855,
+      "step": 12471
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.734760429054768e-06,
+      "loss": 0.8003,
+      "step": 12472
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.733332825044649e-06,
+      "loss": 0.8103,
+      "step": 12473
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.73190524727035e-06,
+      "loss": 0.7749,
+      "step": 12474
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.730477695761434e-06,
+      "loss": 0.8125,
+      "step": 12475
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.72905017054747e-06,
+      "loss": 0.7188,
+      "step": 12476
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.727622671658028e-06,
+      "loss": 0.6626,
+      "step": 12477
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.726195199122671e-06,
+      "loss": 0.7612,
+      "step": 12478
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.724767752970971e-06,
+      "loss": 0.7817,
+      "step": 12479
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.723340333232488e-06,
+      "loss": 0.707,
+      "step": 12480
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.721912939936795e-06,
+      "loss": 0.7373,
+      "step": 12481
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.720485573113447e-06,
+      "loss": 0.8179,
+      "step": 12482
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71905823279202e-06,
+      "loss": 0.7847,
+      "step": 12483
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.717630919002065e-06,
+      "loss": 0.8037,
+      "step": 12484
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.716203631773155e-06,
+      "loss": 0.8208,
+      "step": 12485
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.714776371134848e-06,
+      "loss": 0.7427,
+      "step": 12486
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.713349137116708e-06,
+      "loss": 0.8569,
+      "step": 12487
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.711921929748294e-06,
+      "loss": 0.7871,
+      "step": 12488
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.710494749059173e-06,
+      "loss": 0.6987,
+      "step": 12489
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.709067595078898e-06,
+      "loss": 0.8276,
+      "step": 12490
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.707640467837031e-06,
+      "loss": 0.7209,
+      "step": 12491
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.706213367363136e-06,
+      "loss": 0.7695,
+      "step": 12492
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.704786293686765e-06,
+      "loss": 0.6897,
+      "step": 12493
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.703359246837484e-06,
+      "loss": 0.6733,
+      "step": 12494
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.701932226844842e-06,
+      "loss": 0.7361,
+      "step": 12495
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.700505233738402e-06,
+      "loss": 0.6555,
+      "step": 12496
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.699078267547719e-06,
+      "loss": 0.6807,
+      "step": 12497
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.69765132830235e-06,
+      "loss": 0.7329,
+      "step": 12498
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.696224416031847e-06,
+      "loss": 0.7544,
+      "step": 12499
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.69479753076577e-06,
+      "loss": 0.8345,
+      "step": 12500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.693370672533667e-06,
+      "loss": 0.793,
+      "step": 12501
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.691943841365097e-06,
+      "loss": 0.8254,
+      "step": 12502
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.690517037289611e-06,
+      "loss": 0.6655,
+      "step": 12503
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.689090260336766e-06,
+      "loss": 0.7319,
+      "step": 12504
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.687663510536109e-06,
+      "loss": 0.7812,
+      "step": 12505
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.686236787917188e-06,
+      "loss": 0.7166,
+      "step": 12506
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.684810092509567e-06,
+      "loss": 0.7153,
+      "step": 12507
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.683383424342779e-06,
+      "loss": 0.6721,
+      "step": 12508
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.68195678344639e-06,
+      "loss": 0.7871,
+      "step": 12509
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.680530169849938e-06,
+      "loss": 0.7886,
+      "step": 12510
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.67910358358298e-06,
+      "loss": 0.7588,
+      "step": 12511
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.677677024675056e-06,
+      "loss": 0.7778,
+      "step": 12512
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.676250493155721e-06,
+      "loss": 0.7344,
+      "step": 12513
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.674823989054514e-06,
+      "loss": 0.7358,
+      "step": 12514
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.67339751240099e-06,
+      "loss": 0.8237,
+      "step": 12515
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.67197106322469e-06,
+      "loss": 0.7554,
+      "step": 12516
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.670544641555161e-06,
+      "loss": 0.7861,
+      "step": 12517
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.669118247421946e-06,
+      "loss": 0.7339,
+      "step": 12518
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.667691880854592e-06,
+      "loss": 0.7842,
+      "step": 12519
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.666265541882643e-06,
+      "loss": 0.7437,
+      "step": 12520
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.664839230535635e-06,
+      "loss": 0.7446,
+      "step": 12521
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.663412946843119e-06,
+      "loss": 0.7793,
+      "step": 12522
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.661986690834628e-06,
+      "loss": 0.7627,
+      "step": 12523
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.660560462539714e-06,
+      "loss": 0.7354,
+      "step": 12524
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.659134261987907e-06,
+      "loss": 0.7163,
+      "step": 12525
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.657708089208757e-06,
+      "loss": 0.8096,
+      "step": 12526
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.656281944231794e-06,
+      "loss": 0.7832,
+      "step": 12527
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.654855827086566e-06,
+      "loss": 0.7083,
+      "step": 12528
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.653429737802606e-06,
+      "loss": 0.7126,
+      "step": 12529
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.652003676409452e-06,
+      "loss": 0.8271,
+      "step": 12530
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.65057764293664e-06,
+      "loss": 0.8027,
+      "step": 12531
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.649151637413715e-06,
+      "loss": 0.7915,
+      "step": 12532
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.647725659870203e-06,
+      "loss": 0.7012,
+      "step": 12533
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.646299710335648e-06,
+      "loss": 0.7654,
+      "step": 12534
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.644873788839578e-06,
+      "loss": 0.6956,
+      "step": 12535
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.643447895411528e-06,
+      "loss": 0.7227,
+      "step": 12536
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.642022030081038e-06,
+      "loss": 0.8125,
+      "step": 12537
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.640596192877633e-06,
+      "loss": 0.6616,
+      "step": 12538
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.63917038383085e-06,
+      "loss": 0.6694,
+      "step": 12539
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.637744602970223e-06,
+      "loss": 0.6729,
+      "step": 12540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.636318850325281e-06,
+      "loss": 0.7954,
+      "step": 12541
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.63489312592555e-06,
+      "loss": 0.7505,
+      "step": 12542
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.633467429800573e-06,
+      "loss": 0.6589,
+      "step": 12543
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.632041761979865e-06,
+      "loss": 0.7075,
+      "step": 12544
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.630616122492967e-06,
+      "loss": 0.7239,
+      "step": 12545
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.629190511369399e-06,
+      "loss": 0.7744,
+      "step": 12546
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.627764928638697e-06,
+      "loss": 0.7944,
+      "step": 12547
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.62633937433038e-06,
+      "loss": 0.853,
+      "step": 12548
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.624913848473982e-06,
+      "loss": 0.7231,
+      "step": 12549
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.623488351099026e-06,
+      "loss": 0.7134,
+      "step": 12550
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.622062882235035e-06,
+      "loss": 0.8569,
+      "step": 12551
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.620637441911535e-06,
+      "loss": 0.7939,
+      "step": 12552
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.619212030158054e-06,
+      "loss": 0.7456,
+      "step": 12553
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.617786647004113e-06,
+      "loss": 0.7529,
+      "step": 12554
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.616361292479234e-06,
+      "loss": 0.7817,
+      "step": 12555
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.614935966612946e-06,
+      "loss": 0.7786,
+      "step": 12556
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.613510669434762e-06,
+      "loss": 0.8052,
+      "step": 12557
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.612085400974212e-06,
+      "loss": 0.748,
+      "step": 12558
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.610660161260808e-06,
+      "loss": 0.7334,
+      "step": 12559
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.609234950324082e-06,
+      "loss": 0.666,
+      "step": 12560
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.607809768193538e-06,
+      "loss": 0.8271,
+      "step": 12561
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.606384614898712e-06,
+      "loss": 0.8003,
+      "step": 12562
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.604959490469108e-06,
+      "loss": 0.7686,
+      "step": 12563
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.603534394934254e-06,
+      "loss": 0.7534,
+      "step": 12564
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.602109328323662e-06,
+      "loss": 0.7939,
+      "step": 12565
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.60068429066685e-06,
+      "loss": 0.8191,
+      "step": 12566
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.599259281993333e-06,
+      "loss": 0.7705,
+      "step": 12567
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.597834302332628e-06,
+      "loss": 0.8101,
+      "step": 12568
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.59640935171425e-06,
+      "loss": 0.6909,
+      "step": 12569
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.594984430167711e-06,
+      "loss": 0.7188,
+      "step": 12570
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.59355953772253e-06,
+      "loss": 0.717,
+      "step": 12571
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.592134674408212e-06,
+      "loss": 0.7466,
+      "step": 12572
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.590709840254277e-06,
+      "loss": 0.7163,
+      "step": 12573
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.58928503529023e-06,
+      "loss": 0.7505,
+      "step": 12574
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.587860259545591e-06,
+      "loss": 0.7354,
+      "step": 12575
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.58643551304986e-06,
+      "loss": 0.7734,
+      "step": 12576
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.585010795832557e-06,
+      "loss": 0.7461,
+      "step": 12577
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.583586107923184e-06,
+      "loss": 0.6836,
+      "step": 12578
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.582161449351254e-06,
+      "loss": 0.7356,
+      "step": 12579
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.580736820146274e-06,
+      "loss": 0.7729,
+      "step": 12580
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.57931222033775e-06,
+      "loss": 0.8418,
+      "step": 12581
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.577887649955195e-06,
+      "loss": 0.7417,
+      "step": 12582
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.576463109028104e-06,
+      "loss": 0.8286,
+      "step": 12583
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.575038597585995e-06,
+      "loss": 0.7747,
+      "step": 12584
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.573614115658367e-06,
+      "loss": 0.6777,
+      "step": 12585
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.572189663274725e-06,
+      "loss": 0.8496,
+      "step": 12586
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.570765240464572e-06,
+      "loss": 0.7891,
+      "step": 12587
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.569340847257417e-06,
+      "loss": 0.7476,
+      "step": 12588
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.567916483682752e-06,
+      "loss": 0.8271,
+      "step": 12589
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.566492149770092e-06,
+      "loss": 0.7407,
+      "step": 12590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.565067845548927e-06,
+      "loss": 0.7036,
+      "step": 12591
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.563643571048764e-06,
+      "loss": 0.7935,
+      "step": 12592
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.562219326299103e-06,
+      "loss": 0.7366,
+      "step": 12593
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.560795111329442e-06,
+      "loss": 0.7717,
+      "step": 12594
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.559370926169285e-06,
+      "loss": 0.6636,
+      "step": 12595
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55794677084812e-06,
+      "loss": 0.7283,
+      "step": 12596
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.556522645395455e-06,
+      "loss": 0.8008,
+      "step": 12597
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55509854984078e-06,
+      "loss": 0.7788,
+      "step": 12598
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.553674484213599e-06,
+      "loss": 0.6826,
+      "step": 12599
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.552250448543398e-06,
+      "loss": 0.731,
+      "step": 12600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.550826442859685e-06,
+      "loss": 0.6909,
+      "step": 12601
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.54940246719194e-06,
+      "loss": 0.7949,
+      "step": 12602
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.547978521569672e-06,
+      "loss": 0.7598,
+      "step": 12603
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.54655460602236e-06,
+      "loss": 0.7583,
+      "step": 12604
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.545130720579508e-06,
+      "loss": 0.7363,
+      "step": 12605
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.543706865270604e-06,
+      "loss": 0.8257,
+      "step": 12606
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.54228304012514e-06,
+      "loss": 0.772,
+      "step": 12607
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.540859245172602e-06,
+      "loss": 0.7412,
+      "step": 12608
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.53943548044249e-06,
+      "loss": 0.8091,
+      "step": 12609
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.53801174596429e-06,
+      "loss": 0.7559,
+      "step": 12610
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.536588041767485e-06,
+      "loss": 0.7507,
+      "step": 12611
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.535164367881573e-06,
+      "loss": 0.667,
+      "step": 12612
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.533740724336032e-06,
+      "loss": 0.7139,
+      "step": 12613
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.53231711116036e-06,
+      "loss": 0.7505,
+      "step": 12614
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.530893528384032e-06,
+      "loss": 0.8247,
+      "step": 12615
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.529469976036542e-06,
+      "loss": 0.7185,
+      "step": 12616
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.528046454147373e-06,
+      "loss": 0.7949,
+      "step": 12617
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.526622962746009e-06,
+      "loss": 0.7793,
+      "step": 12618
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.525199501861934e-06,
+      "loss": 0.856,
+      "step": 12619
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.523776071524635e-06,
+      "loss": 0.7109,
+      "step": 12620
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.522352671763587e-06,
+      "loss": 0.7666,
+      "step": 12621
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.520929302608283e-06,
+      "loss": 0.7563,
+      "step": 12622
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.519505964088193e-06,
+      "loss": 0.7559,
+      "step": 12623
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.518082656232809e-06,
+      "loss": 0.6946,
+      "step": 12624
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.516659379071605e-06,
+      "loss": 0.8052,
+      "step": 12625
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.515236132634056e-06,
+      "loss": 0.749,
+      "step": 12626
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.513812916949651e-06,
+      "loss": 0.689,
+      "step": 12627
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.512389732047859e-06,
+      "loss": 0.7891,
+      "step": 12628
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.510966577958163e-06,
+      "loss": 0.7861,
+      "step": 12629
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.509543454710042e-06,
+      "loss": 0.7051,
+      "step": 12630
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.508120362332969e-06,
+      "loss": 0.6841,
+      "step": 12631
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.506697300856417e-06,
+      "loss": 0.748,
+      "step": 12632
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.505274270309869e-06,
+      "loss": 0.7141,
+      "step": 12633
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.503851270722791e-06,
+      "loss": 0.6931,
+      "step": 12634
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.502428302124665e-06,
+      "loss": 0.9341,
+      "step": 12635
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.501005364544956e-06,
+      "loss": 0.791,
+      "step": 12636
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.499582458013146e-06,
+      "loss": 0.7188,
+      "step": 12637
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.498159582558696e-06,
+      "loss": 0.7749,
+      "step": 12638
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.496736738211087e-06,
+      "loss": 0.678,
+      "step": 12639
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.495313924999784e-06,
+      "loss": 0.6658,
+      "step": 12640
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.493891142954256e-06,
+      "loss": 0.8984,
+      "step": 12641
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.492468392103977e-06,
+      "loss": 0.7549,
+      "step": 12642
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.491045672478414e-06,
+      "loss": 0.6709,
+      "step": 12643
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.489622984107031e-06,
+      "loss": 0.6565,
+      "step": 12644
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.4882003270193e-06,
+      "loss": 0.7517,
+      "step": 12645
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.486777701244691e-06,
+      "loss": 0.7788,
+      "step": 12646
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.485355106812661e-06,
+      "loss": 0.8296,
+      "step": 12647
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.483932543752685e-06,
+      "loss": 0.7871,
+      "step": 12648
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.482510012094217e-06,
+      "loss": 0.8037,
+      "step": 12649
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.481087511866732e-06,
+      "loss": 0.8184,
+      "step": 12650
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.479665043099685e-06,
+      "loss": 0.7915,
+      "step": 12651
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.478242605822546e-06,
+      "loss": 0.7588,
+      "step": 12652
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.476820200064774e-06,
+      "loss": 0.7793,
+      "step": 12653
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.475397825855828e-06,
+      "loss": 0.7773,
+      "step": 12654
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.473975483225172e-06,
+      "loss": 0.7324,
+      "step": 12655
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.472553172202263e-06,
+      "loss": 0.7324,
+      "step": 12656
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.471130892816567e-06,
+      "loss": 0.7852,
+      "step": 12657
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.469708645097534e-06,
+      "loss": 0.7444,
+      "step": 12658
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.468286429074634e-06,
+      "loss": 0.6558,
+      "step": 12659
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.466864244777312e-06,
+      "loss": 0.6992,
+      "step": 12660
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.465442092235036e-06,
+      "loss": 0.7402,
+      "step": 12661
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.46401997147725e-06,
+      "loss": 0.7573,
+      "step": 12662
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.462597882533424e-06,
+      "loss": 0.8159,
+      "step": 12663
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.461175825433002e-06,
+      "loss": 0.7275,
+      "step": 12664
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.459753800205444e-06,
+      "loss": 0.6929,
+      "step": 12665
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.458331806880197e-06,
+      "loss": 0.7231,
+      "step": 12666
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.456909845486726e-06,
+      "loss": 0.7754,
+      "step": 12667
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.455487916054472e-06,
+      "loss": 0.7612,
+      "step": 12668
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.454066018612888e-06,
+      "loss": 0.8042,
+      "step": 12669
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.452644153191434e-06,
+      "loss": 0.8296,
+      "step": 12670
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.451222319819549e-06,
+      "loss": 0.7102,
+      "step": 12671
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.449800518526692e-06,
+      "loss": 0.7715,
+      "step": 12672
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.448378749342303e-06,
+      "loss": 0.7964,
+      "step": 12673
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.44695701229584e-06,
+      "loss": 0.8191,
+      "step": 12674
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.445535307416742e-06,
+      "loss": 0.7185,
+      "step": 12675
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.444113634734463e-06,
+      "loss": 0.8149,
+      "step": 12676
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.442691994278444e-06,
+      "loss": 0.7065,
+      "step": 12677
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.441270386078137e-06,
+      "loss": 0.7695,
+      "step": 12678
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.439848810162977e-06,
+      "loss": 0.7529,
+      "step": 12679
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.43842726656242e-06,
+      "loss": 0.8162,
+      "step": 12680
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.437005755305899e-06,
+      "loss": 0.7063,
+      "step": 12681
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.435584276422865e-06,
+      "loss": 0.7817,
+      "step": 12682
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.434162829942759e-06,
+      "loss": 0.7412,
+      "step": 12683
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.432741415895017e-06,
+      "loss": 0.7168,
+      "step": 12684
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.43132003430909e-06,
+      "loss": 0.8032,
+      "step": 12685
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.429898685214406e-06,
+      "loss": 0.7478,
+      "step": 12686
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.428477368640417e-06,
+      "loss": 0.8042,
+      "step": 12687
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.42705608461655e-06,
+      "loss": 0.7031,
+      "step": 12688
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.425634833172257e-06,
+      "loss": 0.7012,
+      "step": 12689
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.424213614336961e-06,
+      "loss": 0.7856,
+      "step": 12690
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.42279242814011e-06,
+      "loss": 0.8398,
+      "step": 12691
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.421371274611134e-06,
+      "loss": 0.7222,
+      "step": 12692
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.41995015377947e-06,
+      "loss": 0.7068,
+      "step": 12693
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.418529065674557e-06,
+      "loss": 0.7759,
+      "step": 12694
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.417108010325826e-06,
+      "loss": 0.8755,
+      "step": 12695
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.415686987762708e-06,
+      "loss": 0.6648,
+      "step": 12696
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.414265998014642e-06,
+      "loss": 0.7583,
+      "step": 12697
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.412845041111057e-06,
+      "loss": 0.8174,
+      "step": 12698
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.411424117081381e-06,
+      "loss": 0.8408,
+      "step": 12699
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.410003225955052e-06,
+      "loss": 0.6758,
+      "step": 12700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.408582367761492e-06,
+      "loss": 0.8159,
+      "step": 12701
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.40716154253014e-06,
+      "loss": 0.7402,
+      "step": 12702
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.405740750290415e-06,
+      "loss": 0.7607,
+      "step": 12703
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.404319991071754e-06,
+      "loss": 0.8228,
+      "step": 12704
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.402899264903577e-06,
+      "loss": 0.7134,
+      "step": 12705
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.401478571815314e-06,
+      "loss": 0.7988,
+      "step": 12706
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.400057911836392e-06,
+      "loss": 0.7563,
+      "step": 12707
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.398637284996237e-06,
+      "loss": 0.7407,
+      "step": 12708
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.397216691324268e-06,
+      "loss": 0.7834,
+      "step": 12709
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.39579613084992e-06,
+      "loss": 0.7383,
+      "step": 12710
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.394375603602602e-06,
+      "loss": 0.6968,
+      "step": 12711
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.392955109611751e-06,
+      "loss": 0.8081,
+      "step": 12712
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.391534648906781e-06,
+      "loss": 0.7324,
+      "step": 12713
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.390114221517111e-06,
+      "loss": 0.6394,
+      "step": 12714
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.388693827472169e-06,
+      "loss": 0.6763,
+      "step": 12715
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.387273466801366e-06,
+      "loss": 0.728,
+      "step": 12716
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.38585313953413e-06,
+      "loss": 0.731,
+      "step": 12717
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.384432845699873e-06,
+      "loss": 0.6392,
+      "step": 12718
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.383012585328017e-06,
+      "loss": 0.7162,
+      "step": 12719
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.381592358447976e-06,
+      "loss": 0.8086,
+      "step": 12720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.38017216508917e-06,
+      "loss": 0.875,
+      "step": 12721
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.378752005281009e-06,
+      "loss": 0.7686,
+      "step": 12722
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.377331879052915e-06,
+      "loss": 0.8076,
+      "step": 12723
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.375911786434297e-06,
+      "loss": 0.7583,
+      "step": 12724
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.374491727454571e-06,
+      "loss": 0.7119,
+      "step": 12725
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.373071702143149e-06,
+      "loss": 0.7671,
+      "step": 12726
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.371651710529448e-06,
+      "loss": 0.7383,
+      "step": 12727
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.370231752642875e-06,
+      "loss": 0.7842,
+      "step": 12728
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.368811828512835e-06,
+      "loss": 0.7734,
+      "step": 12729
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.367391938168748e-06,
+      "loss": 0.6914,
+      "step": 12730
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36597208164002e-06,
+      "loss": 0.7637,
+      "step": 12731
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36455225895606e-06,
+      "loss": 0.8379,
+      "step": 12732
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.363132470146271e-06,
+      "loss": 0.7822,
+      "step": 12733
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.361712715240072e-06,
+      "loss": 0.7651,
+      "step": 12734
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.360292994266857e-06,
+      "loss": 0.6824,
+      "step": 12735
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.358873307256042e-06,
+      "loss": 0.7734,
+      "step": 12736
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.357453654237023e-06,
+      "loss": 0.7163,
+      "step": 12737
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.356034035239214e-06,
+      "loss": 0.7056,
+      "step": 12738
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.35461445029201e-06,
+      "loss": 0.7456,
+      "step": 12739
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.353194899424822e-06,
+      "loss": 0.8682,
+      "step": 12740
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.351775382667043e-06,
+      "loss": 0.7759,
+      "step": 12741
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.350355900048086e-06,
+      "loss": 0.7271,
+      "step": 12742
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.348936451597345e-06,
+      "loss": 0.8359,
+      "step": 12743
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.34751703734422e-06,
+      "loss": 0.7534,
+      "step": 12744
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.346097657318114e-06,
+      "loss": 0.7158,
+      "step": 12745
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.344678311548421e-06,
+      "loss": 0.7759,
+      "step": 12746
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.343259000064547e-06,
+      "loss": 0.7397,
+      "step": 12747
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.34183972289588e-06,
+      "loss": 0.5969,
+      "step": 12748
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.340420480071825e-06,
+      "loss": 0.7969,
+      "step": 12749
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.339001271621772e-06,
+      "loss": 0.7446,
+      "step": 12750
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.337582097575123e-06,
+      "loss": 0.791,
+      "step": 12751
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.336162957961262e-06,
+      "loss": 0.7524,
+      "step": 12752
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.334743852809596e-06,
+      "loss": 0.7808,
+      "step": 12753
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.333324782149506e-06,
+      "loss": 0.7698,
+      "step": 12754
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.331905746010395e-06,
+      "loss": 0.7241,
+      "step": 12755
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.330486744421644e-06,
+      "loss": 0.7366,
+      "step": 12756
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.329067777412656e-06,
+      "loss": 0.7551,
+      "step": 12757
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.327648845012812e-06,
+      "loss": 0.7324,
+      "step": 12758
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.326229947251503e-06,
+      "loss": 0.7612,
+      "step": 12759
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.324811084158123e-06,
+      "loss": 0.7373,
+      "step": 12760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.323392255762053e-06,
+      "loss": 0.6914,
+      "step": 12761
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.321973462092689e-06,
+      "loss": 0.8276,
+      "step": 12762
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.32055470317941e-06,
+      "loss": 0.7571,
+      "step": 12763
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.319135979051606e-06,
+      "loss": 0.7239,
+      "step": 12764
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.317717289738658e-06,
+      "loss": 0.8359,
+      "step": 12765
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.31629863526996e-06,
+      "loss": 0.6665,
+      "step": 12766
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.314880015674883e-06,
+      "loss": 0.7598,
+      "step": 12767
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.313461430982823e-06,
+      "loss": 0.6023,
+      "step": 12768
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.31204288122315e-06,
+      "loss": 0.7524,
+      "step": 12769
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.310624366425256e-06,
+      "loss": 0.6582,
+      "step": 12770
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.309205886618517e-06,
+      "loss": 0.6826,
+      "step": 12771
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.307787441832314e-06,
+      "loss": 0.7471,
+      "step": 12772
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.306369032096029e-06,
+      "loss": 0.856,
+      "step": 12773
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.304950657439034e-06,
+      "loss": 0.7776,
+      "step": 12774
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.303532317890715e-06,
+      "loss": 0.8022,
+      "step": 12775
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.302114013480443e-06,
+      "loss": 0.7847,
+      "step": 12776
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.300695744237601e-06,
+      "loss": 0.7117,
+      "step": 12777
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.299277510191557e-06,
+      "loss": 0.7588,
+      "step": 12778
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.297859311371695e-06,
+      "loss": 0.6992,
+      "step": 12779
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.29644114780738e-06,
+      "loss": 0.7939,
+      "step": 12780
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.295023019527995e-06,
+      "loss": 0.8428,
+      "step": 12781
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.293604926562904e-06,
+      "loss": 0.7656,
+      "step": 12782
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.292186868941485e-06,
+      "loss": 0.791,
+      "step": 12783
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.290768846693109e-06,
+      "loss": 0.6572,
+      "step": 12784
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.289350859847145e-06,
+      "loss": 0.6853,
+      "step": 12785
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.287932908432963e-06,
+      "loss": 0.7437,
+      "step": 12786
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.286514992479937e-06,
+      "loss": 0.6956,
+      "step": 12787
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.285097112017432e-06,
+      "loss": 0.7939,
+      "step": 12788
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.283679267074812e-06,
+      "loss": 0.7773,
+      "step": 12789
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.28226145768145e-06,
+      "loss": 0.7686,
+      "step": 12790
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.280843683866707e-06,
+      "loss": 0.7285,
+      "step": 12791
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.279425945659955e-06,
+      "loss": 0.7378,
+      "step": 12792
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.278008243090551e-06,
+      "loss": 0.7437,
+      "step": 12793
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.27659057618787e-06,
+      "loss": 0.7515,
+      "step": 12794
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.275172944981262e-06,
+      "loss": 0.6868,
+      "step": 12795
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.273755349500101e-06,
+      "loss": 0.6909,
+      "step": 12796
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.272337789773744e-06,
+      "loss": 0.7681,
+      "step": 12797
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.270920265831552e-06,
+      "loss": 0.7549,
+      "step": 12798
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.269502777702885e-06,
+      "loss": 0.7681,
+      "step": 12799
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.268085325417108e-06,
+      "loss": 0.7441,
+      "step": 12800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.26666790900357e-06,
+      "loss": 0.7122,
+      "step": 12801
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.26525052849164e-06,
+      "loss": 0.6809,
+      "step": 12802
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.263833183910672e-06,
+      "loss": 0.7671,
+      "step": 12803
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.262415875290016e-06,
+      "loss": 0.6941,
+      "step": 12804
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.260998602659037e-06,
+      "loss": 0.7993,
+      "step": 12805
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.259581366047083e-06,
+      "loss": 0.749,
+      "step": 12806
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.258164165483514e-06,
+      "loss": 0.812,
+      "step": 12807
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.25674700099768e-06,
+      "loss": 0.6829,
+      "step": 12808
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.255329872618937e-06,
+      "loss": 0.7034,
+      "step": 12809
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.253912780376633e-06,
+      "loss": 0.7227,
+      "step": 12810
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.252495724300129e-06,
+      "loss": 0.6946,
+      "step": 12811
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.251078704418762e-06,
+      "loss": 0.7134,
+      "step": 12812
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.249661720761895e-06,
+      "loss": 0.8389,
+      "step": 12813
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.248244773358866e-06,
+      "loss": 0.7651,
+      "step": 12814
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.246827862239033e-06,
+      "loss": 0.7021,
+      "step": 12815
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.245410987431736e-06,
+      "loss": 0.689,
+      "step": 12816
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.24399414896633e-06,
+      "loss": 0.7124,
+      "step": 12817
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.242577346872155e-06,
+      "loss": 0.7292,
+      "step": 12818
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.241160581178557e-06,
+      "loss": 0.7778,
+      "step": 12819
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.239743851914882e-06,
+      "loss": 0.749,
+      "step": 12820
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.238327159110475e-06,
+      "loss": 0.7285,
+      "step": 12821
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.236910502794678e-06,
+      "loss": 0.7095,
+      "step": 12822
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.23549388299683e-06,
+      "loss": 0.7512,
+      "step": 12823
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.234077299746283e-06,
+      "loss": 0.7832,
+      "step": 12824
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.232660753072366e-06,
+      "loss": 0.6316,
+      "step": 12825
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.231244243004429e-06,
+      "loss": 0.7788,
+      "step": 12826
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2298277695718e-06,
+      "loss": 0.7598,
+      "step": 12827
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.228411332803833e-06,
+      "loss": 0.7939,
+      "step": 12828
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.226994932729852e-06,
+      "loss": 0.7373,
+      "step": 12829
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.225578569379202e-06,
+      "loss": 0.7222,
+      "step": 12830
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.224162242781216e-06,
+      "loss": 0.7593,
+      "step": 12831
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.222745952965232e-06,
+      "loss": 0.6538,
+      "step": 12832
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.221329699960583e-06,
+      "loss": 0.7666,
+      "step": 12833
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.219913483796603e-06,
+      "loss": 0.7007,
+      "step": 12834
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.218497304502627e-06,
+      "loss": 0.8569,
+      "step": 12835
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.217081162107984e-06,
+      "loss": 0.7131,
+      "step": 12836
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.215665056642012e-06,
+      "loss": 0.7944,
+      "step": 12837
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.214248988134035e-06,
+      "loss": 0.7419,
+      "step": 12838
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.21283295661339e-06,
+      "loss": 0.708,
+      "step": 12839
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.211416962109401e-06,
+      "loss": 0.6531,
+      "step": 12840
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.210001004651402e-06,
+      "loss": 0.8037,
+      "step": 12841
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.208585084268715e-06,
+      "loss": 0.7614,
+      "step": 12842
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.207169200990673e-06,
+      "loss": 0.7144,
+      "step": 12843
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.205753354846596e-06,
+      "loss": 0.7725,
+      "step": 12844
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.204337545865817e-06,
+      "loss": 0.7283,
+      "step": 12845
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.202921774077654e-06,
+      "loss": 0.7617,
+      "step": 12846
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.201506039511438e-06,
+      "loss": 0.7402,
+      "step": 12847
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.200090342196488e-06,
+      "loss": 0.7178,
+      "step": 12848
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.198674682162125e-06,
+      "loss": 0.7209,
+      "step": 12849
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.197259059437677e-06,
+      "loss": 0.7163,
+      "step": 12850
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.195843474052456e-06,
+      "loss": 0.8081,
+      "step": 12851
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.194427926035794e-06,
+      "loss": 0.8633,
+      "step": 12852
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.193012415417e-06,
+      "loss": 0.7485,
+      "step": 12853
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.191596942225399e-06,
+      "loss": 0.7944,
+      "step": 12854
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.190181506490303e-06,
+      "loss": 0.7095,
+      "step": 12855
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.188766108241038e-06,
+      "loss": 0.7397,
+      "step": 12856
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.18735074750691e-06,
+      "loss": 0.8267,
+      "step": 12857
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.185935424317245e-06,
+      "loss": 0.6929,
+      "step": 12858
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.184520138701348e-06,
+      "loss": 0.7217,
+      "step": 12859
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.183104890688541e-06,
+      "loss": 0.688,
+      "step": 12860
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.181689680308133e-06,
+      "loss": 0.7466,
+      "step": 12861
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.18027450758944e-06,
+      "loss": 0.7495,
+      "step": 12862
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.178859372561772e-06,
+      "loss": 0.7964,
+      "step": 12863
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.177444275254435e-06,
+      "loss": 0.7339,
+      "step": 12864
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.176029215696747e-06,
+      "loss": 0.7676,
+      "step": 12865
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.174614193918012e-06,
+      "loss": 0.7522,
+      "step": 12866
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.173199209947544e-06,
+      "loss": 0.8228,
+      "step": 12867
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.171784263814642e-06,
+      "loss": 0.7432,
+      "step": 12868
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.170369355548625e-06,
+      "loss": 0.8379,
+      "step": 12869
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.168954485178787e-06,
+      "loss": 0.8503,
+      "step": 12870
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.167539652734442e-06,
+      "loss": 0.7856,
+      "step": 12871
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16612485824489e-06,
+      "loss": 0.7896,
+      "step": 12872
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.164710101739439e-06,
+      "loss": 0.7454,
+      "step": 12873
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.163295383247389e-06,
+      "loss": 0.7383,
+      "step": 12874
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.161880702798042e-06,
+      "loss": 0.7446,
+      "step": 12875
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.160466060420704e-06,
+      "loss": 0.7622,
+      "step": 12876
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.15905145614467e-06,
+      "loss": 0.698,
+      "step": 12877
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.157636889999244e-06,
+      "loss": 0.6816,
+      "step": 12878
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.15622236201372e-06,
+      "loss": 0.6553,
+      "step": 12879
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.154807872217405e-06,
+      "loss": 0.7637,
+      "step": 12880
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.153393420639585e-06,
+      "loss": 0.7354,
+      "step": 12881
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.151979007309571e-06,
+      "loss": 0.8345,
+      "step": 12882
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.150564632256645e-06,
+      "loss": 0.7651,
+      "step": 12883
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.149150295510112e-06,
+      "loss": 0.7207,
+      "step": 12884
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.147735997099261e-06,
+      "loss": 0.7319,
+      "step": 12885
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.14632173705339e-06,
+      "loss": 0.7471,
+      "step": 12886
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.144907515401788e-06,
+      "loss": 0.6682,
+      "step": 12887
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.14349333217375e-06,
+      "loss": 0.7778,
+      "step": 12888
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.142079187398562e-06,
+      "loss": 0.7393,
+      "step": 12889
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.140665081105525e-06,
+      "loss": 0.7227,
+      "step": 12890
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.13925101332392e-06,
+      "loss": 0.7439,
+      "step": 12891
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.137836984083033e-06,
+      "loss": 0.7378,
+      "step": 12892
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.13642299341216e-06,
+      "loss": 0.679,
+      "step": 12893
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.135009041340584e-06,
+      "loss": 0.7681,
+      "step": 12894
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.133595127897595e-06,
+      "loss": 0.7871,
+      "step": 12895
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.132181253112472e-06,
+      "loss": 0.7979,
+      "step": 12896
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.130767417014506e-06,
+      "loss": 0.7346,
+      "step": 12897
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.129353619632978e-06,
+      "loss": 0.7837,
+      "step": 12898
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.127939860997174e-06,
+      "loss": 0.7993,
+      "step": 12899
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.12652614113637e-06,
+      "loss": 0.6426,
+      "step": 12900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.12511246007986e-06,
+      "loss": 0.7705,
+      "step": 12901
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.12369881785691e-06,
+      "loss": 0.832,
+      "step": 12902
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.122285214496814e-06,
+      "loss": 0.7493,
+      "step": 12903
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.120871650028838e-06,
+      "loss": 0.7485,
+      "step": 12904
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.119458124482272e-06,
+      "loss": 0.834,
+      "step": 12905
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.118044637886389e-06,
+      "loss": 0.728,
+      "step": 12906
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.11663119027046e-06,
+      "loss": 0.7598,
+      "step": 12907
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.115217781663769e-06,
+      "loss": 0.8613,
+      "step": 12908
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.113804412095589e-06,
+      "loss": 0.7493,
+      "step": 12909
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.112391081595193e-06,
+      "loss": 0.7129,
+      "step": 12910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.110977790191856e-06,
+      "loss": 0.7896,
+      "step": 12911
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.109564537914849e-06,
+      "loss": 0.7612,
+      "step": 12912
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.108151324793444e-06,
+      "loss": 0.7188,
+      "step": 12913
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.106738150856917e-06,
+      "loss": 0.8193,
+      "step": 12914
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.105325016134531e-06,
+      "loss": 0.6501,
+      "step": 12915
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.103911920655564e-06,
+      "loss": 0.7407,
+      "step": 12916
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.102498864449275e-06,
+      "loss": 0.8042,
+      "step": 12917
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.10108584754494e-06,
+      "loss": 0.6912,
+      "step": 12918
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.099672869971819e-06,
+      "loss": 0.7495,
+      "step": 12919
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.098259931759188e-06,
+      "loss": 0.7458,
+      "step": 12920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.096847032936304e-06,
+      "loss": 0.8198,
+      "step": 12921
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.095434173532433e-06,
+      "loss": 0.7603,
+      "step": 12922
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.094021353576841e-06,
+      "loss": 0.8076,
+      "step": 12923
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.092608573098787e-06,
+      "loss": 0.7651,
+      "step": 12924
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.09119583212754e-06,
+      "loss": 0.7495,
+      "step": 12925
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.089783130692356e-06,
+      "loss": 0.7393,
+      "step": 12926
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.088370468822499e-06,
+      "loss": 0.6741,
+      "step": 12927
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.086957846547223e-06,
+      "loss": 0.8125,
+      "step": 12928
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.085545263895795e-06,
+      "loss": 0.9292,
+      "step": 12929
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.084132720897466e-06,
+      "loss": 0.708,
+      "step": 12930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.0827202175815e-06,
+      "loss": 0.802,
+      "step": 12931
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.081307753977143e-06,
+      "loss": 0.7534,
+      "step": 12932
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.079895330113664e-06,
+      "loss": 0.7407,
+      "step": 12933
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.078482946020305e-06,
+      "loss": 0.8306,
+      "step": 12934
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.077070601726333e-06,
+      "loss": 0.7241,
+      "step": 12935
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.075658297260988e-06,
+      "loss": 0.6997,
+      "step": 12936
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.074246032653529e-06,
+      "loss": 0.769,
+      "step": 12937
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.072833807933212e-06,
+      "loss": 0.6997,
+      "step": 12938
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.071421623129276e-06,
+      "loss": 0.7703,
+      "step": 12939
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.070009478270982e-06,
+      "loss": 0.7175,
+      "step": 12940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.06859737338757e-06,
+      "loss": 0.8262,
+      "step": 12941
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.067185308508299e-06,
+      "loss": 0.7007,
+      "step": 12942
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.065773283662404e-06,
+      "loss": 0.7129,
+      "step": 12943
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.064361298879143e-06,
+      "loss": 0.8286,
+      "step": 12944
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.062949354187751e-06,
+      "loss": 0.6724,
+      "step": 12945
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.061537449617482e-06,
+      "loss": 0.7627,
+      "step": 12946
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.060125585197573e-06,
+      "loss": 0.8481,
+      "step": 12947
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.058713760957275e-06,
+      "loss": 0.6792,
+      "step": 12948
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.057301976925822e-06,
+      "loss": 0.8384,
+      "step": 12949
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.055890233132459e-06,
+      "loss": 0.6792,
+      "step": 12950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.05447852960643e-06,
+      "loss": 0.7263,
+      "step": 12951
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.053066866376969e-06,
+      "loss": 0.8267,
+      "step": 12952
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.05165524347332e-06,
+      "loss": 0.7117,
+      "step": 12953
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.050243660924717e-06,
+      "loss": 0.6882,
+      "step": 12954
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.048832118760403e-06,
+      "loss": 0.731,
+      "step": 12955
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.047420617009607e-06,
+      "loss": 0.7485,
+      "step": 12956
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.046009155701573e-06,
+      "loss": 0.6599,
+      "step": 12957
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.044597734865527e-06,
+      "loss": 0.7129,
+      "step": 12958
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.043186354530712e-06,
+      "loss": 0.8496,
+      "step": 12959
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.041775014726354e-06,
+      "loss": 0.7256,
+      "step": 12960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.040363715481689e-06,
+      "loss": 0.8076,
+      "step": 12961
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.038952456825947e-06,
+      "loss": 0.7905,
+      "step": 12962
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03754123878836e-06,
+      "loss": 0.7725,
+      "step": 12963
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.036130061398155e-06,
+      "loss": 0.8115,
+      "step": 12964
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.034718924684568e-06,
+      "loss": 0.7397,
+      "step": 12965
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.033307828676822e-06,
+      "loss": 0.7295,
+      "step": 12966
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03189677340414e-06,
+      "loss": 0.7385,
+      "step": 12967
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.030485758895757e-06,
+      "loss": 0.7661,
+      "step": 12968
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.02907478518089e-06,
+      "loss": 0.7671,
+      "step": 12969
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.027663852288775e-06,
+      "loss": 0.7021,
+      "step": 12970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.026252960248623e-06,
+      "loss": 0.8506,
+      "step": 12971
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.02484210908967e-06,
+      "loss": 0.7617,
+      "step": 12972
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.023431298841125e-06,
+      "loss": 0.6855,
+      "step": 12973
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.022020529532217e-06,
+      "loss": 0.7458,
+      "step": 12974
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.020609801192168e-06,
+      "loss": 0.8052,
+      "step": 12975
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.019199113850195e-06,
+      "loss": 0.7432,
+      "step": 12976
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.017788467535514e-06,
+      "loss": 0.7441,
+      "step": 12977
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.016377862277352e-06,
+      "loss": 0.7324,
+      "step": 12978
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.014967298104913e-06,
+      "loss": 0.7849,
+      "step": 12979
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.013556775047426e-06,
+      "loss": 0.7334,
+      "step": 12980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.012146293134101e-06,
+      "loss": 0.7676,
+      "step": 12981
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.01073585239415e-06,
+      "loss": 0.7437,
+      "step": 12982
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.009325452856792e-06,
+      "loss": 0.7864,
+      "step": 12983
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.007915094551231e-06,
+      "loss": 0.6953,
+      "step": 12984
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.00650477750669e-06,
+      "loss": 0.7358,
+      "step": 12985
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.005094501752373e-06,
+      "loss": 0.8052,
+      "step": 12986
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.003684267317493e-06,
+      "loss": 0.6782,
+      "step": 12987
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.00227407423126e-06,
+      "loss": 0.7886,
+      "step": 12988
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.00086392252288e-06,
+      "loss": 0.6685,
+      "step": 12989
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.999453812221562e-06,
+      "loss": 0.6855,
+      "step": 12990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.998043743356516e-06,
+      "loss": 0.8081,
+      "step": 12991
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.99663371595694e-06,
+      "loss": 0.7656,
+      "step": 12992
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.995223730052051e-06,
+      "loss": 0.7334,
+      "step": 12993
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.993813785671041e-06,
+      "loss": 0.8545,
+      "step": 12994
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.992403882843125e-06,
+      "loss": 0.7705,
+      "step": 12995
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.990994021597498e-06,
+      "loss": 0.7407,
+      "step": 12996
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.98958420196336e-06,
+      "loss": 0.7212,
+      "step": 12997
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.988174423969917e-06,
+      "loss": 0.8101,
+      "step": 12998
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.986764687646367e-06,
+      "loss": 0.731,
+      "step": 12999
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.985354993021909e-06,
+      "loss": 0.7603,
+      "step": 13000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.98394534012574e-06,
+      "loss": 0.7437,
+      "step": 13001
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.982535728987064e-06,
+      "loss": 0.7126,
+      "step": 13002
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.981126159635068e-06,
+      "loss": 0.6587,
+      "step": 13003
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.979716632098957e-06,
+      "loss": 0.7363,
+      "step": 13004
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.978307146407916e-06,
+      "loss": 0.7905,
+      "step": 13005
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.976897702591146e-06,
+      "loss": 0.8071,
+      "step": 13006
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.975488300677837e-06,
+      "loss": 0.8105,
+      "step": 13007
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.974078940697187e-06,
+      "loss": 0.7473,
+      "step": 13008
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.972669622678376e-06,
+      "loss": 0.7124,
+      "step": 13009
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.971260346650607e-06,
+      "loss": 0.8174,
+      "step": 13010
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.969851112643062e-06,
+      "loss": 0.7771,
+      "step": 13011
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.968441920684929e-06,
+      "loss": 0.7759,
+      "step": 13012
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.9670327708054e-06,
+      "loss": 0.7935,
+      "step": 13013
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.965623663033657e-06,
+      "loss": 0.813,
+      "step": 13014
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.964214597398896e-06,
+      "loss": 0.8042,
+      "step": 13015
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.96280557393029e-06,
+      "loss": 0.7666,
+      "step": 13016
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.961396592657034e-06,
+      "loss": 0.6682,
+      "step": 13017
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.959987653608301e-06,
+      "loss": 0.8428,
+      "step": 13018
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.958578756813283e-06,
+      "loss": 0.7292,
+      "step": 13019
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.957169902301155e-06,
+      "loss": 0.7087,
+      "step": 13020
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.955761090101105e-06,
+      "loss": 0.7056,
+      "step": 13021
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.954352320242303e-06,
+      "loss": 0.7197,
+      "step": 13022
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.952943592753938e-06,
+      "loss": 0.8174,
+      "step": 13023
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.951534907665181e-06,
+      "loss": 0.751,
+      "step": 13024
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.950126265005218e-06,
+      "loss": 0.7964,
+      "step": 13025
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.948717664803216e-06,
+      "loss": 0.8462,
+      "step": 13026
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.947309107088352e-06,
+      "loss": 0.7021,
+      "step": 13027
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.945900591889808e-06,
+      "loss": 0.7173,
+      "step": 13028
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.94449211923675e-06,
+      "loss": 0.7749,
+      "step": 13029
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.943083689158358e-06,
+      "loss": 0.7612,
+      "step": 13030
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.941675301683793e-06,
+      "loss": 0.6855,
+      "step": 13031
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.94026695684224e-06,
+      "loss": 0.6973,
+      "step": 13032
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.938858654662859e-06,
+      "loss": 0.7827,
+      "step": 13033
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.937450395174825e-06,
+      "loss": 0.8213,
+      "step": 13034
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.936042178407302e-06,
+      "loss": 0.8188,
+      "step": 13035
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.934634004389464e-06,
+      "loss": 0.7246,
+      "step": 13036
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.93322587315047e-06,
+      "loss": 0.6121,
+      "step": 13037
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.931817784719492e-06,
+      "loss": 0.6953,
+      "step": 13038
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.930409739125692e-06,
+      "loss": 0.689,
+      "step": 13039
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.929001736398238e-06,
+      "loss": 0.8081,
+      "step": 13040
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.927593776566289e-06,
+      "loss": 0.7466,
+      "step": 13041
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.926185859659006e-06,
+      "loss": 0.6799,
+      "step": 13042
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.924777985705556e-06,
+      "loss": 0.7686,
+      "step": 13043
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.923370154735093e-06,
+      "loss": 0.7227,
+      "step": 13044
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.921962366776785e-06,
+      "loss": 0.7363,
+      "step": 13045
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.920554621859781e-06,
+      "loss": 0.6963,
+      "step": 13046
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91914692001325e-06,
+      "loss": 0.7378,
+      "step": 13047
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.917739261266337e-06,
+      "loss": 0.7004,
+      "step": 13048
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.916331645648208e-06,
+      "loss": 0.769,
+      "step": 13049
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91492407318801e-06,
+      "loss": 0.8174,
+      "step": 13050
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.913516543914906e-06,
+      "loss": 0.8745,
+      "step": 13051
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.912109057858042e-06,
+      "loss": 0.7529,
+      "step": 13052
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.910701615046576e-06,
+      "loss": 0.6621,
+      "step": 13053
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.909294215509653e-06,
+      "loss": 0.7617,
+      "step": 13054
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.907886859276432e-06,
+      "loss": 0.7134,
+      "step": 13055
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.90647954637606e-06,
+      "loss": 0.8276,
+      "step": 13056
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.90507227683768e-06,
+      "loss": 0.8696,
+      "step": 13057
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.903665050690449e-06,
+      "loss": 0.7783,
+      "step": 13058
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.902257867963505e-06,
+      "loss": 0.709,
+      "step": 13059
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.900850728686004e-06,
+      "loss": 0.7334,
+      "step": 13060
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.89944363288708e-06,
+      "loss": 0.7163,
+      "step": 13061
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.898036580595889e-06,
+      "loss": 0.7747,
+      "step": 13062
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.896629571841568e-06,
+      "loss": 0.7598,
+      "step": 13063
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.895222606653261e-06,
+      "loss": 0.759,
+      "step": 13064
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.893815685060109e-06,
+      "loss": 0.6367,
+      "step": 13065
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.892408807091254e-06,
+      "loss": 0.7417,
+      "step": 13066
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.891001972775834e-06,
+      "loss": 0.7964,
+      "step": 13067
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.889595182142995e-06,
+      "loss": 0.7781,
+      "step": 13068
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.888188435221861e-06,
+      "loss": 0.7168,
+      "step": 13069
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.886781732041586e-06,
+      "loss": 0.7988,
+      "step": 13070
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.885375072631297e-06,
+      "loss": 0.6731,
+      "step": 13071
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.883968457020127e-06,
+      "loss": 0.7612,
+      "step": 13072
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.882561885237218e-06,
+      "loss": 0.6758,
+      "step": 13073
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.881155357311694e-06,
+      "loss": 0.7808,
+      "step": 13074
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.879748873272696e-06,
+      "loss": 0.7773,
+      "step": 13075
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.878342433149354e-06,
+      "loss": 0.6125,
+      "step": 13076
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.876936036970799e-06,
+      "loss": 0.7522,
+      "step": 13077
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.875529684766155e-06,
+      "loss": 0.7241,
+      "step": 13078
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.874123376564563e-06,
+      "loss": 0.696,
+      "step": 13079
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.872717112395138e-06,
+      "loss": 0.7354,
+      "step": 13080
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.871310892287017e-06,
+      "loss": 0.7749,
+      "step": 13081
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.86990471626932e-06,
+      "loss": 0.7637,
+      "step": 13082
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.868498584371178e-06,
+      "loss": 0.7114,
+      "step": 13083
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.867092496621713e-06,
+      "loss": 0.7637,
+      "step": 13084
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.865686453050043e-06,
+      "loss": 0.7905,
+      "step": 13085
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.864280453685301e-06,
+      "loss": 0.656,
+      "step": 13086
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.862874498556599e-06,
+      "loss": 0.7324,
+      "step": 13087
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.861468587693065e-06,
+      "loss": 0.7388,
+      "step": 13088
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.860062721123814e-06,
+      "loss": 0.7212,
+      "step": 13089
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.858656898877968e-06,
+      "loss": 0.7158,
+      "step": 13090
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.857251120984644e-06,
+      "loss": 0.7148,
+      "step": 13091
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.855845387472962e-06,
+      "loss": 0.7891,
+      "step": 13092
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.85443969837203e-06,
+      "loss": 0.7422,
+      "step": 13093
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.853034053710975e-06,
+      "loss": 0.8125,
+      "step": 13094
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.8516284535189e-06,
+      "loss": 0.7202,
+      "step": 13095
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.850222897824929e-06,
+      "loss": 0.7827,
+      "step": 13096
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.848817386658164e-06,
+      "loss": 0.6848,
+      "step": 13097
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.847411920047725e-06,
+      "loss": 0.7832,
+      "step": 13098
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.846006498022719e-06,
+      "loss": 0.6768,
+      "step": 13099
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.844601120612256e-06,
+      "loss": 0.7505,
+      "step": 13100
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.843195787845445e-06,
+      "loss": 0.7544,
+      "step": 13101
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.841790499751393e-06,
+      "loss": 0.7229,
+      "step": 13102
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.840385256359212e-06,
+      "loss": 0.7471,
+      "step": 13103
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.838980057697998e-06,
+      "loss": 0.7986,
+      "step": 13104
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.837574903796867e-06,
+      "loss": 0.647,
+      "step": 13105
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.836169794684915e-06,
+      "loss": 0.7026,
+      "step": 13106
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.834764730391254e-06,
+      "loss": 0.7971,
+      "step": 13107
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.833359710944976e-06,
+      "loss": 0.7817,
+      "step": 13108
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.83195473637519e-06,
+      "loss": 0.7104,
+      "step": 13109
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.830549806710994e-06,
+      "loss": 0.6565,
+      "step": 13110
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82914492198149e-06,
+      "loss": 0.7334,
+      "step": 13111
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82774008221577e-06,
+      "loss": 0.8538,
+      "step": 13112
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.826335287442939e-06,
+      "loss": 0.7573,
+      "step": 13113
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82493053769209e-06,
+      "loss": 0.6519,
+      "step": 13114
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.823525832992318e-06,
+      "loss": 0.7017,
+      "step": 13115
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.822121173372723e-06,
+      "loss": 0.8228,
+      "step": 13116
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.820716558862391e-06,
+      "loss": 0.7134,
+      "step": 13117
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.819311989490424e-06,
+      "loss": 0.7451,
+      "step": 13118
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.817907465285907e-06,
+      "loss": 0.7534,
+      "step": 13119
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.816502986277936e-06,
+      "loss": 0.8184,
+      "step": 13120
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.815098552495597e-06,
+      "loss": 0.9785,
+      "step": 13121
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.813694163967984e-06,
+      "loss": 0.7891,
+      "step": 13122
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.81228982072418e-06,
+      "loss": 0.7876,
+      "step": 13123
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.810885522793278e-06,
+      "loss": 0.7729,
+      "step": 13124
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.809481270204357e-06,
+      "loss": 0.7827,
+      "step": 13125
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.808077062986515e-06,
+      "loss": 0.6841,
+      "step": 13126
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.806672901168821e-06,
+      "loss": 0.7583,
+      "step": 13127
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.80526878478037e-06,
+      "loss": 0.8271,
+      "step": 13128
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.803864713850243e-06,
+      "loss": 0.7681,
+      "step": 13129
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.802460688407518e-06,
+      "loss": 0.7061,
+      "step": 13130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.80105670848128e-06,
+      "loss": 0.6758,
+      "step": 13131
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.799652774100603e-06,
+      "loss": 0.7803,
+      "step": 13132
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.798248885294573e-06,
+      "loss": 0.7148,
+      "step": 13133
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.796845042092263e-06,
+      "loss": 0.6702,
+      "step": 13134
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.795441244522754e-06,
+      "loss": 0.7417,
+      "step": 13135
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.794037492615116e-06,
+      "loss": 0.7842,
+      "step": 13136
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.792633786398433e-06,
+      "loss": 0.7651,
+      "step": 13137
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.79123012590177e-06,
+      "loss": 0.7727,
+      "step": 13138
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.789826511154208e-06,
+      "loss": 0.7712,
+      "step": 13139
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.788422942184815e-06,
+      "loss": 0.8057,
+      "step": 13140
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.787019419022663e-06,
+      "loss": 0.7405,
+      "step": 13141
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.785615941696824e-06,
+      "loss": 0.8809,
+      "step": 13142
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.784212510236366e-06,
+      "loss": 0.6233,
+      "step": 13143
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.78280912467036e-06,
+      "loss": 0.7456,
+      "step": 13144
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.781405785027867e-06,
+      "loss": 0.6641,
+      "step": 13145
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.780002491337963e-06,
+      "loss": 0.8872,
+      "step": 13146
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.778599243629704e-06,
+      "loss": 0.7744,
+      "step": 13147
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.777196041932164e-06,
+      "loss": 0.8037,
+      "step": 13148
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.775792886274399e-06,
+      "loss": 0.7305,
+      "step": 13149
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.774389776685478e-06,
+      "loss": 0.7993,
+      "step": 13150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.772986713194457e-06,
+      "loss": 0.7998,
+      "step": 13151
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.7715836958304e-06,
+      "loss": 0.7803,
+      "step": 13152
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.770180724622369e-06,
+      "loss": 0.7373,
+      "step": 13153
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.76877779959942e-06,
+      "loss": 0.7219,
+      "step": 13154
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.767374920790611e-06,
+      "loss": 0.7856,
+      "step": 13155
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.765972088225005e-06,
+      "loss": 0.8003,
+      "step": 13156
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.764569301931647e-06,
+      "loss": 0.7886,
+      "step": 13157
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.763166561939603e-06,
+      "loss": 0.7207,
+      "step": 13158
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.761763868277924e-06,
+      "loss": 0.7007,
+      "step": 13159
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.760361220975656e-06,
+      "loss": 0.7329,
+      "step": 13160
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.758958620061863e-06,
+      "loss": 0.7095,
+      "step": 13161
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.757556065565586e-06,
+      "loss": 0.77,
+      "step": 13162
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.756153557515885e-06,
+      "loss": 0.751,
+      "step": 13163
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.754751095941797e-06,
+      "loss": 0.7593,
+      "step": 13164
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.753348680872382e-06,
+      "loss": 0.7827,
+      "step": 13165
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.751946312336684e-06,
+      "loss": 0.7808,
+      "step": 13166
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.750543990363748e-06,
+      "loss": 0.7363,
+      "step": 13167
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.749141714982619e-06,
+      "loss": 0.7886,
+      "step": 13168
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.747739486222347e-06,
+      "loss": 0.7183,
+      "step": 13169
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.746337304111966e-06,
+      "loss": 0.8101,
+      "step": 13170
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.74493516868053e-06,
+      "loss": 0.7861,
+      "step": 13171
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.743533079957071e-06,
+      "loss": 0.6685,
+      "step": 13172
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.742131037970637e-06,
+      "loss": 0.7388,
+      "step": 13173
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.740729042750264e-06,
+      "loss": 0.5986,
+      "step": 13174
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.739327094324989e-06,
+      "loss": 0.7773,
+      "step": 13175
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.737925192723852e-06,
+      "loss": 0.6853,
+      "step": 13176
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.736523337975893e-06,
+      "loss": 0.8555,
+      "step": 13177
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.735121530110141e-06,
+      "loss": 0.8276,
+      "step": 13178
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.733719769155637e-06,
+      "loss": 0.7305,
+      "step": 13179
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.732318055141413e-06,
+      "loss": 0.8369,
+      "step": 13180
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.7309163880965e-06,
+      "loss": 0.8164,
+      "step": 13181
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.729514768049933e-06,
+      "loss": 0.6653,
+      "step": 13182
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.728113195030739e-06,
+      "loss": 0.8706,
+      "step": 13183
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.726711669067957e-06,
+      "loss": 0.8032,
+      "step": 13184
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.725310190190604e-06,
+      "loss": 0.74,
+      "step": 13185
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.723908758427718e-06,
+      "loss": 0.6812,
+      "step": 13186
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.722507373808318e-06,
+      "loss": 0.7041,
+      "step": 13187
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.721106036361438e-06,
+      "loss": 0.8232,
+      "step": 13188
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.7197047461161e-06,
+      "loss": 0.7966,
+      "step": 13189
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.718303503101325e-06,
+      "loss": 0.7715,
+      "step": 13190
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.71690230734614e-06,
+      "loss": 0.7571,
+      "step": 13191
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.715501158879561e-06,
+      "loss": 0.8076,
+      "step": 13192
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.714100057730621e-06,
+      "loss": 0.7871,
+      "step": 13193
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.71269900392833e-06,
+      "loss": 0.7324,
+      "step": 13194
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.711297997501713e-06,
+      "loss": 0.6274,
+      "step": 13195
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.709897038479784e-06,
+      "loss": 0.8032,
+      "step": 13196
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.708496126891565e-06,
+      "loss": 0.7217,
+      "step": 13197
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.707095262766067e-06,
+      "loss": 0.8862,
+      "step": 13198
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.705694446132311e-06,
+      "loss": 0.7388,
+      "step": 13199
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.704293677019305e-06,
+      "loss": 0.9014,
+      "step": 13200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.702892955456072e-06,
+      "loss": 0.6943,
+      "step": 13201
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.701492281471612e-06,
+      "loss": 0.8218,
+      "step": 13202
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.700091655094947e-06,
+      "loss": 0.7349,
+      "step": 13203
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.698691076355082e-06,
+      "loss": 0.7476,
+      "step": 13204
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.697290545281027e-06,
+      "loss": 0.7373,
+      "step": 13205
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.695890061901792e-06,
+      "loss": 0.6797,
+      "step": 13206
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.694489626246382e-06,
+      "loss": 0.7727,
+      "step": 13207
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.69308923834381e-06,
+      "loss": 0.7021,
+      "step": 13208
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.69168889822307e-06,
+      "loss": 0.7056,
+      "step": 13209
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.69028860591318e-06,
+      "loss": 0.748,
+      "step": 13210
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.68888836144313e-06,
+      "loss": 0.8291,
+      "step": 13211
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.687488164841937e-06,
+      "loss": 0.7351,
+      "step": 13212
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.686088016138588e-06,
+      "loss": 0.7378,
+      "step": 13213
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.684687915362094e-06,
+      "loss": 0.7673,
+      "step": 13214
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.683287862541449e-06,
+      "loss": 0.7124,
+      "step": 13215
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.681887857705653e-06,
+      "loss": 0.8237,
+      "step": 13216
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.680487900883707e-06,
+      "loss": 0.6709,
+      "step": 13217
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.679087992104602e-06,
+      "loss": 0.7075,
+      "step": 13218
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67768813139734e-06,
+      "loss": 0.6548,
+      "step": 13219
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.676288318790906e-06,
+      "loss": 0.7554,
+      "step": 13220
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.674888554314304e-06,
+      "loss": 0.7729,
+      "step": 13221
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67348883799652e-06,
+      "loss": 0.7017,
+      "step": 13222
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67208916986655e-06,
+      "loss": 0.7681,
+      "step": 13223
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.670689549953377e-06,
+      "loss": 0.7871,
+      "step": 13224
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.669289978286001e-06,
+      "loss": 0.6855,
+      "step": 13225
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.6678904548934e-06,
+      "loss": 0.8037,
+      "step": 13226
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.666490979804574e-06,
+      "loss": 0.7603,
+      "step": 13227
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.665091553048497e-06,
+      "loss": 0.7161,
+      "step": 13228
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.66369217465416e-06,
+      "loss": 0.7378,
+      "step": 13229
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.662292844650551e-06,
+      "loss": 0.7695,
+      "step": 13230
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.66089356306665e-06,
+      "loss": 0.6899,
+      "step": 13231
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.659494329931436e-06,
+      "loss": 0.6675,
+      "step": 13232
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.6580951452739e-06,
+      "loss": 0.7778,
+      "step": 13233
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.656696009123016e-06,
+      "loss": 0.7544,
+      "step": 13234
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.655296921507762e-06,
+      "loss": 0.728,
+      "step": 13235
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.653897882457122e-06,
+      "loss": 0.73,
+      "step": 13236
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.652498892000066e-06,
+      "loss": 0.8047,
+      "step": 13237
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.651099950165581e-06,
+      "loss": 0.7432,
+      "step": 13238
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.649701056982632e-06,
+      "loss": 0.7036,
+      "step": 13239
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.648302212480203e-06,
+      "loss": 0.7983,
+      "step": 13240
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64690341668726e-06,
+      "loss": 0.709,
+      "step": 13241
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64550466963278e-06,
+      "loss": 0.7568,
+      "step": 13242
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.644105971345731e-06,
+      "loss": 0.7744,
+      "step": 13243
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.642707321855088e-06,
+      "loss": 0.7234,
+      "step": 13244
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.641308721189816e-06,
+      "loss": 0.8042,
+      "step": 13245
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.639910169378888e-06,
+      "loss": 0.792,
+      "step": 13246
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.638511666451266e-06,
+      "loss": 0.7515,
+      "step": 13247
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.637113212435923e-06,
+      "loss": 0.719,
+      "step": 13248
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.63571480736182e-06,
+      "loss": 0.7551,
+      "step": 13249
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.634316451257919e-06,
+      "loss": 0.8252,
+      "step": 13250
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.632918144153192e-06,
+      "loss": 0.771,
+      "step": 13251
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.63151988607659e-06,
+      "loss": 0.8291,
+      "step": 13252
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.630121677057085e-06,
+      "loss": 0.6978,
+      "step": 13253
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.628723517123632e-06,
+      "loss": 0.7698,
+      "step": 13254
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.62732540630519e-06,
+      "loss": 0.8193,
+      "step": 13255
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.625927344630721e-06,
+      "loss": 0.771,
+      "step": 13256
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.624529332129178e-06,
+      "loss": 0.7852,
+      "step": 13257
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.623131368829518e-06,
+      "loss": 0.6904,
+      "step": 13258
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.621733454760702e-06,
+      "loss": 0.7578,
+      "step": 13259
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.620335589951675e-06,
+      "loss": 0.7207,
+      "step": 13260
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6189377744314e-06,
+      "loss": 0.7183,
+      "step": 13261
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.617540008228819e-06,
+      "loss": 0.7515,
+      "step": 13262
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6161422913728936e-06,
+      "loss": 0.7983,
+      "step": 13263
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.614744623892569e-06,
+      "loss": 0.7793,
+      "step": 13264
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.613347005816789e-06,
+      "loss": 0.7766,
+      "step": 13265
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6119494371745095e-06,
+      "loss": 0.6733,
+      "step": 13266
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6105519179946754e-06,
+      "loss": 0.803,
+      "step": 13267
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.609154448306232e-06,
+      "loss": 0.7998,
+      "step": 13268
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.607757028138123e-06,
+      "loss": 0.7463,
+      "step": 13269
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.606359657519299e-06,
+      "loss": 0.8145,
+      "step": 13270
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.604962336478693e-06,
+      "loss": 0.8057,
+      "step": 13271
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.603565065045256e-06,
+      "loss": 0.793,
+      "step": 13272
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6021678432479225e-06,
+      "loss": 0.7861,
+      "step": 13273
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.600770671115638e-06,
+      "loss": 0.7944,
+      "step": 13274
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.599373548677336e-06,
+      "loss": 0.6445,
+      "step": 13275
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.597976475961959e-06,
+      "loss": 0.689,
+      "step": 13276
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.596579452998439e-06,
+      "loss": 0.8105,
+      "step": 13277
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.595182479815718e-06,
+      "loss": 0.6355,
+      "step": 13278
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.593785556442725e-06,
+      "loss": 0.8315,
+      "step": 13279
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.592388682908397e-06,
+      "loss": 0.8037,
+      "step": 13280
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.590991859241665e-06,
+      "loss": 0.8062,
+      "step": 13281
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.589595085471459e-06,
+      "loss": 0.7998,
+      "step": 13282
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.588198361626717e-06,
+      "loss": 0.6985,
+      "step": 13283
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.58680168773636e-06,
+      "loss": 0.7871,
+      "step": 13284
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5854050638293245e-06,
+      "loss": 0.7681,
+      "step": 13285
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.584008489934529e-06,
+      "loss": 0.7197,
+      "step": 13286
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.582611966080908e-06,
+      "loss": 0.7129,
+      "step": 13287
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.581215492297382e-06,
+      "loss": 0.6968,
+      "step": 13288
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.579819068612879e-06,
+      "loss": 0.8149,
+      "step": 13289
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5784226950563175e-06,
+      "loss": 0.7612,
+      "step": 13290
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.577026371656627e-06,
+      "loss": 0.7119,
+      "step": 13291
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5756300984427234e-06,
+      "loss": 0.8564,
+      "step": 13292
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5742338754435255e-06,
+      "loss": 0.6729,
+      "step": 13293
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.57283770268796e-06,
+      "loss": 0.7405,
+      "step": 13294
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.571441580204936e-06,
+      "loss": 0.8032,
+      "step": 13295
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.570045508023379e-06,
+      "loss": 0.7229,
+      "step": 13296
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.568649486172197e-06,
+      "loss": 0.7439,
+      "step": 13297
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.567253514680315e-06,
+      "loss": 0.8232,
+      "step": 13298
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.565857593576635e-06,
+      "loss": 0.7705,
+      "step": 13299
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.564461722890082e-06,
+      "loss": 0.7822,
+      "step": 13300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.563065902649558e-06,
+      "loss": 0.7891,
+      "step": 13301
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.561670132883983e-06,
+      "loss": 0.8535,
+      "step": 13302
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.560274413622257e-06,
+      "loss": 0.8062,
+      "step": 13303
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.558878744893298e-06,
+      "loss": 0.7537,
+      "step": 13304
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5574831267260055e-06,
+      "loss": 0.7603,
+      "step": 13305
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.556087559149292e-06,
+      "loss": 0.6948,
+      "step": 13306
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.554692042192064e-06,
+      "loss": 0.749,
+      "step": 13307
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.55329657588322e-06,
+      "loss": 0.7681,
+      "step": 13308
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.551901160251671e-06,
+      "loss": 0.7358,
+      "step": 13309
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.550505795326312e-06,
+      "loss": 0.7573,
+      "step": 13310
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.54911048113605e-06,
+      "loss": 0.7461,
+      "step": 13311
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5477152177097825e-06,
+      "loss": 0.7583,
+      "step": 13312
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.546320005076413e-06,
+      "loss": 0.7612,
+      "step": 13313
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.544924843264832e-06,
+      "loss": 0.8135,
+      "step": 13314
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.543529732303946e-06,
+      "loss": 0.8418,
+      "step": 13315
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5421346722226415e-06,
+      "loss": 0.772,
+      "step": 13316
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.540739663049824e-06,
+      "loss": 0.792,
+      "step": 13317
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5393447048143785e-06,
+      "loss": 0.7383,
+      "step": 13318
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.537949797545204e-06,
+      "loss": 0.7219,
+      "step": 13319
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.53655494127119e-06,
+      "loss": 0.6201,
+      "step": 13320
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.535160136021229e-06,
+      "loss": 0.8081,
+      "step": 13321
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.533765381824211e-06,
+      "loss": 0.7612,
+      "step": 13322
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.532370678709019e-06,
+      "loss": 0.7004,
+      "step": 13323
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.530976026704551e-06,
+      "loss": 0.7661,
+      "step": 13324
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.529581425839683e-06,
+      "loss": 0.7485,
+      "step": 13325
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.528186876143308e-06,
+      "loss": 0.6836,
+      "step": 13326
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.526792377644308e-06,
+      "loss": 0.7158,
+      "step": 13327
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.525397930371568e-06,
+      "loss": 0.7007,
+      "step": 13328
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5240035343539655e-06,
+      "loss": 0.6704,
+      "step": 13329
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.522609189620389e-06,
+      "loss": 0.8149,
+      "step": 13330
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.521214896199714e-06,
+      "loss": 0.7822,
+      "step": 13331
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.519820654120822e-06,
+      "loss": 0.7808,
+      "step": 13332
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5184264634125904e-06,
+      "loss": 0.7402,
+      "step": 13333
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.517032324103896e-06,
+      "loss": 0.7339,
+      "step": 13334
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.515638236223615e-06,
+      "loss": 0.8516,
+      "step": 13335
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.514244199800625e-06,
+      "loss": 0.7856,
+      "step": 13336
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.512850214863798e-06,
+      "loss": 0.8096,
+      "step": 13337
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.511456281442002e-06,
+      "loss": 0.7261,
+      "step": 13338
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.510062399564119e-06,
+      "loss": 0.7266,
+      "step": 13339
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.508668569259008e-06,
+      "loss": 0.688,
+      "step": 13340
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.50727479055555e-06,
+      "loss": 0.8398,
+      "step": 13341
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.505881063482604e-06,
+      "loss": 0.8042,
+      "step": 13342
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.504487388069044e-06,
+      "loss": 0.7593,
+      "step": 13343
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.503093764343735e-06,
+      "loss": 0.8247,
+      "step": 13344
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5017001923355415e-06,
+      "loss": 0.7661,
+      "step": 13345
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.500306672073326e-06,
+      "loss": 0.7551,
+      "step": 13346
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.498913203585958e-06,
+      "loss": 0.7578,
+      "step": 13347
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.497519786902291e-06,
+      "loss": 0.6633,
+      "step": 13348
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.496126422051195e-06,
+      "loss": 0.8477,
+      "step": 13349
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.494733109061522e-06,
+      "loss": 0.6216,
+      "step": 13350
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.493339847962138e-06,
+      "loss": 0.6738,
+      "step": 13351
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4919466387819e-06,
+      "loss": 0.7905,
+      "step": 13352
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.490553481549655e-06,
+      "loss": 0.7397,
+      "step": 13353
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4891603762942724e-06,
+      "loss": 0.7617,
+      "step": 13354
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4877673230445944e-06,
+      "loss": 0.7881,
+      "step": 13355
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.486374321829486e-06,
+      "loss": 0.73,
+      "step": 13356
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.484981372677792e-06,
+      "loss": 0.7964,
+      "step": 13357
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.483588475618367e-06,
+      "loss": 0.679,
+      "step": 13358
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.482195630680058e-06,
+      "loss": 0.8081,
+      "step": 13359
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.480802837891723e-06,
+      "loss": 0.8364,
+      "step": 13360
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.479410097282198e-06,
+      "loss": 0.6919,
+      "step": 13361
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4780174088803415e-06,
+      "loss": 0.812,
+      "step": 13362
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.476624772714989e-06,
+      "loss": 0.7349,
+      "step": 13363
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.475232188814996e-06,
+      "loss": 0.7666,
+      "step": 13364
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4738396572091965e-06,
+      "loss": 0.7295,
+      "step": 13365
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.472447177926443e-06,
+      "loss": 0.7944,
+      "step": 13366
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.471054750995568e-06,
+      "loss": 0.7373,
+      "step": 13367
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.469662376445417e-06,
+      "loss": 0.7764,
+      "step": 13368
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4682700543048295e-06,
+      "loss": 0.7676,
+      "step": 13369
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.46687778460264e-06,
+      "loss": 0.7861,
+      "step": 13370
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.465485567367694e-06,
+      "loss": 0.7073,
+      "step": 13371
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4640934026288184e-06,
+      "loss": 0.7739,
+      "step": 13372
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.462701290414856e-06,
+      "loss": 0.7368,
+      "step": 13373
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.461309230754634e-06,
+      "loss": 0.7202,
+      "step": 13374
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4599172236769935e-06,
+      "loss": 0.6475,
+      "step": 13375
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.458525269210758e-06,
+      "loss": 0.709,
+      "step": 13376
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.457133367384765e-06,
+      "loss": 0.6826,
+      "step": 13377
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.455741518227839e-06,
+      "loss": 0.7949,
+      "step": 13378
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.454349721768816e-06,
+      "loss": 0.8423,
+      "step": 13379
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4529579780365125e-06,
+      "loss": 0.6775,
+      "step": 13380
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.451566287059766e-06,
+      "loss": 0.728,
+      "step": 13381
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.450174648867396e-06,
+      "loss": 0.7188,
+      "step": 13382
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.448783063488225e-06,
+      "loss": 0.7061,
+      "step": 13383
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.447391530951082e-06,
+      "loss": 0.7256,
+      "step": 13384
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.446000051284784e-06,
+      "loss": 0.7803,
+      "step": 13385
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.444608624518157e-06,
+      "loss": 0.7876,
+      "step": 13386
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.443217250680015e-06,
+      "loss": 0.7441,
+      "step": 13387
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.441825929799182e-06,
+      "loss": 0.7515,
+      "step": 13388
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.440434661904471e-06,
+      "loss": 0.7319,
+      "step": 13389
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.439043447024705e-06,
+      "loss": 0.7056,
+      "step": 13390
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.43765228518869e-06,
+      "loss": 0.7788,
+      "step": 13391
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.436261176425252e-06,
+      "loss": 0.7598,
+      "step": 13392
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.434870120763193e-06,
+      "loss": 0.6592,
+      "step": 13393
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.433479118231335e-06,
+      "loss": 0.6248,
+      "step": 13394
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.43208816885848e-06,
+      "loss": 0.6604,
+      "step": 13395
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.430697272673446e-06,
+      "loss": 0.8467,
+      "step": 13396
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.42930642970504e-06,
+      "loss": 0.6792,
+      "step": 13397
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.427915639982064e-06,
+      "loss": 0.7896,
+      "step": 13398
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4265249035333345e-06,
+      "loss": 0.833,
+      "step": 13399
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.425134220387645e-06,
+      "loss": 0.7432,
+      "step": 13400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.423743590573813e-06,
+      "loss": 0.8628,
+      "step": 13401
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4223530141206305e-06,
+      "loss": 0.7637,
+      "step": 13402
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.420962491056909e-06,
+      "loss": 0.7729,
+      "step": 13403
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4195720214114404e-06,
+      "loss": 0.752,
+      "step": 13404
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.418181605213035e-06,
+      "loss": 0.751,
+      "step": 13405
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.416791242490483e-06,
+      "loss": 0.7837,
+      "step": 13406
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.415400933272587e-06,
+      "loss": 0.8374,
+      "step": 13407
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.414010677588143e-06,
+      "loss": 0.7891,
+      "step": 13408
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.412620475465947e-06,
+      "loss": 0.7959,
+      "step": 13409
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.411230326934793e-06,
+      "loss": 0.8018,
+      "step": 13410
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.409840232023473e-06,
+      "loss": 0.7578,
+      "step": 13411
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.408450190760783e-06,
+      "loss": 0.7476,
+      "step": 13412
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.407060203175508e-06,
+      "loss": 0.752,
+      "step": 13413
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.405670269296445e-06,
+      "loss": 0.7544,
+      "step": 13414
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4042803891523765e-06,
+      "loss": 0.8159,
+      "step": 13415
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.402890562772098e-06,
+      "loss": 0.8154,
+      "step": 13416
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.401500790184387e-06,
+      "loss": 0.7188,
+      "step": 13417
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.400111071418039e-06,
+      "loss": 0.689,
+      "step": 13418
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.39872140650183e-06,
+      "loss": 0.7466,
+      "step": 13419
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.397331795464547e-06,
+      "loss": 0.8052,
+      "step": 13420
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.395942238334974e-06,
+      "loss": 0.8394,
+      "step": 13421
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.39455273514189e-06,
+      "loss": 0.6719,
+      "step": 13422
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.393163285914073e-06,
+      "loss": 0.7949,
+      "step": 13423
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.391773890680309e-06,
+      "loss": 0.7256,
+      "step": 13424
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.390384549469367e-06,
+      "loss": 0.7666,
+      "step": 13425
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3889952623100326e-06,
+      "loss": 0.7061,
+      "step": 13426
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.387606029231076e-06,
+      "loss": 0.7661,
+      "step": 13427
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.386216850261268e-06,
+      "loss": 0.8027,
+      "step": 13428
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.384827725429391e-06,
+      "loss": 0.8179,
+      "step": 13429
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.383438654764208e-06,
+      "loss": 0.7207,
+      "step": 13430
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3820496382944995e-06,
+      "loss": 0.7456,
+      "step": 13431
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.380660676049026e-06,
+      "loss": 0.7156,
+      "step": 13432
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.379271768056564e-06,
+      "loss": 0.7012,
+      "step": 13433
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.377882914345877e-06,
+      "loss": 0.8193,
+      "step": 13434
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3764941149457335e-06,
+      "loss": 0.8179,
+      "step": 13435
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.375105369884895e-06,
+      "loss": 0.717,
+      "step": 13436
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3737166791921345e-06,
+      "loss": 0.7949,
+      "step": 13437
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.372328042896205e-06,
+      "loss": 0.6887,
+      "step": 13438
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3709394610258785e-06,
+      "loss": 0.7539,
+      "step": 13439
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3695509336099056e-06,
+      "loss": 0.7695,
+      "step": 13440
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.368162460677056e-06,
+      "loss": 0.729,
+      "step": 13441
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.366774042256084e-06,
+      "loss": 0.8857,
+      "step": 13442
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.365385678375743e-06,
+      "loss": 0.7441,
+      "step": 13443
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.363997369064796e-06,
+      "loss": 0.7671,
+      "step": 13444
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.362609114351995e-06,
+      "loss": 0.7183,
+      "step": 13445
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3612209142660964e-06,
+      "loss": 0.7476,
+      "step": 13446
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.359832768835848e-06,
+      "loss": 0.7031,
+      "step": 13447
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.358444678090011e-06,
+      "loss": 0.8418,
+      "step": 13448
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.357056642057327e-06,
+      "loss": 0.7012,
+      "step": 13449
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.355668660766553e-06,
+      "loss": 0.717,
+      "step": 13450
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.35428073424643e-06,
+      "loss": 0.7793,
+      "step": 13451
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.352892862525714e-06,
+      "loss": 0.8384,
+      "step": 13452
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3515050456331404e-06,
+      "loss": 0.8057,
+      "step": 13453
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.350117283597468e-06,
+      "loss": 0.7148,
+      "step": 13454
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.348729576447427e-06,
+      "loss": 0.7024,
+      "step": 13455
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.347341924211772e-06,
+      "loss": 0.7271,
+      "step": 13456
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.345954326919237e-06,
+      "loss": 0.792,
+      "step": 13457
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.344566784598565e-06,
+      "loss": 0.6648,
+      "step": 13458
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.343179297278495e-06,
+      "loss": 0.7139,
+      "step": 13459
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.341791864987764e-06,
+      "loss": 0.7014,
+      "step": 13460
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.340404487755115e-06,
+      "loss": 0.7793,
+      "step": 13461
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.339017165609275e-06,
+      "loss": 0.6562,
+      "step": 13462
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.337629898578988e-06,
+      "loss": 0.6714,
+      "step": 13463
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.336242686692981e-06,
+      "loss": 0.7109,
+      "step": 13464
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3348555299799915e-06,
+      "loss": 0.7598,
+      "step": 13465
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.333468428468745e-06,
+      "loss": 0.708,
+      "step": 13466
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3320813821879796e-06,
+      "loss": 0.7039,
+      "step": 13467
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.330694391166416e-06,
+      "loss": 0.7095,
+      "step": 13468
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3293074554327905e-06,
+      "loss": 0.7466,
+      "step": 13469
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.327920575015822e-06,
+      "loss": 0.7856,
+      "step": 13470
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.326533749944245e-06,
+      "loss": 0.7397,
+      "step": 13471
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3251469802467756e-06,
+      "loss": 0.7671,
+      "step": 13472
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.32376026595214e-06,
+      "loss": 0.834,
+      "step": 13473
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.322373607089065e-06,
+      "loss": 0.699,
+      "step": 13474
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.320987003686264e-06,
+      "loss": 0.728,
+      "step": 13475
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.319600455772465e-06,
+      "loss": 0.752,
+      "step": 13476
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3182139633763795e-06,
+      "loss": 0.7109,
+      "step": 13477
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.316827526526733e-06,
+      "loss": 0.7646,
+      "step": 13478
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.315441145252232e-06,
+      "loss": 0.7817,
+      "step": 13479
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.314054819581603e-06,
+      "loss": 0.7444,
+      "step": 13480
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.31266854954355e-06,
+      "loss": 0.7793,
+      "step": 13481
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.311282335166796e-06,
+      "loss": 0.6953,
+      "step": 13482
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.309896176480042e-06,
+      "loss": 0.7559,
+      "step": 13483
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.3085100735120085e-06,
+      "loss": 0.7065,
+      "step": 13484
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.307124026291401e-06,
+      "loss": 0.7593,
+      "step": 13485
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.305738034846928e-06,
+      "loss": 0.718,
+      "step": 13486
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.304352099207298e-06,
+      "loss": 0.8091,
+      "step": 13487
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.302966219401213e-06,
+      "loss": 0.7029,
+      "step": 13488
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.301580395457385e-06,
+      "loss": 0.7144,
+      "step": 13489
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.300194627404511e-06,
+      "loss": 0.8276,
+      "step": 13490
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.298808915271299e-06,
+      "loss": 0.7656,
+      "step": 13491
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.297423259086444e-06,
+      "loss": 0.6943,
+      "step": 13492
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.296037658878657e-06,
+      "loss": 0.7769,
+      "step": 13493
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2946521146766255e-06,
+      "loss": 0.6829,
+      "step": 13494
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.293266626509058e-06,
+      "loss": 0.7373,
+      "step": 13495
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.291881194404642e-06,
+      "loss": 0.77,
+      "step": 13496
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.290495818392079e-06,
+      "loss": 0.7935,
+      "step": 13497
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.28911049850006e-06,
+      "loss": 0.6943,
+      "step": 13498
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.287725234757283e-06,
+      "loss": 0.7339,
+      "step": 13499
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2863400271924404e-06,
+      "loss": 0.7297,
+      "step": 13500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.284954875834216e-06,
+      "loss": 0.7202,
+      "step": 13501
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.283569780711308e-06,
+      "loss": 0.7471,
+      "step": 13502
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.282184741852396e-06,
+      "loss": 0.7197,
+      "step": 13503
+    },
+    {
+      "epoch": 3.0,
+      "eval_cIoU": 0.7983784675598145,
+      "eval_gIoU": 0.7953092455863953,
+      "eval_runtime": 120.3032,
+      "eval_samples_per_second": 31.678,
+      "eval_steps_per_second": 3.965,
+      "step": 13503
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.280799759286179e-06,
+      "loss": 0.5361,
+      "step": 13504
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2794148330413336e-06,
+      "loss": 0.6558,
+      "step": 13505
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.278029963146554e-06,
+      "loss": 0.5664,
+      "step": 13506
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.276645149630512e-06,
+      "loss": 0.5181,
+      "step": 13507
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2752603925219035e-06,
+      "loss": 0.5569,
+      "step": 13508
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.273875691849401e-06,
+      "loss": 0.6106,
+      "step": 13509
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.27249104764169e-06,
+      "loss": 0.5493,
+      "step": 13510
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.271106459927448e-06,
+      "loss": 0.5046,
+      "step": 13511
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.269721928735356e-06,
+      "loss": 0.5396,
+      "step": 13512
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.268337454094085e-06,
+      "loss": 0.6528,
+      "step": 13513
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.266953036032318e-06,
+      "loss": 0.6035,
+      "step": 13514
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.265568674578729e-06,
+      "loss": 0.6116,
+      "step": 13515
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2641843697619826e-06,
+      "loss": 0.554,
+      "step": 13516
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.262800121610763e-06,
+      "loss": 0.5405,
+      "step": 13517
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.261415930153733e-06,
+      "loss": 0.5217,
+      "step": 13518
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.260031795419566e-06,
+      "loss": 0.55,
+      "step": 13519
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.25864771743693e-06,
+      "loss": 0.5767,
+      "step": 13520
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.257263696234493e-06,
+      "loss": 0.6421,
+      "step": 13521
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.255879731840924e-06,
+      "loss": 0.5742,
+      "step": 13522
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.254495824284884e-06,
+      "loss": 0.5669,
+      "step": 13523
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.253111973595038e-06,
+      "loss": 0.5923,
+      "step": 13524
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.251728179800055e-06,
+      "loss": 0.5662,
+      "step": 13525
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.250344442928588e-06,
+      "loss": 0.5664,
+      "step": 13526
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.248960763009303e-06,
+      "loss": 0.6494,
+      "step": 13527
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.247577140070856e-06,
+      "loss": 0.5808,
+      "step": 13528
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.246193574141911e-06,
+      "loss": 0.5042,
+      "step": 13529
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2448100652511225e-06,
+      "loss": 0.5474,
+      "step": 13530
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.24342661342714e-06,
+      "loss": 0.5122,
+      "step": 13531
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.242043218698627e-06,
+      "loss": 0.5488,
+      "step": 13532
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.240659881094231e-06,
+      "loss": 0.5791,
+      "step": 13533
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2392766006426084e-06,
+      "loss": 0.4924,
+      "step": 13534
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.237893377372408e-06,
+      "loss": 0.5989,
+      "step": 13535
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.236510211312281e-06,
+      "loss": 0.5806,
+      "step": 13536
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.235127102490873e-06,
+      "loss": 0.5354,
+      "step": 13537
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.233744050936841e-06,
+      "loss": 0.6079,
+      "step": 13538
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.232361056678818e-06,
+      "loss": 0.5527,
+      "step": 13539
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.230978119745462e-06,
+      "loss": 0.6489,
+      "step": 13540
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.229595240165406e-06,
+      "loss": 0.5692,
+      "step": 13541
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.228212417967303e-06,
+      "loss": 0.5837,
+      "step": 13542
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.226829653179785e-06,
+      "loss": 0.6025,
+      "step": 13543
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.225446945831502e-06,
+      "loss": 0.5647,
+      "step": 13544
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.224064295951087e-06,
+      "loss": 0.5388,
+      "step": 13545
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.222681703567177e-06,
+      "loss": 0.5764,
+      "step": 13546
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.221299168708414e-06,
+      "loss": 0.6304,
+      "step": 13547
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.219916691403428e-06,
+      "loss": 0.5696,
+      "step": 13548
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.218534271680859e-06,
+      "loss": 0.5754,
+      "step": 13549
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.217151909569337e-06,
+      "loss": 0.5154,
+      "step": 13550
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.215769605097498e-06,
+      "loss": 0.5688,
+      "step": 13551
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.214387358293966e-06,
+      "loss": 0.5627,
+      "step": 13552
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.21300516918738e-06,
+      "loss": 0.5613,
+      "step": 13553
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.211623037806357e-06,
+      "loss": 0.6062,
+      "step": 13554
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.210240964179538e-06,
+      "loss": 0.5977,
+      "step": 13555
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2088589483355355e-06,
+      "loss": 0.5886,
+      "step": 13556
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.207476990302986e-06,
+      "loss": 0.5552,
+      "step": 13557
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.206095090110504e-06,
+      "loss": 0.5615,
+      "step": 13558
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.204713247786721e-06,
+      "loss": 0.5554,
+      "step": 13559
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.203331463360251e-06,
+      "loss": 0.5681,
+      "step": 13560
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.201949736859716e-06,
+      "loss": 0.5669,
+      "step": 13561
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.200568068313739e-06,
+      "loss": 0.5767,
+      "step": 13562
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.199186457750931e-06,
+      "loss": 0.5396,
+      "step": 13563
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.197804905199917e-06,
+      "loss": 0.6667,
+      "step": 13564
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.196423410689301e-06,
+      "loss": 0.4976,
+      "step": 13565
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.1950419742477094e-06,
+      "loss": 0.543,
+      "step": 13566
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.193660595903747e-06,
+      "loss": 0.6648,
+      "step": 13567
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.192279275686031e-06,
+      "loss": 0.4871,
+      "step": 13568
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.190898013623164e-06,
+      "loss": 0.5583,
+      "step": 13569
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.189516809743766e-06,
+      "loss": 0.6516,
+      "step": 13570
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.188135664076434e-06,
+      "loss": 0.5486,
+      "step": 13571
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.186754576649786e-06,
+      "loss": 0.5928,
+      "step": 13572
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.185373547492418e-06,
+      "loss": 0.592,
+      "step": 13573
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.183992576632941e-06,
+      "loss": 0.585,
+      "step": 13574
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.182611664099958e-06,
+      "loss": 0.6074,
+      "step": 13575
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.181230809922065e-06,
+      "loss": 0.5337,
+      "step": 13576
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.179850014127871e-06,
+      "loss": 0.6089,
+      "step": 13577
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.17846927674597e-06,
+      "loss": 0.6418,
+      "step": 13578
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1770885978049645e-06,
+      "loss": 0.5742,
+      "step": 13579
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.175707977333445e-06,
+      "loss": 0.6172,
+      "step": 13580
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.174327415360017e-06,
+      "loss": 0.5903,
+      "step": 13581
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1729469119132675e-06,
+      "loss": 0.4456,
+      "step": 13582
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1715664670217955e-06,
+      "loss": 0.5486,
+      "step": 13583
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.170186080714188e-06,
+      "loss": 0.5376,
+      "step": 13584
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1688057530190446e-06,
+      "loss": 0.5471,
+      "step": 13585
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.167425483964945e-06,
+      "loss": 0.5364,
+      "step": 13586
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.166045273580485e-06,
+      "loss": 0.4854,
+      "step": 13587
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.164665121894251e-06,
+      "loss": 0.4514,
+      "step": 13588
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.163285028934828e-06,
+      "loss": 0.6516,
+      "step": 13589
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.161904994730804e-06,
+      "loss": 0.5623,
+      "step": 13590
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1605250193107574e-06,
+      "loss": 0.6851,
+      "step": 13591
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.159145102703278e-06,
+      "loss": 0.5828,
+      "step": 13592
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.157765244936939e-06,
+      "loss": 0.552,
+      "step": 13593
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1563854460403295e-06,
+      "loss": 0.5898,
+      "step": 13594
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.15500570604202e-06,
+      "loss": 0.5962,
+      "step": 13595
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.153626024970597e-06,
+      "loss": 0.5083,
+      "step": 13596
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.152246402854628e-06,
+      "loss": 0.5701,
+      "step": 13597
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.150866839722696e-06,
+      "loss": 0.6311,
+      "step": 13598
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1494873356033714e-06,
+      "loss": 0.6152,
+      "step": 13599
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.148107890525229e-06,
+      "loss": 0.5579,
+      "step": 13600
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.146728504516838e-06,
+      "loss": 0.5679,
+      "step": 13601
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.145349177606773e-06,
+      "loss": 0.5737,
+      "step": 13602
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.143969909823596e-06,
+      "loss": 0.5464,
+      "step": 13603
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.142590701195886e-06,
+      "loss": 0.6169,
+      "step": 13604
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.141211551752203e-06,
+      "loss": 0.6113,
+      "step": 13605
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.139832461521108e-06,
+      "loss": 0.6233,
+      "step": 13606
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.138453430531176e-06,
+      "loss": 0.5759,
+      "step": 13607
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.13707445881096e-06,
+      "loss": 0.5498,
+      "step": 13608
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.135695546389031e-06,
+      "loss": 0.5908,
+      "step": 13609
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1343166932939415e-06,
+      "loss": 0.5569,
+      "step": 13610
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.132937899554256e-06,
+      "loss": 0.574,
+      "step": 13611
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.131559165198533e-06,
+      "loss": 0.5237,
+      "step": 13612
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.130180490255327e-06,
+      "loss": 0.5767,
+      "step": 13613
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.128801874753193e-06,
+      "loss": 0.5632,
+      "step": 13614
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.127423318720691e-06,
+      "loss": 0.552,
+      "step": 13615
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.126044822186366e-06,
+      "loss": 0.6516,
+      "step": 13616
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.124666385178781e-06,
+      "loss": 0.5562,
+      "step": 13617
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.123288007726475e-06,
+      "loss": 0.5566,
+      "step": 13618
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.121909689858009e-06,
+      "loss": 0.5303,
+      "step": 13619
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1205314316019245e-06,
+      "loss": 0.6069,
+      "step": 13620
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.119153232986764e-06,
+      "loss": 0.5386,
+      "step": 13621
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1177750940410865e-06,
+      "loss": 0.5962,
+      "step": 13622
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.116397014793422e-06,
+      "loss": 0.6152,
+      "step": 13623
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.115018995272326e-06,
+      "loss": 0.5767,
+      "step": 13624
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.113641035506335e-06,
+      "loss": 0.5815,
+      "step": 13625
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.11226313552399e-06,
+      "loss": 0.5791,
+      "step": 13626
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.110885295353831e-06,
+      "loss": 0.5513,
+      "step": 13627
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.109507515024401e-06,
+      "loss": 0.5564,
+      "step": 13628
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.108129794564228e-06,
+      "loss": 0.6172,
+      "step": 13629
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.106752134001859e-06,
+      "loss": 0.5632,
+      "step": 13630
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.105374533365819e-06,
+      "loss": 0.6235,
+      "step": 13631
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.103996992684648e-06,
+      "loss": 0.4971,
+      "step": 13632
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1026195119868736e-06,
+      "loss": 0.5779,
+      "step": 13633
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.101242091301032e-06,
+      "loss": 0.5535,
+      "step": 13634
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.09986473065565e-06,
+      "loss": 0.6262,
+      "step": 13635
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.098487430079255e-06,
+      "loss": 0.5403,
+      "step": 13636
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.097110189600376e-06,
+      "loss": 0.6191,
+      "step": 13637
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.095733009247536e-06,
+      "loss": 0.5471,
+      "step": 13638
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.094355889049267e-06,
+      "loss": 0.5657,
+      "step": 13639
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.092978829034085e-06,
+      "loss": 0.5903,
+      "step": 13640
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.091601829230519e-06,
+      "loss": 0.5308,
+      "step": 13641
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.090224889667081e-06,
+      "loss": 0.5674,
+      "step": 13642
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.088848010372302e-06,
+      "loss": 0.5186,
+      "step": 13643
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.087471191374691e-06,
+      "loss": 0.6228,
+      "step": 13644
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.086094432702772e-06,
+      "loss": 0.6929,
+      "step": 13645
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0847177343850555e-06,
+      "loss": 0.6516,
+      "step": 13646
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.083341096450063e-06,
+      "loss": 0.5916,
+      "step": 13647
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.081964518926301e-06,
+      "loss": 0.6162,
+      "step": 13648
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.080588001842288e-06,
+      "loss": 0.6196,
+      "step": 13649
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.079211545226529e-06,
+      "loss": 0.5974,
+      "step": 13650
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.077835149107537e-06,
+      "loss": 0.604,
+      "step": 13651
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.076458813513825e-06,
+      "loss": 0.595,
+      "step": 13652
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.07508253847389e-06,
+      "loss": 0.5583,
+      "step": 13653
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.073706324016249e-06,
+      "loss": 0.5874,
+      "step": 13654
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.072330170169397e-06,
+      "loss": 0.4839,
+      "step": 13655
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.070954076961847e-06,
+      "loss": 0.6086,
+      "step": 13656
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.069578044422092e-06,
+      "loss": 0.5933,
+      "step": 13657
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.068202072578642e-06,
+      "loss": 0.5889,
+      "step": 13658
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.066826161459989e-06,
+      "loss": 0.616,
+      "step": 13659
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.065450311094639e-06,
+      "loss": 0.5745,
+      "step": 13660
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.064074521511079e-06,
+      "loss": 0.5354,
+      "step": 13661
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.062698792737819e-06,
+      "loss": 0.542,
+      "step": 13662
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.06132312480334e-06,
+      "loss": 0.5898,
+      "step": 13663
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.059947517736143e-06,
+      "loss": 0.5544,
+      "step": 13664
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.058571971564722e-06,
+      "loss": 0.6414,
+      "step": 13665
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.057196486317559e-06,
+      "loss": 0.5691,
+      "step": 13666
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.055821062023156e-06,
+      "loss": 0.5708,
+      "step": 13667
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.054445698709988e-06,
+      "loss": 0.6072,
+      "step": 13668
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.053070396406555e-06,
+      "loss": 0.5779,
+      "step": 13669
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.051695155141333e-06,
+      "loss": 0.5503,
+      "step": 13670
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0503199749428145e-06,
+      "loss": 0.5891,
+      "step": 13671
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.048944855839473e-06,
+      "loss": 0.5764,
+      "step": 13672
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.047569797859803e-06,
+      "loss": 0.5618,
+      "step": 13673
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.046194801032273e-06,
+      "loss": 0.5044,
+      "step": 13674
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.044819865385373e-06,
+      "loss": 0.5598,
+      "step": 13675
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.043444990947576e-06,
+      "loss": 0.501,
+      "step": 13676
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.04207017774736e-06,
+      "loss": 0.6094,
+      "step": 13677
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.040695425813197e-06,
+      "loss": 0.6533,
+      "step": 13678
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0393207351735715e-06,
+      "loss": 0.5623,
+      "step": 13679
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.03794610585695e-06,
+      "loss": 0.5498,
+      "step": 13680
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.036571537891801e-06,
+      "loss": 0.5132,
+      "step": 13681
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.035197031306602e-06,
+      "loss": 0.5454,
+      "step": 13682
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.033822586129817e-06,
+      "loss": 0.5188,
+      "step": 13683
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.032448202389921e-06,
+      "loss": 0.6179,
+      "step": 13684
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.031073880115371e-06,
+      "loss": 0.6089,
+      "step": 13685
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.029699619334644e-06,
+      "loss": 0.4995,
+      "step": 13686
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0283254200761944e-06,
+      "loss": 0.5659,
+      "step": 13687
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.026951282368491e-06,
+      "loss": 0.5576,
+      "step": 13688
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.025577206239996e-06,
+      "loss": 0.6631,
+      "step": 13689
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0242031917191655e-06,
+      "loss": 0.5557,
+      "step": 13690
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.022829238834461e-06,
+      "loss": 0.6724,
+      "step": 13691
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.021455347614343e-06,
+      "loss": 0.5735,
+      "step": 13692
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.020081518087263e-06,
+      "loss": 0.562,
+      "step": 13693
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.018707750281684e-06,
+      "loss": 0.5764,
+      "step": 13694
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.017334044226055e-06,
+      "loss": 0.6943,
+      "step": 13695
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.015960399948825e-06,
+      "loss": 0.5013,
+      "step": 13696
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.014586817478455e-06,
+      "loss": 0.6338,
+      "step": 13697
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0132132968433865e-06,
+      "loss": 0.5872,
+      "step": 13698
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.011839838072074e-06,
+      "loss": 0.5381,
+      "step": 13699
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.010466441192962e-06,
+      "loss": 0.553,
+      "step": 13700
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.009093106234499e-06,
+      "loss": 0.6465,
+      "step": 13701
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0077198332251304e-06,
+      "loss": 0.53,
+      "step": 13702
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.006346622193299e-06,
+      "loss": 0.5759,
+      "step": 13703
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0049734731674445e-06,
+      "loss": 0.5264,
+      "step": 13704
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.003600386176017e-06,
+      "loss": 0.5056,
+      "step": 13705
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.002227361247446e-06,
+      "loss": 0.5186,
+      "step": 13706
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.000854398410182e-06,
+      "loss": 0.6287,
+      "step": 13707
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.999481497692648e-06,
+      "loss": 0.5662,
+      "step": 13708
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.998108659123294e-06,
+      "loss": 0.6069,
+      "step": 13709
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.996735882730548e-06,
+      "loss": 0.5908,
+      "step": 13710
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.99536316854284e-06,
+      "loss": 0.5771,
+      "step": 13711
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.993990516588609e-06,
+      "loss": 0.6548,
+      "step": 13712
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.992617926896284e-06,
+      "loss": 0.5405,
+      "step": 13713
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9912453994942955e-06,
+      "loss": 0.6089,
+      "step": 13714
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.989872934411066e-06,
+      "loss": 0.5361,
+      "step": 13715
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.988500531675034e-06,
+      "loss": 0.5496,
+      "step": 13716
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.987128191314614e-06,
+      "loss": 0.5381,
+      "step": 13717
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.98575591335824e-06,
+      "loss": 0.5764,
+      "step": 13718
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.984383697834325e-06,
+      "loss": 0.5283,
+      "step": 13719
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.983011544771303e-06,
+      "loss": 0.5247,
+      "step": 13720
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.981639454197583e-06,
+      "loss": 0.623,
+      "step": 13721
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9802674261415935e-06,
+      "loss": 0.5483,
+      "step": 13722
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9788954606317505e-06,
+      "loss": 0.5383,
+      "step": 13723
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.977523557696464e-06,
+      "loss": 0.542,
+      "step": 13724
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.976151717364157e-06,
+      "loss": 0.5742,
+      "step": 13725
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.974779939663241e-06,
+      "loss": 0.573,
+      "step": 13726
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.97340822462213e-06,
+      "loss": 0.6021,
+      "step": 13727
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.972036572269233e-06,
+      "loss": 0.4724,
+      "step": 13728
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.970664982632967e-06,
+      "loss": 0.5566,
+      "step": 13729
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.96929345574173e-06,
+      "loss": 0.5576,
+      "step": 13730
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.967921991623942e-06,
+      "loss": 0.5374,
+      "step": 13731
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.966550590307999e-06,
+      "loss": 0.6829,
+      "step": 13732
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.965179251822314e-06,
+      "loss": 0.6267,
+      "step": 13733
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.963807976195283e-06,
+      "loss": 0.5798,
+      "step": 13734
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.962436763455318e-06,
+      "loss": 0.5208,
+      "step": 13735
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.961065613630811e-06,
+      "loss": 0.6492,
+      "step": 13736
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.959694526750169e-06,
+      "loss": 0.5918,
+      "step": 13737
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.958323502841786e-06,
+      "loss": 0.5808,
+      "step": 13738
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9569525419340615e-06,
+      "loss": 0.6909,
+      "step": 13739
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9555816440553916e-06,
+      "loss": 0.4797,
+      "step": 13740
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.954210809234167e-06,
+      "loss": 0.5649,
+      "step": 13741
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.952840037498789e-06,
+      "loss": 0.6123,
+      "step": 13742
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9514693288776405e-06,
+      "loss": 0.5352,
+      "step": 13743
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.950098683399121e-06,
+      "loss": 0.6318,
+      "step": 13744
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.948728101091613e-06,
+      "loss": 0.6189,
+      "step": 13745
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.94735758198351e-06,
+      "loss": 0.6504,
+      "step": 13746
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.945987126103193e-06,
+      "loss": 0.5649,
+      "step": 13747
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.944616733479055e-06,
+      "loss": 0.5444,
+      "step": 13748
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.943246404139472e-06,
+      "loss": 0.6284,
+      "step": 13749
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.941876138112835e-06,
+      "loss": 0.5938,
+      "step": 13750
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.940505935427517e-06,
+      "loss": 0.5876,
+      "step": 13751
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.939135796111906e-06,
+      "loss": 0.6196,
+      "step": 13752
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.93776572019438e-06,
+      "loss": 0.627,
+      "step": 13753
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.936395707703309e-06,
+      "loss": 0.5549,
+      "step": 13754
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.93502575866708e-06,
+      "loss": 0.5786,
+      "step": 13755
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.933655873114058e-06,
+      "loss": 0.5481,
+      "step": 13756
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.932286051072627e-06,
+      "loss": 0.5435,
+      "step": 13757
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9309162925711495e-06,
+      "loss": 0.4558,
+      "step": 13758
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.929546597638005e-06,
+      "loss": 0.5625,
+      "step": 13759
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.928176966301555e-06,
+      "loss": 0.6201,
+      "step": 13760
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.926807398590177e-06,
+      "loss": 0.5747,
+      "step": 13761
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.92543789453223e-06,
+      "loss": 0.571,
+      "step": 13762
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9240684541560875e-06,
+      "loss": 0.5044,
+      "step": 13763
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.922699077490105e-06,
+      "loss": 0.5688,
+      "step": 13764
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.921329764562652e-06,
+      "loss": 0.5791,
+      "step": 13765
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.919960515402091e-06,
+      "loss": 0.6326,
+      "step": 13766
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.918591330036779e-06,
+      "loss": 0.5583,
+      "step": 13767
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9172222084950794e-06,
+      "loss": 0.5854,
+      "step": 13768
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.915853150805343e-06,
+      "loss": 0.5347,
+      "step": 13769
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.914484156995934e-06,
+      "loss": 0.584,
+      "step": 13770
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.913115227095202e-06,
+      "loss": 0.573,
+      "step": 13771
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.911746361131506e-06,
+      "loss": 0.501,
+      "step": 13772
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.910377559133192e-06,
+      "loss": 0.5806,
+      "step": 13773
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.90900882112862e-06,
+      "loss": 0.5913,
+      "step": 13774
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.907640147146131e-06,
+      "loss": 0.5714,
+      "step": 13775
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.906271537214082e-06,
+      "loss": 0.5354,
+      "step": 13776
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.904902991360812e-06,
+      "loss": 0.5854,
+      "step": 13777
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.903534509614673e-06,
+      "loss": 0.593,
+      "step": 13778
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.902166092004008e-06,
+      "loss": 0.5798,
+      "step": 13779
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.900797738557159e-06,
+      "loss": 0.593,
+      "step": 13780
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.899429449302469e-06,
+      "loss": 0.5801,
+      "step": 13781
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.898061224268281e-06,
+      "loss": 0.5605,
+      "step": 13782
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.896693063482934e-06,
+      "loss": 0.6196,
+      "step": 13783
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.895324966974757e-06,
+      "loss": 0.5315,
+      "step": 13784
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.893956934772099e-06,
+      "loss": 0.6167,
+      "step": 13785
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.892588966903288e-06,
+      "loss": 0.4775,
+      "step": 13786
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.891221063396662e-06,
+      "loss": 0.571,
+      "step": 13787
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.889853224280548e-06,
+      "loss": 0.5327,
+      "step": 13788
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.888485449583284e-06,
+      "loss": 0.6047,
+      "step": 13789
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.887117739333195e-06,
+      "loss": 0.6697,
+      "step": 13790
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.885750093558614e-06,
+      "loss": 0.6162,
+      "step": 13791
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.884382512287864e-06,
+      "loss": 0.6194,
+      "step": 13792
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.883014995549277e-06,
+      "loss": 0.6111,
+      "step": 13793
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.881647543371169e-06,
+      "loss": 0.5784,
+      "step": 13794
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.880280155781871e-06,
+      "loss": 0.5828,
+      "step": 13795
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8789128328097e-06,
+      "loss": 0.5859,
+      "step": 13796
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.877545574482983e-06,
+      "loss": 0.627,
+      "step": 13797
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.876178380830033e-06,
+      "loss": 0.5227,
+      "step": 13798
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.874811251879168e-06,
+      "loss": 0.5947,
+      "step": 13799
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.873444187658709e-06,
+      "loss": 0.5464,
+      "step": 13800
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.872077188196965e-06,
+      "loss": 0.5166,
+      "step": 13801
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.870710253522257e-06,
+      "loss": 0.5583,
+      "step": 13802
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.869343383662893e-06,
+      "loss": 0.5569,
+      "step": 13803
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.867976578647186e-06,
+      "loss": 0.5093,
+      "step": 13804
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.866609838503444e-06,
+      "loss": 0.5579,
+      "step": 13805
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.865243163259982e-06,
+      "loss": 0.5813,
+      "step": 13806
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.863876552945097e-06,
+      "loss": 0.5945,
+      "step": 13807
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.862510007587107e-06,
+      "loss": 0.585,
+      "step": 13808
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.861143527214302e-06,
+      "loss": 0.6128,
+      "step": 13809
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.859777111855e-06,
+      "loss": 0.5649,
+      "step": 13810
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8584107615374905e-06,
+      "loss": 0.5947,
+      "step": 13811
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.857044476290085e-06,
+      "loss": 0.5552,
+      "step": 13812
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.855678256141072e-06,
+      "loss": 0.6387,
+      "step": 13813
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.854312101118756e-06,
+      "loss": 0.5623,
+      "step": 13814
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8529460112514335e-06,
+      "loss": 0.5671,
+      "step": 13815
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.851579986567396e-06,
+      "loss": 0.5071,
+      "step": 13816
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.850214027094939e-06,
+      "loss": 0.614,
+      "step": 13817
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.848848132862353e-06,
+      "loss": 0.6047,
+      "step": 13818
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.847482303897936e-06,
+      "loss": 0.5364,
+      "step": 13819
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.846116540229969e-06,
+      "loss": 0.6296,
+      "step": 13820
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.844750841886748e-06,
+      "loss": 0.5781,
+      "step": 13821
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8433852088965515e-06,
+      "loss": 0.5232,
+      "step": 13822
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.842019641287675e-06,
+      "loss": 0.5361,
+      "step": 13823
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.840654139088393e-06,
+      "loss": 0.5833,
+      "step": 13824
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.839288702326997e-06,
+      "loss": 0.562,
+      "step": 13825
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.837923331031761e-06,
+      "loss": 0.5654,
+      "step": 13826
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.836558025230973e-06,
+      "loss": 0.5884,
+      "step": 13827
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.835192784952904e-06,
+      "loss": 0.49,
+      "step": 13828
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.833827610225835e-06,
+      "loss": 0.5803,
+      "step": 13829
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.832462501078045e-06,
+      "loss": 0.6101,
+      "step": 13830
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.831097457537803e-06,
+      "loss": 0.6094,
+      "step": 13831
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.829732479633391e-06,
+      "loss": 0.5286,
+      "step": 13832
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.82836756739307e-06,
+      "loss": 0.616,
+      "step": 13833
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.827002720845121e-06,
+      "loss": 0.522,
+      "step": 13834
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.825637940017804e-06,
+      "loss": 0.606,
+      "step": 13835
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.824273224939397e-06,
+      "loss": 0.5808,
+      "step": 13836
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.822908575638156e-06,
+      "loss": 0.55,
+      "step": 13837
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.821543992142357e-06,
+      "loss": 0.5298,
+      "step": 13838
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.820179474480253e-06,
+      "loss": 0.5708,
+      "step": 13839
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.818815022680116e-06,
+      "loss": 0.5864,
+      "step": 13840
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.8174506367702024e-06,
+      "loss": 0.5408,
+      "step": 13841
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.816086316778772e-06,
+      "loss": 0.6582,
+      "step": 13842
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.814722062734087e-06,
+      "loss": 0.6477,
+      "step": 13843
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.813357874664397e-06,
+      "loss": 0.5393,
+      "step": 13844
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.8119937525979675e-06,
+      "loss": 0.5068,
+      "step": 13845
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.810629696563043e-06,
+      "loss": 0.5728,
+      "step": 13846
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.809265706587884e-06,
+      "loss": 0.6641,
+      "step": 13847
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.807901782700736e-06,
+      "loss": 0.6357,
+      "step": 13848
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.806537924929858e-06,
+      "loss": 0.5061,
+      "step": 13849
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.805174133303488e-06,
+      "loss": 0.6372,
+      "step": 13850
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.803810407849883e-06,
+      "loss": 0.572,
+      "step": 13851
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.802446748597282e-06,
+      "loss": 0.5713,
+      "step": 13852
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.801083155573935e-06,
+      "loss": 0.5139,
+      "step": 13853
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.799719628808081e-06,
+      "loss": 0.5786,
+      "step": 13854
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.798356168327966e-06,
+      "loss": 0.5942,
+      "step": 13855
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.796992774161828e-06,
+      "loss": 0.5317,
+      "step": 13856
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.795629446337908e-06,
+      "loss": 0.5615,
+      "step": 13857
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7942661848844435e-06,
+      "loss": 0.5698,
+      "step": 13858
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.792902989829668e-06,
+      "loss": 0.5803,
+      "step": 13859
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.791539861201823e-06,
+      "loss": 0.5894,
+      "step": 13860
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.790176799029134e-06,
+      "loss": 0.5422,
+      "step": 13861
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.788813803339842e-06,
+      "loss": 0.5513,
+      "step": 13862
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.787450874162169e-06,
+      "loss": 0.6333,
+      "step": 13863
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.786088011524356e-06,
+      "loss": 0.6831,
+      "step": 13864
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.784725215454619e-06,
+      "loss": 0.5442,
+      "step": 13865
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.783362485981193e-06,
+      "loss": 0.5781,
+      "step": 13866
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7819998231323e-06,
+      "loss": 0.5779,
+      "step": 13867
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.780637226936166e-06,
+      "loss": 0.5808,
+      "step": 13868
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.779274697421012e-06,
+      "loss": 0.5347,
+      "step": 13869
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.777912234615064e-06,
+      "loss": 0.5981,
+      "step": 13870
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.776549838546534e-06,
+      "loss": 0.5847,
+      "step": 13871
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.775187509243649e-06,
+      "loss": 0.5459,
+      "step": 13872
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.773825246734622e-06,
+      "loss": 0.688,
+      "step": 13873
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.772463051047667e-06,
+      "loss": 0.6089,
+      "step": 13874
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.771100922211001e-06,
+      "loss": 0.541,
+      "step": 13875
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.769738860252835e-06,
+      "loss": 0.5813,
+      "step": 13876
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.768376865201385e-06,
+      "loss": 0.6335,
+      "step": 13877
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.767014937084855e-06,
+      "loss": 0.5728,
+      "step": 13878
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.76565307593146e-06,
+      "loss": 0.5603,
+      "step": 13879
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.764291281769404e-06,
+      "loss": 0.5969,
+      "step": 13880
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7629295546268945e-06,
+      "loss": 0.6055,
+      "step": 13881
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.761567894532135e-06,
+      "loss": 0.5498,
+      "step": 13882
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7602063015133324e-06,
+      "loss": 0.5361,
+      "step": 13883
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.758844775598681e-06,
+      "loss": 0.5115,
+      "step": 13884
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.757483316816391e-06,
+      "loss": 0.6855,
+      "step": 13885
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.756121925194654e-06,
+      "loss": 0.5684,
+      "step": 13886
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7547606007616735e-06,
+      "loss": 0.5942,
+      "step": 13887
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.753399343545644e-06,
+      "loss": 0.5303,
+      "step": 13888
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.752038153574755e-06,
+      "loss": 0.5105,
+      "step": 13889
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.750677030877206e-06,
+      "loss": 0.5579,
+      "step": 13890
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.74931597548119e-06,
+      "loss": 0.5342,
+      "step": 13891
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.747954987414894e-06,
+      "loss": 0.5498,
+      "step": 13892
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.74659406670651e-06,
+      "loss": 0.5696,
+      "step": 13893
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.745233213384225e-06,
+      "loss": 0.627,
+      "step": 13894
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.743872427476224e-06,
+      "loss": 0.5364,
+      "step": 13895
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7425117090106974e-06,
+      "loss": 0.5229,
+      "step": 13896
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.741151058015823e-06,
+      "loss": 0.5354,
+      "step": 13897
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.73979047451979e-06,
+      "loss": 0.5918,
+      "step": 13898
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.73842995855077e-06,
+      "loss": 0.5466,
+      "step": 13899
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.737069510136953e-06,
+      "loss": 0.6062,
+      "step": 13900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.735709129306509e-06,
+      "loss": 0.6304,
+      "step": 13901
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.734348816087621e-06,
+      "loss": 0.6055,
+      "step": 13902
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.732988570508461e-06,
+      "loss": 0.5559,
+      "step": 13903
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.731628392597204e-06,
+      "loss": 0.5657,
+      "step": 13904
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.730268282382022e-06,
+      "loss": 0.5627,
+      "step": 13905
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.728908239891083e-06,
+      "loss": 0.5225,
+      "step": 13906
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.727548265152567e-06,
+      "loss": 0.5676,
+      "step": 13907
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.726188358194631e-06,
+      "loss": 0.5703,
+      "step": 13908
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.724828519045451e-06,
+      "loss": 0.6047,
+      "step": 13909
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7234687477331836e-06,
+      "loss": 0.635,
+      "step": 13910
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.722109044286005e-06,
+      "loss": 0.5979,
+      "step": 13911
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.720749408732065e-06,
+      "loss": 0.5693,
+      "step": 13912
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.719389841099535e-06,
+      "loss": 0.6382,
+      "step": 13913
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.718030341416568e-06,
+      "loss": 0.5991,
+      "step": 13914
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.716670909711329e-06,
+      "loss": 0.5637,
+      "step": 13915
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.715311546011969e-06,
+      "loss": 0.5806,
+      "step": 13916
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7139522503466506e-06,
+      "loss": 0.541,
+      "step": 13917
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.712593022743522e-06,
+      "loss": 0.5212,
+      "step": 13918
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.711233863230738e-06,
+      "loss": 0.4686,
+      "step": 13919
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.709874771836451e-06,
+      "loss": 0.5769,
+      "step": 13920
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.70851574858881e-06,
+      "loss": 0.5771,
+      "step": 13921
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.707156793515968e-06,
+      "loss": 0.5244,
+      "step": 13922
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.705797906646063e-06,
+      "loss": 0.571,
+      "step": 13923
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.704439088007253e-06,
+      "loss": 0.5474,
+      "step": 13924
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.703080337627672e-06,
+      "loss": 0.6333,
+      "step": 13925
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.701721655535471e-06,
+      "loss": 0.5405,
+      "step": 13926
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.700363041758784e-06,
+      "loss": 0.5879,
+      "step": 13927
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.69900449632576e-06,
+      "loss": 0.5166,
+      "step": 13928
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.697646019264529e-06,
+      "loss": 0.6328,
+      "step": 13929
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.696287610603236e-06,
+      "loss": 0.6401,
+      "step": 13930
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.694929270370012e-06,
+      "loss": 0.5886,
+      "step": 13931
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.693570998592991e-06,
+      "loss": 0.519,
+      "step": 13932
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.692212795300312e-06,
+      "loss": 0.5134,
+      "step": 13933
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.690854660520101e-06,
+      "loss": 0.614,
+      "step": 13934
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6894965942804935e-06,
+      "loss": 0.5405,
+      "step": 13935
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.68813859660961e-06,
+      "loss": 0.5862,
+      "step": 13936
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.686780667535589e-06,
+      "loss": 0.5547,
+      "step": 13937
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.685422807086545e-06,
+      "loss": 0.5735,
+      "step": 13938
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.684065015290615e-06,
+      "loss": 0.572,
+      "step": 13939
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.682707292175912e-06,
+      "loss": 0.583,
+      "step": 13940
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.681349637770565e-06,
+      "loss": 0.6467,
+      "step": 13941
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.679992052102687e-06,
+      "loss": 0.615,
+      "step": 13942
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6786345352004016e-06,
+      "loss": 0.5559,
+      "step": 13943
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.677277087091827e-06,
+      "loss": 0.6221,
+      "step": 13944
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.675919707805077e-06,
+      "loss": 0.5264,
+      "step": 13945
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.674562397368272e-06,
+      "loss": 0.6147,
+      "step": 13946
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6732051558095145e-06,
+      "loss": 0.5254,
+      "step": 13947
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.671847983156926e-06,
+      "loss": 0.5906,
+      "step": 13948
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.67049087943861e-06,
+      "loss": 0.5349,
+      "step": 13949
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.669133844682684e-06,
+      "loss": 0.5911,
+      "step": 13950
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6677768789172445e-06,
+      "loss": 0.6165,
+      "step": 13951
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.666419982170409e-06,
+      "loss": 0.6475,
+      "step": 13952
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.665063154470273e-06,
+      "loss": 0.5791,
+      "step": 13953
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6637063958449465e-06,
+      "loss": 0.6479,
+      "step": 13954
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.662349706322524e-06,
+      "loss": 0.5259,
+      "step": 13955
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.660993085931113e-06,
+      "loss": 0.5383,
+      "step": 13956
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.659636534698809e-06,
+      "loss": 0.614,
+      "step": 13957
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6582800526537116e-06,
+      "loss": 0.5898,
+      "step": 13958
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.656923639823911e-06,
+      "loss": 0.5454,
+      "step": 13959
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6555672962375125e-06,
+      "loss": 0.5862,
+      "step": 13960
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.654211021922603e-06,
+      "loss": 0.6011,
+      "step": 13961
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.65285481690727e-06,
+      "loss": 0.6211,
+      "step": 13962
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.651498681219612e-06,
+      "loss": 0.5715,
+      "step": 13963
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.650142614887712e-06,
+      "loss": 0.5103,
+      "step": 13964
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6487866179396634e-06,
+      "loss": 0.5481,
+      "step": 13965
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6474306904035445e-06,
+      "loss": 0.6016,
+      "step": 13966
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.646074832307445e-06,
+      "loss": 0.6331,
+      "step": 13967
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.644719043679448e-06,
+      "loss": 0.5068,
+      "step": 13968
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.643363324547635e-06,
+      "loss": 0.5549,
+      "step": 13969
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.642007674940085e-06,
+      "loss": 0.554,
+      "step": 13970
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.640652094884877e-06,
+      "loss": 0.5596,
+      "step": 13971
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.639296584410089e-06,
+      "loss": 0.5381,
+      "step": 13972
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.637941143543799e-06,
+      "loss": 0.606,
+      "step": 13973
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.636585772314077e-06,
+      "loss": 0.6267,
+      "step": 13974
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.635230470749001e-06,
+      "loss": 0.5166,
+      "step": 13975
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.633875238876641e-06,
+      "loss": 0.5857,
+      "step": 13976
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6325200767250644e-06,
+      "loss": 0.5034,
+      "step": 13977
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.631164984322344e-06,
+      "loss": 0.5803,
+      "step": 13978
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.62980996169654e-06,
+      "loss": 0.5615,
+      "step": 13979
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.628455008875728e-06,
+      "loss": 0.5361,
+      "step": 13980
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6271001258879675e-06,
+      "loss": 0.5522,
+      "step": 13981
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.62574531276132e-06,
+      "loss": 0.4854,
+      "step": 13982
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.624390569523849e-06,
+      "loss": 0.5195,
+      "step": 13983
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.623035896203618e-06,
+      "loss": 0.5947,
+      "step": 13984
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6216812928286786e-06,
+      "loss": 0.5579,
+      "step": 13985
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.620326759427094e-06,
+      "loss": 0.5632,
+      "step": 13986
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.618972296026915e-06,
+      "loss": 0.5674,
+      "step": 13987
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.617617902656201e-06,
+      "loss": 0.5703,
+      "step": 13988
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.616263579343e-06,
+      "loss": 0.5974,
+      "step": 13989
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6149093261153686e-06,
+      "loss": 0.5627,
+      "step": 13990
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.613555143001354e-06,
+      "loss": 0.5571,
+      "step": 13991
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6122010300289996e-06,
+      "loss": 0.6294,
+      "step": 13992
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.61084698722636e-06,
+      "loss": 0.6719,
+      "step": 13993
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.609493014621478e-06,
+      "loss": 0.5894,
+      "step": 13994
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.608139112242398e-06,
+      "loss": 0.5718,
+      "step": 13995
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.606785280117161e-06,
+      "loss": 0.5945,
+      "step": 13996
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.605431518273812e-06,
+      "loss": 0.5574,
+      "step": 13997
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6040778267403846e-06,
+      "loss": 0.5505,
+      "step": 13998
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.602724205544926e-06,
+      "loss": 0.5674,
+      "step": 13999
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.601370654715465e-06,
+      "loss": 0.6013,
+      "step": 14000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.600017174280042e-06,
+      "loss": 0.5947,
+      "step": 14001
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.598663764266686e-06,
+      "loss": 0.5356,
+      "step": 14002
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5973104247034355e-06,
+      "loss": 0.5112,
+      "step": 14003
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.595957155618315e-06,
+      "loss": 0.6138,
+      "step": 14004
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.594603957039362e-06,
+      "loss": 0.5681,
+      "step": 14005
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.593250828994597e-06,
+      "loss": 0.564,
+      "step": 14006
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.59189777151205e-06,
+      "loss": 0.519,
+      "step": 14007
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.590544784619746e-06,
+      "loss": 0.5537,
+      "step": 14008
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5891918683457064e-06,
+      "loss": 0.491,
+      "step": 14009
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.587839022717961e-06,
+      "loss": 0.4878,
+      "step": 14010
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.586486247764519e-06,
+      "loss": 0.5813,
+      "step": 14011
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.585133543513411e-06,
+      "loss": 0.4736,
+      "step": 14012
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.583780909992646e-06,
+      "loss": 0.5444,
+      "step": 14013
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5824283472302484e-06,
+      "loss": 0.6099,
+      "step": 14014
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.581075855254224e-06,
+      "loss": 0.5579,
+      "step": 14015
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.579723434092598e-06,
+      "loss": 0.605,
+      "step": 14016
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.578371083773368e-06,
+      "loss": 0.5605,
+      "step": 14017
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5770188043245575e-06,
+      "loss": 0.5542,
+      "step": 14018
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.575666595774166e-06,
+      "loss": 0.5134,
+      "step": 14019
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.574314458150208e-06,
+      "loss": 0.5869,
+      "step": 14020
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.572962391480689e-06,
+      "loss": 0.5354,
+      "step": 14021
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.571610395793606e-06,
+      "loss": 0.6157,
+      "step": 14022
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.570258471116973e-06,
+      "loss": 0.594,
+      "step": 14023
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.568906617478781e-06,
+      "loss": 0.5818,
+      "step": 14024
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.567554834907041e-06,
+      "loss": 0.5569,
+      "step": 14025
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.566203123429742e-06,
+      "loss": 0.575,
+      "step": 14026
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.56485148307489e-06,
+      "loss": 0.5227,
+      "step": 14027
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.563499913870471e-06,
+      "loss": 0.6064,
+      "step": 14028
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.562148415844491e-06,
+      "loss": 0.5242,
+      "step": 14029
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5607969890249315e-06,
+      "loss": 0.5486,
+      "step": 14030
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.559445633439794e-06,
+      "loss": 0.6204,
+      "step": 14031
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.558094349117059e-06,
+      "loss": 0.603,
+      "step": 14032
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.556743136084722e-06,
+      "loss": 0.5737,
+      "step": 14033
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.555391994370767e-06,
+      "loss": 0.511,
+      "step": 14034
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.55404092400318e-06,
+      "loss": 0.6392,
+      "step": 14035
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.552689925009947e-06,
+      "loss": 0.5024,
+      "step": 14036
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.551338997419044e-06,
+      "loss": 0.6704,
+      "step": 14037
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.549988141258462e-06,
+      "loss": 0.5281,
+      "step": 14038
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.548637356556171e-06,
+      "loss": 0.5854,
+      "step": 14039
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5472866433401565e-06,
+      "loss": 0.5615,
+      "step": 14040
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5459360016383875e-06,
+      "loss": 0.5266,
+      "step": 14041
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.544585431478849e-06,
+      "loss": 0.5188,
+      "step": 14042
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5432349328895036e-06,
+      "loss": 0.6226,
+      "step": 14043
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.541884505898333e-06,
+      "loss": 0.542,
+      "step": 14044
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.540534150533302e-06,
+      "loss": 0.5469,
+      "step": 14045
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.539183866822383e-06,
+      "loss": 0.5356,
+      "step": 14046
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.537833654793542e-06,
+      "loss": 0.5574,
+      "step": 14047
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.536483514474745e-06,
+      "loss": 0.5752,
+      "step": 14048
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5351334458939565e-06,
+      "loss": 0.6355,
+      "step": 14049
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.533783449079144e-06,
+      "loss": 0.5327,
+      "step": 14050
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.532433524058266e-06,
+      "loss": 0.5455,
+      "step": 14051
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.53108367085928e-06,
+      "loss": 0.5603,
+      "step": 14052
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.52973388951015e-06,
+      "loss": 0.5454,
+      "step": 14053
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5283841800388295e-06,
+      "loss": 0.5955,
+      "step": 14054
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.527034542473278e-06,
+      "loss": 0.5396,
+      "step": 14055
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.525684976841442e-06,
+      "loss": 0.468,
+      "step": 14056
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.524335483171285e-06,
+      "loss": 0.5808,
+      "step": 14057
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.52298606149075e-06,
+      "loss": 0.582,
+      "step": 14058
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.521636711827792e-06,
+      "loss": 0.573,
+      "step": 14059
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.520287434210357e-06,
+      "loss": 0.6284,
+      "step": 14060
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.518938228666395e-06,
+      "loss": 0.5227,
+      "step": 14061
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.517589095223844e-06,
+      "loss": 0.656,
+      "step": 14062
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.516240033910658e-06,
+      "loss": 0.5901,
+      "step": 14063
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.51489104475477e-06,
+      "loss": 0.5386,
+      "step": 14064
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5135421277841285e-06,
+      "loss": 0.6406,
+      "step": 14065
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.512193283026668e-06,
+      "loss": 0.5632,
+      "step": 14066
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.510844510510327e-06,
+      "loss": 0.5496,
+      "step": 14067
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.509495810263044e-06,
+      "loss": 0.561,
+      "step": 14068
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.5081471823127515e-06,
+      "loss": 0.5459,
+      "step": 14069
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.506798626687384e-06,
+      "loss": 0.5007,
+      "step": 14070
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.505450143414874e-06,
+      "loss": 0.5164,
+      "step": 14071
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.504101732523151e-06,
+      "loss": 0.5779,
+      "step": 14072
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.502753394040144e-06,
+      "loss": 0.5212,
+      "step": 14073
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.501405127993783e-06,
+      "loss": 0.5088,
+      "step": 14074
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.500056934411989e-06,
+      "loss": 0.563,
+      "step": 14075
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.498708813322694e-06,
+      "loss": 0.614,
+      "step": 14076
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.49736076475381e-06,
+      "loss": 0.5859,
+      "step": 14077
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.49601278873327e-06,
+      "loss": 0.592,
+      "step": 14078
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.494664885288984e-06,
+      "loss": 0.5713,
+      "step": 14079
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.493317054448879e-06,
+      "loss": 0.5576,
+      "step": 14080
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.491969296240865e-06,
+      "loss": 0.4946,
+      "step": 14081
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.490621610692862e-06,
+      "loss": 0.6289,
+      "step": 14082
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.489273997832781e-06,
+      "loss": 0.4836,
+      "step": 14083
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.487926457688536e-06,
+      "loss": 0.6509,
+      "step": 14084
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.486578990288037e-06,
+      "loss": 0.5967,
+      "step": 14085
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.485231595659191e-06,
+      "loss": 0.5679,
+      "step": 14086
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.483884273829914e-06,
+      "loss": 0.6211,
+      "step": 14087
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4825370248281024e-06,
+      "loss": 0.5745,
+      "step": 14088
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.48118984868167e-06,
+      "loss": 0.5974,
+      "step": 14089
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.479842745418512e-06,
+      "loss": 0.5227,
+      "step": 14090
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.478495715066539e-06,
+      "loss": 0.6108,
+      "step": 14091
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.477148757653642e-06,
+      "loss": 0.5642,
+      "step": 14092
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4758018732077274e-06,
+      "loss": 0.6052,
+      "step": 14093
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.474455061756686e-06,
+      "loss": 0.532,
+      "step": 14094
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.473108323328421e-06,
+      "loss": 0.5771,
+      "step": 14095
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.471761657950821e-06,
+      "loss": 0.5154,
+      "step": 14096
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.470415065651778e-06,
+      "loss": 0.5027,
+      "step": 14097
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.469068546459191e-06,
+      "loss": 0.6111,
+      "step": 14098
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.46772210040094e-06,
+      "loss": 0.6182,
+      "step": 14099
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.466375727504921e-06,
+      "loss": 0.5591,
+      "step": 14100
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.465029427799013e-06,
+      "loss": 0.6226,
+      "step": 14101
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.463683201311111e-06,
+      "loss": 0.5396,
+      "step": 14102
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.46233704806909e-06,
+      "loss": 0.6152,
+      "step": 14103
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4609909681008375e-06,
+      "loss": 0.593,
+      "step": 14104
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.459644961434228e-06,
+      "loss": 0.5339,
+      "step": 14105
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.45829902809715e-06,
+      "loss": 0.5835,
+      "step": 14106
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.456953168117472e-06,
+      "loss": 0.5747,
+      "step": 14107
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.455607381523079e-06,
+      "loss": 0.6042,
+      "step": 14108
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.454261668341835e-06,
+      "loss": 0.5898,
+      "step": 14109
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.452916028601621e-06,
+      "loss": 0.6091,
+      "step": 14110
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.451570462330308e-06,
+      "loss": 0.554,
+      "step": 14111
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.45022496955576e-06,
+      "loss": 0.5662,
+      "step": 14112
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.448879550305855e-06,
+      "loss": 0.5486,
+      "step": 14113
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.44753420460845e-06,
+      "loss": 0.5669,
+      "step": 14114
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.446188932491419e-06,
+      "loss": 0.5381,
+      "step": 14115
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.444843733982617e-06,
+      "loss": 0.616,
+      "step": 14116
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4434986091099175e-06,
+      "loss": 0.6094,
+      "step": 14117
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.442153557901169e-06,
+      "loss": 0.5972,
+      "step": 14118
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.440808580384243e-06,
+      "loss": 0.6089,
+      "step": 14119
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.439463676586988e-06,
+      "loss": 0.5171,
+      "step": 14120
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.438118846537264e-06,
+      "loss": 0.5217,
+      "step": 14121
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.436774090262925e-06,
+      "loss": 0.6904,
+      "step": 14122
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.435429407791827e-06,
+      "loss": 0.5701,
+      "step": 14123
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4340847991518175e-06,
+      "loss": 0.5828,
+      "step": 14124
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4327402643707495e-06,
+      "loss": 0.5171,
+      "step": 14125
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.431395803476472e-06,
+      "loss": 0.5764,
+      "step": 14126
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.430051416496828e-06,
+      "loss": 0.6067,
+      "step": 14127
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4287071034596695e-06,
+      "loss": 0.6062,
+      "step": 14128
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.427362864392832e-06,
+      "loss": 0.5977,
+      "step": 14129
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.426018699324168e-06,
+      "loss": 0.5146,
+      "step": 14130
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4246746082815104e-06,
+      "loss": 0.5637,
+      "step": 14131
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.423330591292705e-06,
+      "loss": 0.5774,
+      "step": 14132
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.421986648385582e-06,
+      "loss": 0.5872,
+      "step": 14133
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.420642779587984e-06,
+      "loss": 0.4836,
+      "step": 14134
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.419298984927746e-06,
+      "loss": 0.6196,
+      "step": 14135
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.417955264432698e-06,
+      "loss": 0.5435,
+      "step": 14136
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.416611618130671e-06,
+      "loss": 0.5227,
+      "step": 14137
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.415268046049503e-06,
+      "loss": 0.5859,
+      "step": 14138
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.413924548217017e-06,
+      "loss": 0.5562,
+      "step": 14139
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.412581124661035e-06,
+      "loss": 0.6077,
+      "step": 14140
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.411237775409392e-06,
+      "loss": 0.7036,
+      "step": 14141
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.409894500489906e-06,
+      "loss": 0.6362,
+      "step": 14142
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.408551299930403e-06,
+      "loss": 0.6328,
+      "step": 14143
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.407208173758701e-06,
+      "loss": 0.5464,
+      "step": 14144
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4058651220026235e-06,
+      "loss": 0.5469,
+      "step": 14145
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.404522144689984e-06,
+      "loss": 0.5642,
+      "step": 14146
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4031792418486e-06,
+      "loss": 0.499,
+      "step": 14147
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.40183641350629e-06,
+      "loss": 0.554,
+      "step": 14148
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.400493659690864e-06,
+      "loss": 0.553,
+      "step": 14149
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.399150980430131e-06,
+      "loss": 0.5315,
+      "step": 14150
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.397808375751911e-06,
+      "loss": 0.5593,
+      "step": 14151
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.3964658456840005e-06,
+      "loss": 0.5601,
+      "step": 14152
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.395123390254216e-06,
+      "loss": 0.5403,
+      "step": 14153
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.393781009490361e-06,
+      "loss": 0.6021,
+      "step": 14154
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.392438703420233e-06,
+      "loss": 0.5962,
+      "step": 14155
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3910964720716425e-06,
+      "loss": 0.5432,
+      "step": 14156
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.389754315472385e-06,
+      "loss": 0.5854,
+      "step": 14157
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.388412233650264e-06,
+      "loss": 0.6011,
+      "step": 14158
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.387070226633076e-06,
+      "loss": 0.6294,
+      "step": 14159
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.385728294448617e-06,
+      "loss": 0.563,
+      "step": 14160
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.38438643712468e-06,
+      "loss": 0.5701,
+      "step": 14161
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.383044654689061e-06,
+      "loss": 0.5564,
+      "step": 14162
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.381702947169548e-06,
+      "loss": 0.5225,
+      "step": 14163
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.380361314593938e-06,
+      "loss": 0.5542,
+      "step": 14164
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.379019756990012e-06,
+      "loss": 0.6035,
+      "step": 14165
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.377678274385562e-06,
+      "loss": 0.5051,
+      "step": 14166
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.376336866808369e-06,
+      "loss": 0.519,
+      "step": 14167
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3749955342862235e-06,
+      "loss": 0.5781,
+      "step": 14168
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3736542768469036e-06,
+      "loss": 0.5586,
+      "step": 14169
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.372313094518185e-06,
+      "loss": 0.5791,
+      "step": 14170
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.370971987327855e-06,
+      "loss": 0.5632,
+      "step": 14171
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.36963095530369e-06,
+      "loss": 0.6196,
+      "step": 14172
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.368289998473462e-06,
+      "loss": 0.5918,
+      "step": 14173
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.366949116864947e-06,
+      "loss": 0.5457,
+      "step": 14174
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.365608310505922e-06,
+      "loss": 0.6091,
+      "step": 14175
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.364267579424153e-06,
+      "loss": 0.5425,
+      "step": 14176
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.362926923647414e-06,
+      "loss": 0.5615,
+      "step": 14177
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3615863432034695e-06,
+      "loss": 0.5291,
+      "step": 14178
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.360245838120092e-06,
+      "loss": 0.6211,
+      "step": 14179
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.358905408425038e-06,
+      "loss": 0.5598,
+      "step": 14180
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.357565054146082e-06,
+      "loss": 0.613,
+      "step": 14181
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3562247753109754e-06,
+      "loss": 0.5745,
+      "step": 14182
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.354884571947486e-06,
+      "loss": 0.4937,
+      "step": 14183
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.35354444408337e-06,
+      "loss": 0.6045,
+      "step": 14184
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.352204391746384e-06,
+      "loss": 0.5935,
+      "step": 14185
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.350864414964285e-06,
+      "loss": 0.5837,
+      "step": 14186
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.349524513764824e-06,
+      "loss": 0.5894,
+      "step": 14187
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.348184688175763e-06,
+      "loss": 0.575,
+      "step": 14188
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.346844938224841e-06,
+      "loss": 0.6475,
+      "step": 14189
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3455052639398176e-06,
+      "loss": 0.5605,
+      "step": 14190
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3441656653484325e-06,
+      "loss": 0.6328,
+      "step": 14191
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.342826142478441e-06,
+      "loss": 0.531,
+      "step": 14192
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.341486695357578e-06,
+      "loss": 0.5205,
+      "step": 14193
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.340147324013597e-06,
+      "loss": 0.5649,
+      "step": 14194
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.338808028474229e-06,
+      "loss": 0.5686,
+      "step": 14195
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.337468808767224e-06,
+      "loss": 0.5386,
+      "step": 14196
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.336129664920313e-06,
+      "loss": 0.5369,
+      "step": 14197
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.334790596961238e-06,
+      "loss": 0.5715,
+      "step": 14198
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.333451604917734e-06,
+      "loss": 0.5483,
+      "step": 14199
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.33211268881753e-06,
+      "loss": 0.6658,
+      "step": 14200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3307738486883655e-06,
+      "loss": 0.4924,
+      "step": 14201
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.329435084557964e-06,
+      "loss": 0.6287,
+      "step": 14202
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.328096396454062e-06,
+      "loss": 0.5238,
+      "step": 14203
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.326757784404378e-06,
+      "loss": 0.5759,
+      "step": 14204
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.325419248436648e-06,
+      "loss": 0.499,
+      "step": 14205
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.324080788578586e-06,
+      "loss": 0.575,
+      "step": 14206
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.322742404857926e-06,
+      "loss": 0.6218,
+      "step": 14207
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.321404097302379e-06,
+      "loss": 0.5654,
+      "step": 14208
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.320065865939673e-06,
+      "loss": 0.6003,
+      "step": 14209
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.318727710797519e-06,
+      "loss": 0.4836,
+      "step": 14210
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.317389631903638e-06,
+      "loss": 0.5796,
+      "step": 14211
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.316051629285744e-06,
+      "loss": 0.5627,
+      "step": 14212
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.31471370297155e-06,
+      "loss": 0.5784,
+      "step": 14213
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.31337585298877e-06,
+      "loss": 0.5916,
+      "step": 14214
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.312038079365108e-06,
+      "loss": 0.5388,
+      "step": 14215
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.31070038212828e-06,
+      "loss": 0.5935,
+      "step": 14216
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3093627613059864e-06,
+      "loss": 0.6516,
+      "step": 14217
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3080252169259396e-06,
+      "loss": 0.6196,
+      "step": 14218
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.306687749015837e-06,
+      "loss": 0.5566,
+      "step": 14219
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.305350357603386e-06,
+      "loss": 0.5996,
+      "step": 14220
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3040130427162805e-06,
+      "loss": 0.592,
+      "step": 14221
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.302675804382229e-06,
+      "loss": 0.5938,
+      "step": 14222
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.301338642628921e-06,
+      "loss": 0.5713,
+      "step": 14223
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.300001557484054e-06,
+      "loss": 0.5857,
+      "step": 14224
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.298664548975327e-06,
+      "loss": 0.5393,
+      "step": 14225
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.297327617130429e-06,
+      "loss": 0.562,
+      "step": 14226
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.29599076197705e-06,
+      "loss": 0.6316,
+      "step": 14227
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.294653983542884e-06,
+      "loss": 0.6558,
+      "step": 14228
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2933172818556175e-06,
+      "loss": 0.5408,
+      "step": 14229
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.291980656942932e-06,
+      "loss": 0.5674,
+      "step": 14230
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.29064410883252e-06,
+      "loss": 0.5188,
+      "step": 14231
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.289307637552056e-06,
+      "loss": 0.6035,
+      "step": 14232
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2879712431292316e-06,
+      "loss": 0.5867,
+      "step": 14233
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.286634925591716e-06,
+      "loss": 0.6228,
+      "step": 14234
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.285298684967196e-06,
+      "loss": 0.5415,
+      "step": 14235
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.283962521283346e-06,
+      "loss": 0.5547,
+      "step": 14236
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.282626434567841e-06,
+      "loss": 0.5574,
+      "step": 14237
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.281290424848354e-06,
+      "loss": 0.5991,
+      "step": 14238
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.279954492152559e-06,
+      "loss": 0.595,
+      "step": 14239
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.278618636508123e-06,
+      "loss": 0.676,
+      "step": 14240
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2772828579427206e-06,
+      "loss": 0.5859,
+      "step": 14241
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2759471564840114e-06,
+      "loss": 0.5471,
+      "step": 14242
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.274611532159669e-06,
+      "loss": 0.6084,
+      "step": 14243
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.273275984997356e-06,
+      "loss": 0.5488,
+      "step": 14244
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.271940515024726e-06,
+      "loss": 0.5295,
+      "step": 14245
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.270605122269451e-06,
+      "loss": 0.5549,
+      "step": 14246
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.269269806759183e-06,
+      "loss": 0.5413,
+      "step": 14247
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.267934568521583e-06,
+      "loss": 0.6013,
+      "step": 14248
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.266599407584308e-06,
+      "loss": 0.5706,
+      "step": 14249
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.265264323975011e-06,
+      "loss": 0.5925,
+      "step": 14250
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.263929317721342e-06,
+      "loss": 0.5474,
+      "step": 14251
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.262594388850961e-06,
+      "loss": 0.6213,
+      "step": 14252
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.261259537391506e-06,
+      "loss": 0.5525,
+      "step": 14253
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.259924763370637e-06,
+      "loss": 0.5967,
+      "step": 14254
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.258590066815991e-06,
+      "loss": 0.6108,
+      "step": 14255
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.257255447755217e-06,
+      "loss": 0.5894,
+      "step": 14256
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.255920906215957e-06,
+      "loss": 0.5872,
+      "step": 14257
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.254586442225856e-06,
+      "loss": 0.5286,
+      "step": 14258
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.253252055812551e-06,
+      "loss": 0.5725,
+      "step": 14259
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.251917747003677e-06,
+      "loss": 0.5327,
+      "step": 14260
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.250583515826878e-06,
+      "loss": 0.5413,
+      "step": 14261
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.249249362309785e-06,
+      "loss": 0.6077,
+      "step": 14262
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.247915286480033e-06,
+      "loss": 0.5701,
+      "step": 14263
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.24658128836525e-06,
+      "loss": 0.6072,
+      "step": 14264
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.245247367993076e-06,
+      "loss": 0.574,
+      "step": 14265
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.243913525391129e-06,
+      "loss": 0.5286,
+      "step": 14266
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.242579760587044e-06,
+      "loss": 0.5117,
+      "step": 14267
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.241246073608441e-06,
+      "loss": 0.53,
+      "step": 14268
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.23991246448295e-06,
+      "loss": 0.5198,
+      "step": 14269
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.238578933238186e-06,
+      "loss": 0.5781,
+      "step": 14270
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.237245479901776e-06,
+      "loss": 0.5796,
+      "step": 14271
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.235912104501336e-06,
+      "loss": 0.5881,
+      "step": 14272
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.234578807064487e-06,
+      "loss": 0.6489,
+      "step": 14273
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.233245587618839e-06,
+      "loss": 0.6084,
+      "step": 14274
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.231912446192008e-06,
+      "loss": 0.5742,
+      "step": 14275
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.230579382811611e-06,
+      "loss": 0.5222,
+      "step": 14276
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.229246397505253e-06,
+      "loss": 0.6541,
+      "step": 14277
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.22791349030055e-06,
+      "loss": 0.5674,
+      "step": 14278
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.226580661225103e-06,
+      "loss": 0.5642,
+      "step": 14279
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2252479103065265e-06,
+      "loss": 0.5911,
+      "step": 14280
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2239152375724155e-06,
+      "loss": 0.5728,
+      "step": 14281
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.222582643050381e-06,
+      "loss": 0.6436,
+      "step": 14282
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.221250126768017e-06,
+      "loss": 0.5898,
+      "step": 14283
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.21991768875293e-06,
+      "loss": 0.5334,
+      "step": 14284
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.218585329032714e-06,
+      "loss": 0.6309,
+      "step": 14285
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.217253047634968e-06,
+      "loss": 0.5603,
+      "step": 14286
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.215920844587282e-06,
+      "loss": 0.5991,
+      "step": 14287
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.214588719917256e-06,
+      "loss": 0.5295,
+      "step": 14288
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.213256673652479e-06,
+      "loss": 0.541,
+      "step": 14289
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.211924705820538e-06,
+      "loss": 0.4993,
+      "step": 14290
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.210592816449026e-06,
+      "loss": 0.5073,
+      "step": 14291
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.209261005565525e-06,
+      "loss": 0.6016,
+      "step": 14292
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.207929273197625e-06,
+      "loss": 0.6113,
+      "step": 14293
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.206597619372904e-06,
+      "loss": 0.5884,
+      "step": 14294
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.2052660441189494e-06,
+      "loss": 0.5376,
+      "step": 14295
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.203934547463337e-06,
+      "loss": 0.5688,
+      "step": 14296
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.202603129433651e-06,
+      "loss": 0.5618,
+      "step": 14297
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.201271790057459e-06,
+      "loss": 0.5999,
+      "step": 14298
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.199940529362347e-06,
+      "loss": 0.562,
+      "step": 14299
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.198609347375879e-06,
+      "loss": 0.5803,
+      "step": 14300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1972782441256355e-06,
+      "loss": 0.5608,
+      "step": 14301
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.195947219639181e-06,
+      "loss": 0.5623,
+      "step": 14302
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.194616273944088e-06,
+      "loss": 0.5732,
+      "step": 14303
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.193285407067924e-06,
+      "loss": 0.5432,
+      "step": 14304
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.191954619038249e-06,
+      "loss": 0.6567,
+      "step": 14305
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.190623909882634e-06,
+      "loss": 0.5044,
+      "step": 14306
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.189293279628634e-06,
+      "loss": 0.6172,
+      "step": 14307
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.187962728303818e-06,
+      "loss": 0.5793,
+      "step": 14308
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.186632255935736e-06,
+      "loss": 0.5159,
+      "step": 14309
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.185301862551955e-06,
+      "loss": 0.5864,
+      "step": 14310
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.183971548180021e-06,
+      "loss": 0.5215,
+      "step": 14311
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.182641312847495e-06,
+      "loss": 0.51,
+      "step": 14312
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1813111565819275e-06,
+      "loss": 0.6396,
+      "step": 14313
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.179981079410868e-06,
+      "loss": 0.5903,
+      "step": 14314
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.178651081361868e-06,
+      "loss": 0.5508,
+      "step": 14315
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.177321162462474e-06,
+      "loss": 0.5356,
+      "step": 14316
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.17599132274023e-06,
+      "loss": 0.5273,
+      "step": 14317
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1746615622226854e-06,
+      "loss": 0.5757,
+      "step": 14318
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.173331880937381e-06,
+      "loss": 0.6072,
+      "step": 14319
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.172002278911851e-06,
+      "loss": 0.5906,
+      "step": 14320
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.170672756173643e-06,
+      "loss": 0.5298,
+      "step": 14321
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.16934331275029e-06,
+      "loss": 0.6304,
+      "step": 14322
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.168013948669333e-06,
+      "loss": 0.615,
+      "step": 14323
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.166684663958299e-06,
+      "loss": 0.5613,
+      "step": 14324
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.165355458644727e-06,
+      "loss": 0.5432,
+      "step": 14325
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.164026332756148e-06,
+      "loss": 0.6436,
+      "step": 14326
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.162697286320087e-06,
+      "loss": 0.5867,
+      "step": 14327
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.161368319364074e-06,
+      "loss": 0.5227,
+      "step": 14328
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.16003943191564e-06,
+      "loss": 0.5579,
+      "step": 14329
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1587106240023e-06,
+      "loss": 0.5999,
+      "step": 14330
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1573818956515875e-06,
+      "loss": 0.481,
+      "step": 14331
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.156053246891014e-06,
+      "loss": 0.626,
+      "step": 14332
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.154724677748107e-06,
+      "loss": 0.583,
+      "step": 14333
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1533961882503825e-06,
+      "loss": 0.5452,
+      "step": 14334
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.15206777842535e-06,
+      "loss": 0.562,
+      "step": 14335
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.150739448300533e-06,
+      "loss": 0.6465,
+      "step": 14336
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.149411197903438e-06,
+      "loss": 0.6548,
+      "step": 14337
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.148083027261582e-06,
+      "loss": 0.6196,
+      "step": 14338
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.146754936402471e-06,
+      "loss": 0.542,
+      "step": 14339
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.145426925353615e-06,
+      "loss": 0.6091,
+      "step": 14340
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.144098994142516e-06,
+      "loss": 0.5371,
+      "step": 14341
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.142771142796688e-06,
+      "loss": 0.5974,
+      "step": 14342
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.141443371343623e-06,
+      "loss": 0.5879,
+      "step": 14343
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.140115679810833e-06,
+      "loss": 0.627,
+      "step": 14344
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.138788068225807e-06,
+      "loss": 0.5752,
+      "step": 14345
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.137460536616055e-06,
+      "loss": 0.551,
+      "step": 14346
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.136133085009066e-06,
+      "loss": 0.4771,
+      "step": 14347
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1348057134323315e-06,
+      "loss": 0.51,
+      "step": 14348
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.133478421913352e-06,
+      "loss": 0.4646,
+      "step": 14349
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.132151210479616e-06,
+      "loss": 0.574,
+      "step": 14350
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.130824079158612e-06,
+      "loss": 0.5837,
+      "step": 14351
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.129497027977829e-06,
+      "loss": 0.5657,
+      "step": 14352
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.128170056964758e-06,
+      "loss": 0.6221,
+      "step": 14353
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.126843166146875e-06,
+      "loss": 0.5352,
+      "step": 14354
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.125516355551673e-06,
+      "loss": 0.6069,
+      "step": 14355
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.124189625206625e-06,
+      "loss": 0.5972,
+      "step": 14356
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.122862975139219e-06,
+      "loss": 0.5676,
+      "step": 14357
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.121536405376924e-06,
+      "loss": 0.6135,
+      "step": 14358
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.120209915947226e-06,
+      "loss": 0.5369,
+      "step": 14359
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.118883506877591e-06,
+      "loss": 0.5935,
+      "step": 14360
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1175571781955e-06,
+      "loss": 0.5337,
+      "step": 14361
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.116230929928419e-06,
+      "loss": 0.5039,
+      "step": 14362
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1149047621038215e-06,
+      "loss": 0.585,
+      "step": 14363
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.113578674749173e-06,
+      "loss": 0.5532,
+      "step": 14364
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1122526678919395e-06,
+      "loss": 0.5376,
+      "step": 14365
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.11092674155959e-06,
+      "loss": 0.6169,
+      "step": 14366
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.109600895779584e-06,
+      "loss": 0.5278,
+      "step": 14367
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1082751305793864e-06,
+      "loss": 0.6355,
+      "step": 14368
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.106949445986452e-06,
+      "loss": 0.5198,
+      "step": 14369
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.105623842028246e-06,
+      "loss": 0.5562,
+      "step": 14370
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.104298318732218e-06,
+      "loss": 0.6357,
+      "step": 14371
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.102972876125828e-06,
+      "loss": 0.5103,
+      "step": 14372
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.101647514236524e-06,
+      "loss": 0.5745,
+      "step": 14373
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1003222330917645e-06,
+      "loss": 0.5046,
+      "step": 14374
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.098997032718992e-06,
+      "loss": 0.5984,
+      "step": 14375
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.097671913145661e-06,
+      "loss": 0.6245,
+      "step": 14376
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.096346874399214e-06,
+      "loss": 0.5688,
+      "step": 14377
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.0950219165070945e-06,
+      "loss": 0.5398,
+      "step": 14378
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.093697039496752e-06,
+      "loss": 0.5554,
+      "step": 14379
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.092372243395621e-06,
+      "loss": 0.5134,
+      "step": 14380
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.091047528231148e-06,
+      "loss": 0.6055,
+      "step": 14381
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.089722894030763e-06,
+      "loss": 0.5496,
+      "step": 14382
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.088398340821913e-06,
+      "loss": 0.5874,
+      "step": 14383
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.087073868632021e-06,
+      "loss": 0.5396,
+      "step": 14384
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0857494774885295e-06,
+      "loss": 0.5662,
+      "step": 14385
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.084425167418864e-06,
+      "loss": 0.5583,
+      "step": 14386
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.083100938450459e-06,
+      "loss": 0.5801,
+      "step": 14387
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.081776790610737e-06,
+      "loss": 0.5562,
+      "step": 14388
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.08045272392713e-06,
+      "loss": 0.6111,
+      "step": 14389
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.07912873842706e-06,
+      "loss": 0.623,
+      "step": 14390
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.07780483413795e-06,
+      "loss": 0.5066,
+      "step": 14391
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.076481011087223e-06,
+      "loss": 0.6113,
+      "step": 14392
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.075157269302294e-06,
+      "loss": 0.592,
+      "step": 14393
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.073833608810589e-06,
+      "loss": 0.584,
+      "step": 14394
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.072510029639513e-06,
+      "loss": 0.5769,
+      "step": 14395
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.071186531816493e-06,
+      "loss": 0.5874,
+      "step": 14396
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.069863115368932e-06,
+      "loss": 0.6641,
+      "step": 14397
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.06853978032425e-06,
+      "loss": 0.5977,
+      "step": 14398
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.067216526709847e-06,
+      "loss": 0.5825,
+      "step": 14399
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.06589335455314e-06,
+      "loss": 0.5127,
+      "step": 14400
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.064570263881526e-06,
+      "loss": 0.5679,
+      "step": 14401
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.063247254722418e-06,
+      "loss": 0.6145,
+      "step": 14402
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.061924327103213e-06,
+      "loss": 0.5715,
+      "step": 14403
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.060601481051315e-06,
+      "loss": 0.5737,
+      "step": 14404
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.059278716594122e-06,
+      "loss": 0.6189,
+      "step": 14405
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.057956033759033e-06,
+      "loss": 0.5681,
+      "step": 14406
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.056633432573446e-06,
+      "loss": 0.5476,
+      "step": 14407
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.055310913064749e-06,
+      "loss": 0.5786,
+      "step": 14408
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.053988475260343e-06,
+      "loss": 0.6216,
+      "step": 14409
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0526661191876094e-06,
+      "loss": 0.5828,
+      "step": 14410
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.051343844873946e-06,
+      "loss": 0.6152,
+      "step": 14411
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.050021652346734e-06,
+      "loss": 0.5962,
+      "step": 14412
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.048699541633365e-06,
+      "loss": 0.5457,
+      "step": 14413
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.047377512761216e-06,
+      "loss": 0.5173,
+      "step": 14414
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.046055565757678e-06,
+      "loss": 0.6284,
+      "step": 14415
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.044733700650126e-06,
+      "loss": 0.5649,
+      "step": 14416
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.043411917465941e-06,
+      "loss": 0.6116,
+      "step": 14417
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.042090216232498e-06,
+      "loss": 0.4917,
+      "step": 14418
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.040768596977179e-06,
+      "loss": 0.5522,
+      "step": 14419
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0394470597273504e-06,
+      "loss": 0.5967,
+      "step": 14420
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.038125604510392e-06,
+      "loss": 0.624,
+      "step": 14421
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.036804231353669e-06,
+      "loss": 0.6543,
+      "step": 14422
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.035482940284549e-06,
+      "loss": 0.5449,
+      "step": 14423
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0341617313304055e-06,
+      "loss": 0.5942,
+      "step": 14424
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.032840604518596e-06,
+      "loss": 0.5891,
+      "step": 14425
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.031519559876491e-06,
+      "loss": 0.5586,
+      "step": 14426
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.03019859743145e-06,
+      "loss": 0.5569,
+      "step": 14427
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0288777172108345e-06,
+      "loss": 0.6243,
+      "step": 14428
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0275569192420005e-06,
+      "loss": 0.4905,
+      "step": 14429
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0262362035523105e-06,
+      "loss": 0.5476,
+      "step": 14430
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.024915570169112e-06,
+      "loss": 0.5393,
+      "step": 14431
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.023595019119768e-06,
+      "loss": 0.5664,
+      "step": 14432
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.022274550431619e-06,
+      "loss": 0.6196,
+      "step": 14433
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.020954164132027e-06,
+      "loss": 0.6309,
+      "step": 14434
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.019633860248331e-06,
+      "loss": 0.5681,
+      "step": 14435
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.018313638807884e-06,
+      "loss": 0.6641,
+      "step": 14436
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.016993499838031e-06,
+      "loss": 0.5239,
+      "step": 14437
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.015673443366108e-06,
+      "loss": 0.5361,
+      "step": 14438
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.014353469419463e-06,
+      "loss": 0.5537,
+      "step": 14439
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.013033578025436e-06,
+      "loss": 0.6218,
+      "step": 14440
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0117137692113625e-06,
+      "loss": 0.6016,
+      "step": 14441
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.01039404300458e-06,
+      "loss": 0.5881,
+      "step": 14442
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.009074399432425e-06,
+      "loss": 0.5701,
+      "step": 14443
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.007754838522227e-06,
+      "loss": 0.5852,
+      "step": 14444
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.006435360301324e-06,
+      "loss": 0.5315,
+      "step": 14445
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.005115964797036e-06,
+      "loss": 0.5376,
+      "step": 14446
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0037966520367e-06,
+      "loss": 0.5312,
+      "step": 14447
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0024774220476365e-06,
+      "loss": 0.552,
+      "step": 14448
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.001158274857175e-06,
+      "loss": 0.5156,
+      "step": 14449
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.99983921049263e-06,
+      "loss": 0.5701,
+      "step": 14450
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.998520228981333e-06,
+      "loss": 0.4897,
+      "step": 14451
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.997201330350597e-06,
+      "loss": 0.5856,
+      "step": 14452
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.99588251462774e-06,
+      "loss": 0.5945,
+      "step": 14453
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.99456378184008e-06,
+      "loss": 0.5068,
+      "step": 14454
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.993245132014925e-06,
+      "loss": 0.5894,
+      "step": 14455
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9919265651796e-06,
+      "loss": 0.4725,
+      "step": 14456
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.990608081361402e-06,
+      "loss": 0.6138,
+      "step": 14457
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.989289680587651e-06,
+      "loss": 0.5989,
+      "step": 14458
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.987971362885648e-06,
+      "loss": 0.5884,
+      "step": 14459
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.986653128282702e-06,
+      "loss": 0.5491,
+      "step": 14460
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.985334976806112e-06,
+      "loss": 0.5396,
+      "step": 14461
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.984016908483187e-06,
+      "loss": 0.5427,
+      "step": 14462
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.98269892334122e-06,
+      "loss": 0.5073,
+      "step": 14463
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.981381021407518e-06,
+      "loss": 0.5388,
+      "step": 14464
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.98006320270937e-06,
+      "loss": 0.6155,
+      "step": 14465
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.978745467274076e-06,
+      "loss": 0.5103,
+      "step": 14466
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.977427815128932e-06,
+      "loss": 0.5183,
+      "step": 14467
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.976110246301222e-06,
+      "loss": 0.6035,
+      "step": 14468
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.974792760818244e-06,
+      "loss": 0.5662,
+      "step": 14469
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9734753587072795e-06,
+      "loss": 0.6113,
+      "step": 14470
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.972158039995622e-06,
+      "loss": 0.6426,
+      "step": 14471
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9708408047105495e-06,
+      "loss": 0.5891,
+      "step": 14472
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.969523652879353e-06,
+      "loss": 0.5527,
+      "step": 14473
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.968206584529306e-06,
+      "loss": 0.5732,
+      "step": 14474
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9668895996876955e-06,
+      "loss": 0.552,
+      "step": 14475
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.965572698381791e-06,
+      "loss": 0.6411,
+      "step": 14476
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.964255880638879e-06,
+      "loss": 0.5247,
+      "step": 14477
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9629391464862245e-06,
+      "loss": 0.5706,
+      "step": 14478
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.961622495951105e-06,
+      "loss": 0.5671,
+      "step": 14479
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9603059290607925e-06,
+      "loss": 0.5696,
+      "step": 14480
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.958989445842555e-06,
+      "loss": 0.4851,
+      "step": 14481
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9576730463236636e-06,
+      "loss": 0.5112,
+      "step": 14482
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.956356730531375e-06,
+      "loss": 0.4888,
+      "step": 14483
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.955040498492965e-06,
+      "loss": 0.5818,
+      "step": 14484
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.953724350235684e-06,
+      "loss": 0.5024,
+      "step": 14485
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.952408285786805e-06,
+      "loss": 0.5239,
+      "step": 14486
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.951092305173576e-06,
+      "loss": 0.5403,
+      "step": 14487
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9497764084232625e-06,
+      "loss": 0.5413,
+      "step": 14488
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.948460595563113e-06,
+      "loss": 0.5627,
+      "step": 14489
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.947144866620389e-06,
+      "loss": 0.499,
+      "step": 14490
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9458292216223345e-06,
+      "loss": 0.5159,
+      "step": 14491
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.944513660596205e-06,
+      "loss": 0.5261,
+      "step": 14492
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.943198183569248e-06,
+      "loss": 0.5203,
+      "step": 14493
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.941882790568709e-06,
+      "loss": 0.6165,
+      "step": 14494
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.940567481621833e-06,
+      "loss": 0.5583,
+      "step": 14495
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.939252256755869e-06,
+      "loss": 0.6194,
+      "step": 14496
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9379371159980526e-06,
+      "loss": 0.5483,
+      "step": 14497
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.936622059375621e-06,
+      "loss": 0.6086,
+      "step": 14498
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.93530708691582e-06,
+      "loss": 0.5522,
+      "step": 14499
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.933992198645879e-06,
+      "loss": 0.5935,
+      "step": 14500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.932677394593038e-06,
+      "loss": 0.4946,
+      "step": 14501
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.931362674784526e-06,
+      "loss": 0.5784,
+      "step": 14502
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.930048039247575e-06,
+      "loss": 0.5181,
+      "step": 14503
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.928733488009417e-06,
+      "loss": 0.5781,
+      "step": 14504
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.927419021097276e-06,
+      "loss": 0.5327,
+      "step": 14505
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.926104638538379e-06,
+      "loss": 0.5923,
+      "step": 14506
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.924790340359954e-06,
+      "loss": 0.5435,
+      "step": 14507
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.923476126589216e-06,
+      "loss": 0.5278,
+      "step": 14508
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.922161997253393e-06,
+      "loss": 0.6033,
+      "step": 14509
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.920847952379696e-06,
+      "loss": 0.5447,
+      "step": 14510
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.919533991995352e-06,
+      "loss": 0.5435,
+      "step": 14511
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.918220116127571e-06,
+      "loss": 0.6157,
+      "step": 14512
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.916906324803562e-06,
+      "loss": 0.5771,
+      "step": 14513
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.915592618050545e-06,
+      "loss": 0.5725,
+      "step": 14514
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9142789958957225e-06,
+      "loss": 0.5713,
+      "step": 14515
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.91296545836631e-06,
+      "loss": 0.5549,
+      "step": 14516
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.91165200548951e-06,
+      "loss": 0.5486,
+      "step": 14517
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.91033863729253e-06,
+      "loss": 0.5884,
+      "step": 14518
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.909025353802568e-06,
+      "loss": 0.5574,
+      "step": 14519
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.907712155046833e-06,
+      "loss": 0.5872,
+      "step": 14520
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.906399041052517e-06,
+      "loss": 0.5935,
+      "step": 14521
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.905086011846825e-06,
+      "loss": 0.6357,
+      "step": 14522
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.903773067456946e-06,
+      "loss": 0.5237,
+      "step": 14523
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.902460207910081e-06,
+      "loss": 0.5342,
+      "step": 14524
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.901147433233417e-06,
+      "loss": 0.5979,
+      "step": 14525
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.89983474345415e-06,
+      "loss": 0.5542,
+      "step": 14526
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.898522138599465e-06,
+      "loss": 0.5884,
+      "step": 14527
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.897209618696547e-06,
+      "loss": 0.5442,
+      "step": 14528
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.895897183772588e-06,
+      "loss": 0.5737,
+      "step": 14529
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.894584833854769e-06,
+      "loss": 0.5732,
+      "step": 14530
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8932725689702716e-06,
+      "loss": 0.5364,
+      "step": 14531
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.891960389146274e-06,
+      "loss": 0.5376,
+      "step": 14532
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.890648294409961e-06,
+      "loss": 0.6218,
+      "step": 14533
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.889336284788502e-06,
+      "loss": 0.563,
+      "step": 14534
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.888024360309078e-06,
+      "loss": 0.6379,
+      "step": 14535
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.886712520998856e-06,
+      "loss": 0.5222,
+      "step": 14536
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.885400766885016e-06,
+      "loss": 0.5862,
+      "step": 14537
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.884089097994717e-06,
+      "loss": 0.5801,
+      "step": 14538
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.882777514355138e-06,
+      "loss": 0.5693,
+      "step": 14539
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.881466015993435e-06,
+      "loss": 0.4897,
+      "step": 14540
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.880154602936781e-06,
+      "loss": 0.5918,
+      "step": 14541
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.878843275212333e-06,
+      "loss": 0.6169,
+      "step": 14542
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.877532032847252e-06,
+      "loss": 0.5627,
+      "step": 14543
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.876220875868701e-06,
+      "loss": 0.5469,
+      "step": 14544
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.874909804303832e-06,
+      "loss": 0.5635,
+      "step": 14545
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.873598818179809e-06,
+      "loss": 0.5996,
+      "step": 14546
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.872287917523774e-06,
+      "loss": 0.6042,
+      "step": 14547
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.87097710236289e-06,
+      "loss": 0.5635,
+      "step": 14548
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.869666372724299e-06,
+      "loss": 0.5164,
+      "step": 14549
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.868355728635155e-06,
+      "loss": 0.5728,
+      "step": 14550
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8670451701226005e-06,
+      "loss": 0.6023,
+      "step": 14551
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8657346972137864e-06,
+      "loss": 0.5593,
+      "step": 14552
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.864424309935848e-06,
+      "loss": 0.6162,
+      "step": 14553
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.863114008315933e-06,
+      "loss": 0.5881,
+      "step": 14554
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.861803792381176e-06,
+      "loss": 0.5273,
+      "step": 14555
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.86049366215872e-06,
+      "loss": 0.5066,
+      "step": 14556
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.859183617675699e-06,
+      "loss": 0.6272,
+      "step": 14557
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.857873658959245e-06,
+      "loss": 0.5488,
+      "step": 14558
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.856563786036493e-06,
+      "loss": 0.5698,
+      "step": 14559
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.855253998934571e-06,
+      "loss": 0.5312,
+      "step": 14560
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.853944297680614e-06,
+      "loss": 0.637,
+      "step": 14561
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.852634682301741e-06,
+      "loss": 0.6145,
+      "step": 14562
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8513251528250854e-06,
+      "loss": 0.5427,
+      "step": 14563
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.850015709277763e-06,
+      "loss": 0.4993,
+      "step": 14564
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.848706351686903e-06,
+      "loss": 0.5933,
+      "step": 14565
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.847397080079617e-06,
+      "loss": 0.551,
+      "step": 14566
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.846087894483034e-06,
+      "loss": 0.6152,
+      "step": 14567
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.84477879492426e-06,
+      "loss": 0.4619,
+      "step": 14568
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.843469781430417e-06,
+      "loss": 0.5847,
+      "step": 14569
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8421608540286156e-06,
+      "loss": 0.5542,
+      "step": 14570
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8408520127459635e-06,
+      "loss": 0.5388,
+      "step": 14571
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.839543257609578e-06,
+      "loss": 0.5234,
+      "step": 14572
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.838234588646556e-06,
+      "loss": 0.5605,
+      "step": 14573
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.836926005884014e-06,
+      "loss": 0.5879,
+      "step": 14574
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.835617509349047e-06,
+      "loss": 0.4912,
+      "step": 14575
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.834309099068767e-06,
+      "loss": 0.5789,
+      "step": 14576
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.833000775070262e-06,
+      "loss": 0.4893,
+      "step": 14577
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.831692537380642e-06,
+      "loss": 0.5515,
+      "step": 14578
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.830384386026996e-06,
+      "loss": 0.5972,
+      "step": 14579
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.829076321036426e-06,
+      "loss": 0.606,
+      "step": 14580
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.827768342436016e-06,
+      "loss": 0.4905,
+      "step": 14581
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8264604502528685e-06,
+      "loss": 0.53,
+      "step": 14582
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.825152644514064e-06,
+      "loss": 0.5305,
+      "step": 14583
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.823844925246698e-06,
+      "loss": 0.5925,
+      "step": 14584
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.822537292477848e-06,
+      "loss": 0.5732,
+      "step": 14585
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8212297462346065e-06,
+      "loss": 0.6143,
+      "step": 14586
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.819922286544049e-06,
+      "loss": 0.6155,
+      "step": 14587
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.818614913433261e-06,
+      "loss": 0.5215,
+      "step": 14588
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.817307626929322e-06,
+      "loss": 0.572,
+      "step": 14589
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.816000427059304e-06,
+      "loss": 0.5977,
+      "step": 14590
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.81469331385029e-06,
+      "loss": 0.5491,
+      "step": 14591
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.813386287329346e-06,
+      "loss": 0.6199,
+      "step": 14592
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.812079347523549e-06,
+      "loss": 0.5164,
+      "step": 14593
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.810772494459964e-06,
+      "loss": 0.5508,
+      "step": 14594
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.809465728165666e-06,
+      "loss": 0.5649,
+      "step": 14595
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.808159048667714e-06,
+      "loss": 0.6167,
+      "step": 14596
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8068524559931794e-06,
+      "loss": 0.5349,
+      "step": 14597
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.805545950169118e-06,
+      "loss": 0.5542,
+      "step": 14598
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8042395312225975e-06,
+      "loss": 0.603,
+      "step": 14599
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.802933199180676e-06,
+      "loss": 0.5415,
+      "step": 14600
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.801626954070403e-06,
+      "loss": 0.615,
+      "step": 14601
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.800320795918845e-06,
+      "loss": 0.627,
+      "step": 14602
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.799014724753045e-06,
+      "loss": 0.5066,
+      "step": 14603
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.7977087406000664e-06,
+      "loss": 0.6145,
+      "step": 14604
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.7964028434869494e-06,
+      "loss": 0.5808,
+      "step": 14605
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.79509703344075e-06,
+      "loss": 0.5142,
+      "step": 14606
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.793791310488507e-06,
+      "loss": 0.5317,
+      "step": 14607
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.792485674657273e-06,
+      "loss": 0.5959,
+      "step": 14608
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.791180125974085e-06,
+      "loss": 0.4971,
+      "step": 14609
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.789874664465991e-06,
+      "loss": 0.5964,
+      "step": 14610
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.78856929016002e-06,
+      "loss": 0.5684,
+      "step": 14611
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.787264003083217e-06,
+      "loss": 0.52,
+      "step": 14612
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.785958803262619e-06,
+      "loss": 0.6316,
+      "step": 14613
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.784653690725255e-06,
+      "loss": 0.5022,
+      "step": 14614
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.78334866549816e-06,
+      "loss": 0.5811,
+      "step": 14615
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7820437276083626e-06,
+      "loss": 0.5806,
+      "step": 14616
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7807388770828964e-06,
+      "loss": 0.4978,
+      "step": 14617
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.779434113948779e-06,
+      "loss": 0.5186,
+      "step": 14618
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.778129438233043e-06,
+      "loss": 0.5701,
+      "step": 14619
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.776824849962706e-06,
+      "loss": 0.5388,
+      "step": 14620
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.775520349164796e-06,
+      "loss": 0.5852,
+      "step": 14621
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.774215935866323e-06,
+      "loss": 0.6057,
+      "step": 14622
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.772911610094316e-06,
+      "loss": 0.5732,
+      "step": 14623
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.77160737187578e-06,
+      "loss": 0.5901,
+      "step": 14624
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.770303221237738e-06,
+      "loss": 0.6704,
+      "step": 14625
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7689991582071934e-06,
+      "loss": 0.5212,
+      "step": 14626
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.767695182811166e-06,
+      "loss": 0.522,
+      "step": 14627
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.766391295076655e-06,
+      "loss": 0.5193,
+      "step": 14628
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7650874950306754e-06,
+      "loss": 0.5735,
+      "step": 14629
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.76378378270023e-06,
+      "loss": 0.5898,
+      "step": 14630
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7624801581123146e-06,
+      "loss": 0.6558,
+      "step": 14631
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.761176621293941e-06,
+      "loss": 0.6216,
+      "step": 14632
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7598731722721e-06,
+      "loss": 0.5767,
+      "step": 14633
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.758569811073798e-06,
+      "loss": 0.5559,
+      "step": 14634
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.757266537726021e-06,
+      "loss": 0.5959,
+      "step": 14635
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.755963352255771e-06,
+      "loss": 0.6226,
+      "step": 14636
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.754660254690035e-06,
+      "loss": 0.5173,
+      "step": 14637
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.753357245055805e-06,
+      "loss": 0.5305,
+      "step": 14638
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.752054323380074e-06,
+      "loss": 0.5688,
+      "step": 14639
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.750751489689821e-06,
+      "loss": 0.5828,
+      "step": 14640
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7494487440120365e-06,
+      "loss": 0.5806,
+      "step": 14641
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.748146086373704e-06,
+      "loss": 0.5417,
+      "step": 14642
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7468435168017985e-06,
+      "loss": 0.5732,
+      "step": 14643
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7455410353233084e-06,
+      "loss": 0.5793,
+      "step": 14644
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7442386419652075e-06,
+      "loss": 0.5032,
+      "step": 14645
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.742936336754465e-06,
+      "loss": 0.5349,
+      "step": 14646
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7416341197180646e-06,
+      "loss": 0.6204,
+      "step": 14647
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.740331990882973e-06,
+      "loss": 0.5759,
+      "step": 14648
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7390299502761624e-06,
+      "loss": 0.5593,
+      "step": 14649
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.737727997924598e-06,
+      "loss": 0.5874,
+      "step": 14650
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.736426133855254e-06,
+      "loss": 0.5271,
+      "step": 14651
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.735124358095085e-06,
+      "loss": 0.6018,
+      "step": 14652
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.733822670671066e-06,
+      "loss": 0.6387,
+      "step": 14653
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.732521071610146e-06,
+      "loss": 0.5325,
+      "step": 14654
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.731219560939294e-06,
+      "loss": 0.5396,
+      "step": 14655
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.72991813868546e-06,
+      "loss": 0.5657,
+      "step": 14656
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.728616804875606e-06,
+      "loss": 0.6177,
+      "step": 14657
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.727315559536679e-06,
+      "loss": 0.5991,
+      "step": 14658
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.726014402695641e-06,
+      "loss": 0.582,
+      "step": 14659
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.724713334379435e-06,
+      "loss": 0.5215,
+      "step": 14660
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.723412354615007e-06,
+      "loss": 0.5208,
+      "step": 14661
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.722111463429306e-06,
+      "loss": 0.5344,
+      "step": 14662
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.720810660849283e-06,
+      "loss": 0.5535,
+      "step": 14663
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.719509946901871e-06,
+      "loss": 0.595,
+      "step": 14664
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.718209321614016e-06,
+      "loss": 0.6184,
+      "step": 14665
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.71690878501266e-06,
+      "loss": 0.5933,
+      "step": 14666
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.715608337124734e-06,
+      "loss": 0.4568,
+      "step": 14667
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7143079779771805e-06,
+      "loss": 0.5916,
+      "step": 14668
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.713007707596925e-06,
+      "loss": 0.5552,
+      "step": 14669
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.711707526010907e-06,
+      "loss": 0.5464,
+      "step": 14670
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.71040743324605e-06,
+      "loss": 0.5737,
+      "step": 14671
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.709107429329288e-06,
+      "loss": 0.5288,
+      "step": 14672
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.707807514287542e-06,
+      "loss": 0.5767,
+      "step": 14673
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.706507688147742e-06,
+      "loss": 0.5825,
+      "step": 14674
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7052079509368086e-06,
+      "loss": 0.5979,
+      "step": 14675
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.703908302681658e-06,
+      "loss": 0.645,
+      "step": 14676
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.702608743409215e-06,
+      "loss": 0.5884,
+      "step": 14677
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.701309273146392e-06,
+      "loss": 0.5569,
+      "step": 14678
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.700009891920108e-06,
+      "loss": 0.5732,
+      "step": 14679
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.698710599757275e-06,
+      "loss": 0.5366,
+      "step": 14680
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.697411396684807e-06,
+      "loss": 0.4893,
+      "step": 14681
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.696112282729607e-06,
+      "loss": 0.5933,
+      "step": 14682
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.69481325791859e-06,
+      "loss": 0.5747,
+      "step": 14683
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.693514322278657e-06,
+      "loss": 0.5637,
+      "step": 14684
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.692215475836718e-06,
+      "loss": 0.4954,
+      "step": 14685
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.690916718619668e-06,
+      "loss": 0.5278,
+      "step": 14686
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.689618050654414e-06,
+      "loss": 0.5845,
+      "step": 14687
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.688319471967849e-06,
+      "loss": 0.6091,
+      "step": 14688
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.687020982586873e-06,
+      "loss": 0.4883,
+      "step": 14689
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.685722582538383e-06,
+      "loss": 0.6804,
+      "step": 14690
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.684424271849267e-06,
+      "loss": 0.6033,
+      "step": 14691
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.683126050546423e-06,
+      "loss": 0.647,
+      "step": 14692
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.681827918656732e-06,
+      "loss": 0.5781,
+      "step": 14693
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.680529876207089e-06,
+      "loss": 0.509,
+      "step": 14694
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.679231923224373e-06,
+      "loss": 0.5947,
+      "step": 14695
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.677934059735476e-06,
+      "loss": 0.6204,
+      "step": 14696
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.676636285767271e-06,
+      "loss": 0.5669,
+      "step": 14697
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.675338601346646e-06,
+      "loss": 0.6094,
+      "step": 14698
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.674041006500471e-06,
+      "loss": 0.5906,
+      "step": 14699
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.672743501255633e-06,
+      "loss": 0.5669,
+      "step": 14700
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.671446085638994e-06,
+      "loss": 0.5281,
+      "step": 14701
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.67014875967744e-06,
+      "loss": 0.5879,
+      "step": 14702
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.668851523397829e-06,
+      "loss": 0.5898,
+      "step": 14703
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6675543768270415e-06,
+      "loss": 0.5977,
+      "step": 14704
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.666257319991939e-06,
+      "loss": 0.5813,
+      "step": 14705
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6649603529193816e-06,
+      "loss": 0.5408,
+      "step": 14706
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.663663475636243e-06,
+      "loss": 0.575,
+      "step": 14707
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.662366688169377e-06,
+      "loss": 0.5044,
+      "step": 14708
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.661069990545649e-06,
+      "loss": 0.4758,
+      "step": 14709
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6597733827919095e-06,
+      "loss": 0.5461,
+      "step": 14710
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6584768649350245e-06,
+      "loss": 0.5632,
+      "step": 14711
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.657180437001837e-06,
+      "loss": 0.5039,
+      "step": 14712
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6558840990192085e-06,
+      "loss": 0.5381,
+      "step": 14713
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.654587851013983e-06,
+      "loss": 0.5381,
+      "step": 14714
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6532916930130096e-06,
+      "loss": 0.5842,
+      "step": 14715
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.651995625043143e-06,
+      "loss": 0.6665,
+      "step": 14716
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.650699647131217e-06,
+      "loss": 0.6155,
+      "step": 14717
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.64940375930408e-06,
+      "loss": 0.5098,
+      "step": 14718
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.648107961588576e-06,
+      "loss": 0.5581,
+      "step": 14719
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.646812254011541e-06,
+      "loss": 0.5811,
+      "step": 14720
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.645516636599808e-06,
+      "loss": 0.5842,
+      "step": 14721
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.64422110938022e-06,
+      "loss": 0.583,
+      "step": 14722
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6429256723796035e-06,
+      "loss": 0.5676,
+      "step": 14723
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.641630325624797e-06,
+      "loss": 0.6362,
+      "step": 14724
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.640335069142624e-06,
+      "loss": 0.583,
+      "step": 14725
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.639039902959919e-06,
+      "loss": 0.5415,
+      "step": 14726
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.637744827103501e-06,
+      "loss": 0.5574,
+      "step": 14727
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.636449841600202e-06,
+      "loss": 0.5684,
+      "step": 14728
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.635154946476835e-06,
+      "loss": 0.595,
+      "step": 14729
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.63386014176023e-06,
+      "loss": 0.5525,
+      "step": 14730
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.632565427477199e-06,
+      "loss": 0.5667,
+      "step": 14731
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.631270803654564e-06,
+      "loss": 0.5652,
+      "step": 14732
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.629976270319133e-06,
+      "loss": 0.6069,
+      "step": 14733
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.628681827497726e-06,
+      "loss": 0.5757,
+      "step": 14734
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.627387475217152e-06,
+      "loss": 0.5459,
+      "step": 14735
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.626093213504216e-06,
+      "loss": 0.604,
+      "step": 14736
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.624799042385732e-06,
+      "loss": 0.5137,
+      "step": 14737
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.623504961888497e-06,
+      "loss": 0.5625,
+      "step": 14738
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6222109720393215e-06,
+      "loss": 0.5291,
+      "step": 14739
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.620917072865009e-06,
+      "loss": 0.5244,
+      "step": 14740
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.619623264392351e-06,
+      "loss": 0.5989,
+      "step": 14741
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.618329546648152e-06,
+      "loss": 0.5979,
+      "step": 14742
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6170359196592095e-06,
+      "loss": 0.5256,
+      "step": 14743
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.615742383452311e-06,
+      "loss": 0.6323,
+      "step": 14744
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.614448938054255e-06,
+      "loss": 0.5391,
+      "step": 14745
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.613155583491827e-06,
+      "loss": 0.5474,
+      "step": 14746
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.611862319791822e-06,
+      "loss": 0.5193,
+      "step": 14747
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.610569146981017e-06,
+      "loss": 0.52,
+      "step": 14748
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.609276065086209e-06,
+      "loss": 0.5771,
+      "step": 14749
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.607983074134173e-06,
+      "loss": 0.4998,
+      "step": 14750
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6066901741516885e-06,
+      "loss": 0.6082,
+      "step": 14751
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.605397365165541e-06,
+      "loss": 0.6562,
+      "step": 14752
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.604104647202502e-06,
+      "loss": 0.55,
+      "step": 14753
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.602812020289351e-06,
+      "loss": 0.6292,
+      "step": 14754
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.601519484452857e-06,
+      "loss": 0.5691,
+      "step": 14755
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.600227039719799e-06,
+      "loss": 0.5864,
+      "step": 14756
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.598934686116937e-06,
+      "loss": 0.6138,
+      "step": 14757
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.597642423671049e-06,
+      "loss": 0.5955,
+      "step": 14758
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.596350252408891e-06,
+      "loss": 0.564,
+      "step": 14759
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.595058172357237e-06,
+      "loss": 0.4973,
+      "step": 14760
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.593766183542842e-06,
+      "loss": 0.5637,
+      "step": 14761
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5924742859924685e-06,
+      "loss": 0.5286,
+      "step": 14762
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.591182479732874e-06,
+      "loss": 0.541,
+      "step": 14763
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.589890764790819e-06,
+      "loss": 0.5747,
+      "step": 14764
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.588599141193052e-06,
+      "loss": 0.5427,
+      "step": 14765
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.587307608966328e-06,
+      "loss": 0.6306,
+      "step": 14766
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.586016168137403e-06,
+      "loss": 0.6052,
+      "step": 14767
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5847248187330165e-06,
+      "loss": 0.5828,
+      "step": 14768
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5834335607799255e-06,
+      "loss": 0.627,
+      "step": 14769
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5821423943048655e-06,
+      "loss": 0.55,
+      "step": 14770
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.58085131933459e-06,
+      "loss": 0.5029,
+      "step": 14771
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.57956033589583e-06,
+      "loss": 0.4968,
+      "step": 14772
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.578269444015334e-06,
+      "loss": 0.6294,
+      "step": 14773
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.576978643719831e-06,
+      "loss": 0.5957,
+      "step": 14774
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.575687935036066e-06,
+      "loss": 0.5298,
+      "step": 14775
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.574397317990763e-06,
+      "loss": 0.6619,
+      "step": 14776
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.573106792610664e-06,
+      "loss": 0.6108,
+      "step": 14777
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5718163589224926e-06,
+      "loss": 0.5925,
+      "step": 14778
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.570526016952975e-06,
+      "loss": 0.5464,
+      "step": 14779
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.569235766728844e-06,
+      "loss": 0.55,
+      "step": 14780
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.567945608276815e-06,
+      "loss": 0.583,
+      "step": 14781
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.566655541623622e-06,
+      "loss": 0.564,
+      "step": 14782
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.565365566795974e-06,
+      "loss": 0.5767,
+      "step": 14783
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.564075683820598e-06,
+      "loss": 0.6016,
+      "step": 14784
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.562785892724205e-06,
+      "loss": 0.5286,
+      "step": 14785
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.561496193533516e-06,
+      "loss": 0.5979,
+      "step": 14786
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.560206586275234e-06,
+      "loss": 0.5542,
+      "step": 14787
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.558917070976082e-06,
+      "loss": 0.5767,
+      "step": 14788
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5576276476627575e-06,
+      "loss": 0.5376,
+      "step": 14789
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.556338316361978e-06,
+      "loss": 0.6167,
+      "step": 14790
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5550490771004405e-06,
+      "loss": 0.6929,
+      "step": 14791
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.553759929904851e-06,
+      "loss": 0.5305,
+      "step": 14792
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.552470874801916e-06,
+      "loss": 0.5706,
+      "step": 14793
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.551181911818327e-06,
+      "loss": 0.5469,
+      "step": 14794
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.54989304098079e-06,
+      "loss": 0.5239,
+      "step": 14795
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.548604262315991e-06,
+      "loss": 0.5876,
+      "step": 14796
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.547315575850632e-06,
+      "loss": 0.5908,
+      "step": 14797
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.546026981611399e-06,
+      "loss": 0.532,
+      "step": 14798
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.544738479624989e-06,
+      "loss": 0.5117,
+      "step": 14799
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.543450069918082e-06,
+      "loss": 0.5571,
+      "step": 14800
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.54216175251737e-06,
+      "loss": 0.5374,
+      "step": 14801
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.540873527449533e-06,
+      "loss": 0.5671,
+      "step": 14802
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.539585394741259e-06,
+      "loss": 0.5728,
+      "step": 14803
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.538297354419221e-06,
+      "loss": 0.6326,
+      "step": 14804
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.537009406510105e-06,
+      "loss": 0.6494,
+      "step": 14805
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5357215510405815e-06,
+      "loss": 0.562,
+      "step": 14806
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.53443378803733e-06,
+      "loss": 0.6631,
+      "step": 14807
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.533146117527023e-06,
+      "loss": 0.5815,
+      "step": 14808
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5318585395363235e-06,
+      "loss": 0.5586,
+      "step": 14809
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.530571054091911e-06,
+      "loss": 0.5574,
+      "step": 14810
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.529283661220443e-06,
+      "loss": 0.6157,
+      "step": 14811
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.527996360948595e-06,
+      "loss": 0.5286,
+      "step": 14812
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5267091533030206e-06,
+      "loss": 0.6208,
+      "step": 14813
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.525422038310389e-06,
+      "loss": 0.6321,
+      "step": 14814
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.52413501599735e-06,
+      "loss": 0.5701,
+      "step": 14815
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.522848086390568e-06,
+      "loss": 0.5825,
+      "step": 14816
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5215612495167e-06,
+      "loss": 0.6484,
+      "step": 14817
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.520274505402393e-06,
+      "loss": 0.574,
+      "step": 14818
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.518987854074303e-06,
+      "loss": 0.5312,
+      "step": 14819
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.517701295559082e-06,
+      "loss": 0.6245,
+      "step": 14820
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.516414829883372e-06,
+      "loss": 0.6072,
+      "step": 14821
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.515128457073826e-06,
+      "loss": 0.5618,
+      "step": 14822
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.513842177157083e-06,
+      "loss": 0.4857,
+      "step": 14823
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.512555990159783e-06,
+      "loss": 0.502,
+      "step": 14824
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.511269896108571e-06,
+      "loss": 0.5852,
+      "step": 14825
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.50998389503008e-06,
+      "loss": 0.603,
+      "step": 14826
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5086979869509524e-06,
+      "loss": 0.613,
+      "step": 14827
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.507412171897817e-06,
+      "loss": 0.6145,
+      "step": 14828
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.506126449897312e-06,
+      "loss": 0.6772,
+      "step": 14829
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.50484082097606e-06,
+      "loss": 0.4771,
+      "step": 14830
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.5035552851606995e-06,
+      "loss": 0.6091,
+      "step": 14831
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.502269842477847e-06,
+      "loss": 0.5686,
+      "step": 14832
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.500984492954138e-06,
+      "loss": 0.6396,
+      "step": 14833
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.499699236616183e-06,
+      "loss": 0.5623,
+      "step": 14834
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4984140734906145e-06,
+      "loss": 0.5437,
+      "step": 14835
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.497129003604042e-06,
+      "loss": 0.6016,
+      "step": 14836
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4958440269830895e-06,
+      "loss": 0.666,
+      "step": 14837
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.49455914365437e-06,
+      "loss": 0.5591,
+      "step": 14838
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.493274353644491e-06,
+      "loss": 0.5928,
+      "step": 14839
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.491989656980068e-06,
+      "loss": 0.6748,
+      "step": 14840
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.490705053687716e-06,
+      "loss": 0.5237,
+      "step": 14841
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.489420543794033e-06,
+      "loss": 0.7183,
+      "step": 14842
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.488136127325627e-06,
+      "loss": 0.6052,
+      "step": 14843
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.486851804309108e-06,
+      "loss": 0.5261,
+      "step": 14844
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.485567574771067e-06,
+      "loss": 0.6357,
+      "step": 14845
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.484283438738114e-06,
+      "loss": 0.519,
+      "step": 14846
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.482999396236837e-06,
+      "loss": 0.5164,
+      "step": 14847
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.48171544729384e-06,
+      "loss": 0.6238,
+      "step": 14848
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.48043159193571e-06,
+      "loss": 0.6116,
+      "step": 14849
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.479147830189044e-06,
+      "loss": 0.5908,
+      "step": 14850
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.477864162080428e-06,
+      "loss": 0.522,
+      "step": 14851
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.476580587636456e-06,
+      "loss": 0.5708,
+      "step": 14852
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.475297106883707e-06,
+      "loss": 0.5544,
+      "step": 14853
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.474013719848765e-06,
+      "loss": 0.5166,
+      "step": 14854
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4727304265582195e-06,
+      "loss": 0.5623,
+      "step": 14855
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.47144722703864e-06,
+      "loss": 0.5326,
+      "step": 14856
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.470164121316617e-06,
+      "loss": 0.5742,
+      "step": 14857
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.468881109418716e-06,
+      "loss": 0.5918,
+      "step": 14858
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4675981913715195e-06,
+      "loss": 0.5439,
+      "step": 14859
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.466315367201592e-06,
+      "loss": 0.6643,
+      "step": 14860
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.465032636935512e-06,
+      "loss": 0.5808,
+      "step": 14861
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4637500005998425e-06,
+      "loss": 0.5474,
+      "step": 14862
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4624674582211545e-06,
+      "loss": 0.5154,
+      "step": 14863
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.461185009826006e-06,
+      "loss": 0.5867,
+      "step": 14864
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.459902655440967e-06,
+      "loss": 0.4846,
+      "step": 14865
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4586203950925935e-06,
+      "loss": 0.5339,
+      "step": 14866
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.457338228807447e-06,
+      "loss": 0.5405,
+      "step": 14867
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4560561566120795e-06,
+      "loss": 0.5603,
+      "step": 14868
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.454774178533051e-06,
+      "loss": 0.5208,
+      "step": 14869
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.453492294596916e-06,
+      "loss": 0.5635,
+      "step": 14870
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4522105048302186e-06,
+      "loss": 0.5386,
+      "step": 14871
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.450928809259514e-06,
+      "loss": 0.5798,
+      "step": 14872
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.449647207911346e-06,
+      "loss": 0.6094,
+      "step": 14873
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.448365700812264e-06,
+      "loss": 0.5115,
+      "step": 14874
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.447084287988802e-06,
+      "loss": 0.5398,
+      "step": 14875
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.445802969467513e-06,
+      "loss": 0.6338,
+      "step": 14876
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4445217452749254e-06,
+      "loss": 0.5349,
+      "step": 14877
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.443240615437586e-06,
+      "loss": 0.6062,
+      "step": 14878
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.441959579982021e-06,
+      "loss": 0.5526,
+      "step": 14879
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.440678638934773e-06,
+      "loss": 0.6897,
+      "step": 14880
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.439397792322365e-06,
+      "loss": 0.5371,
+      "step": 14881
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.438117040171335e-06,
+      "loss": 0.5435,
+      "step": 14882
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.436836382508206e-06,
+      "loss": 0.6333,
+      "step": 14883
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4355558193594994e-06,
+      "loss": 0.623,
+      "step": 14884
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4342753507517455e-06,
+      "loss": 0.5815,
+      "step": 14885
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.432994976711461e-06,
+      "loss": 0.55,
+      "step": 14886
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.431714697265172e-06,
+      "loss": 0.625,
+      "step": 14887
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4304345124393886e-06,
+      "loss": 0.6372,
+      "step": 14888
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.429154422260635e-06,
+      "loss": 0.5588,
+      "step": 14889
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.427874426755415e-06,
+      "loss": 0.5979,
+      "step": 14890
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.426594525950252e-06,
+      "loss": 0.6121,
+      "step": 14891
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.425314719871644e-06,
+      "loss": 0.6025,
+      "step": 14892
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.424035008546106e-06,
+      "loss": 0.5632,
+      "step": 14893
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4227553920001465e-06,
+      "loss": 0.6313,
+      "step": 14894
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.421475870260261e-06,
+      "loss": 0.5645,
+      "step": 14895
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.420196443352957e-06,
+      "loss": 0.5745,
+      "step": 14896
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.418917111304738e-06,
+      "loss": 0.5046,
+      "step": 14897
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.417637874142099e-06,
+      "loss": 0.5232,
+      "step": 14898
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.41635873189153e-06,
+      "loss": 0.4725,
+      "step": 14899
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.415079684579536e-06,
+      "loss": 0.5029,
+      "step": 14900
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.413800732232598e-06,
+      "loss": 0.4954,
+      "step": 14901
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.412521874877217e-06,
+      "loss": 0.6624,
+      "step": 14902
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.411243112539872e-06,
+      "loss": 0.5586,
+      "step": 14903
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.409964445247058e-06,
+      "loss": 0.5906,
+      "step": 14904
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.408685873025251e-06,
+      "loss": 0.6138,
+      "step": 14905
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.407407395900942e-06,
+      "loss": 0.553,
+      "step": 14906
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.406129013900603e-06,
+      "loss": 0.6138,
+      "step": 14907
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.404850727050719e-06,
+      "loss": 0.5801,
+      "step": 14908
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.40357253537776e-06,
+      "loss": 0.6179,
+      "step": 14909
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4022944389082085e-06,
+      "loss": 0.5669,
+      "step": 14910
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.401016437668529e-06,
+      "loss": 0.5298,
+      "step": 14911
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.399738531685202e-06,
+      "loss": 0.5627,
+      "step": 14912
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3984607209846875e-06,
+      "loss": 0.5894,
+      "step": 14913
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3971830055934514e-06,
+      "loss": 0.5776,
+      "step": 14914
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.395905385537965e-06,
+      "loss": 0.5305,
+      "step": 14915
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.394627860844686e-06,
+      "loss": 0.6064,
+      "step": 14916
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.393350431540074e-06,
+      "loss": 0.6128,
+      "step": 14917
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.392073097650596e-06,
+      "loss": 0.5562,
+      "step": 14918
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.390795859202698e-06,
+      "loss": 0.5752,
+      "step": 14919
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.389518716222841e-06,
+      "loss": 0.4873,
+      "step": 14920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3882416687374806e-06,
+      "loss": 0.5276,
+      "step": 14921
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.38696471677306e-06,
+      "loss": 0.5278,
+      "step": 14922
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.385687860356035e-06,
+      "loss": 0.5693,
+      "step": 14923
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.384411099512845e-06,
+      "loss": 0.4919,
+      "step": 14924
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3831344342699455e-06,
+      "loss": 0.5991,
+      "step": 14925
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.381857864653767e-06,
+      "loss": 0.6562,
+      "step": 14926
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.38058139069076e-06,
+      "loss": 0.5552,
+      "step": 14927
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.379305012407361e-06,
+      "loss": 0.5789,
+      "step": 14928
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.378028729830003e-06,
+      "loss": 0.6479,
+      "step": 14929
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.376752542985126e-06,
+      "loss": 0.5205,
+      "step": 14930
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3754764518991575e-06,
+      "loss": 0.5674,
+      "step": 14931
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3742004565985365e-06,
+      "loss": 0.5505,
+      "step": 14932
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.372924557109682e-06,
+      "loss": 0.5703,
+      "step": 14933
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.371648753459031e-06,
+      "loss": 0.5669,
+      "step": 14934
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3703730456730005e-06,
+      "loss": 0.5957,
+      "step": 14935
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.369097433778021e-06,
+      "loss": 0.532,
+      "step": 14936
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.367821917800506e-06,
+      "loss": 0.5706,
+      "step": 14937
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3665464977668805e-06,
+      "loss": 0.5618,
+      "step": 14938
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.365271173703557e-06,
+      "loss": 0.636,
+      "step": 14939
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.363995945636956e-06,
+      "loss": 0.5461,
+      "step": 14940
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.362720813593484e-06,
+      "loss": 0.4956,
+      "step": 14941
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.36144577759956e-06,
+      "loss": 0.5718,
+      "step": 14942
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3601708376815845e-06,
+      "loss": 0.6548,
+      "step": 14943
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.358895993865973e-06,
+      "loss": 0.593,
+      "step": 14944
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.357621246179123e-06,
+      "loss": 0.5438,
+      "step": 14945
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.356346594647442e-06,
+      "loss": 0.585,
+      "step": 14946
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.355072039297333e-06,
+      "loss": 0.6392,
+      "step": 14947
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.35379758015519e-06,
+      "loss": 0.5869,
+      "step": 14948
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3525232172474165e-06,
+      "loss": 0.5891,
+      "step": 14949
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3512489506004e-06,
+      "loss": 0.6384,
+      "step": 14950
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.349974780240543e-06,
+      "loss": 0.592,
+      "step": 14951
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.348700706194227e-06,
+      "loss": 0.5664,
+      "step": 14952
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.34742672848785e-06,
+      "loss": 0.5044,
+      "step": 14953
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.346152847147792e-06,
+      "loss": 0.5256,
+      "step": 14954
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.344879062200444e-06,
+      "loss": 0.6294,
+      "step": 14955
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.343605373672182e-06,
+      "loss": 0.5513,
+      "step": 14956
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.342331781589398e-06,
+      "loss": 0.5259,
+      "step": 14957
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.341058285978464e-06,
+      "loss": 0.5889,
+      "step": 14958
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3397848868657555e-06,
+      "loss": 0.5415,
+      "step": 14959
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3385115842776535e-06,
+      "loss": 0.5969,
+      "step": 14960
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.337238378240525e-06,
+      "loss": 0.6035,
+      "step": 14961
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3359652687807476e-06,
+      "loss": 0.5586,
+      "step": 14962
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.334692255924685e-06,
+      "loss": 0.5923,
+      "step": 14963
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.333419339698709e-06,
+      "loss": 0.4927,
+      "step": 14964
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.33214652012918e-06,
+      "loss": 0.6306,
+      "step": 14965
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.330873797242468e-06,
+      "loss": 0.5789,
+      "step": 14966
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.329601171064927e-06,
+      "loss": 0.521,
+      "step": 14967
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3283286416229244e-06,
+      "loss": 0.574,
+      "step": 14968
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.327056208942807e-06,
+      "loss": 0.6436,
+      "step": 14969
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.325783873050936e-06,
+      "loss": 0.5796,
+      "step": 14970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.324511633973668e-06,
+      "loss": 0.6057,
+      "step": 14971
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.323239491737347e-06,
+      "loss": 0.5396,
+      "step": 14972
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3219674463683305e-06,
+      "loss": 0.5908,
+      "step": 14973
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.320695497892956e-06,
+      "loss": 0.5747,
+      "step": 14974
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.319423646337577e-06,
+      "loss": 0.5603,
+      "step": 14975
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.318151891728532e-06,
+      "loss": 0.5745,
+      "step": 14976
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.316880234092165e-06,
+      "loss": 0.5583,
+      "step": 14977
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.315608673454809e-06,
+      "loss": 0.6079,
+      "step": 14978
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.314337209842813e-06,
+      "loss": 0.5867,
+      "step": 14979
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.313065843282499e-06,
+      "loss": 0.5469,
+      "step": 14980
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.311794573800211e-06,
+      "loss": 0.5872,
+      "step": 14981
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.31052340142227e-06,
+      "loss": 0.5278,
+      "step": 14982
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.309252326175016e-06,
+      "loss": 0.5127,
+      "step": 14983
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.307981348084766e-06,
+      "loss": 0.6619,
+      "step": 14984
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.306710467177854e-06,
+      "loss": 0.541,
+      "step": 14985
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3054396834806e-06,
+      "loss": 0.5188,
+      "step": 14986
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3041689970193175e-06,
+      "loss": 0.5076,
+      "step": 14987
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.302898407820338e-06,
+      "loss": 0.5214,
+      "step": 14988
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.301627915909967e-06,
+      "loss": 0.5513,
+      "step": 14989
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.30035752131453e-06,
+      "loss": 0.5391,
+      "step": 14990
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.299087224060331e-06,
+      "loss": 0.5146,
+      "step": 14991
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.297817024173688e-06,
+      "loss": 0.5486,
+      "step": 14992
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2965469216809055e-06,
+      "loss": 0.5259,
+      "step": 14993
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2952769166082896e-06,
+      "loss": 0.5186,
+      "step": 14994
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.294007008982151e-06,
+      "loss": 0.5938,
+      "step": 14995
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.292737198828788e-06,
+      "loss": 0.5469,
+      "step": 14996
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2914674861745016e-06,
+      "loss": 0.6348,
+      "step": 14997
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.290197871045597e-06,
+      "loss": 0.5842,
+      "step": 14998
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.288928353468361e-06,
+      "loss": 0.6223,
+      "step": 14999
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.287658933469096e-06,
+      "loss": 0.606,
+      "step": 15000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.286389611074094e-06,
+      "loss": 0.5696,
+      "step": 15001
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.28512038630964e-06,
+      "loss": 0.5464,
+      "step": 15002
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.283851259202031e-06,
+      "loss": 0.5864,
+      "step": 15003
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.282582229777546e-06,
+      "loss": 0.5498,
+      "step": 15004
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.281313298062476e-06,
+      "loss": 0.48,
+      "step": 15005
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.280044464083098e-06,
+      "loss": 0.5386,
+      "step": 15006
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.278775727865699e-06,
+      "loss": 0.5186,
+      "step": 15007
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.277507089436552e-06,
+      "loss": 0.6101,
+      "step": 15008
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.276238548821939e-06,
+      "loss": 0.5063,
+      "step": 15009
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2749701060481285e-06,
+      "loss": 0.5796,
+      "step": 15010
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2737017611414e-06,
+      "loss": 0.5352,
+      "step": 15011
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.272433514128017e-06,
+      "loss": 0.5449,
+      "step": 15012
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.271165365034254e-06,
+      "loss": 0.6958,
+      "step": 15013
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.269897313886372e-06,
+      "loss": 0.5796,
+      "step": 15014
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.268629360710642e-06,
+      "loss": 0.5876,
+      "step": 15015
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2673615055333235e-06,
+      "loss": 0.5598,
+      "step": 15016
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.266093748380672e-06,
+      "loss": 0.5813,
+      "step": 15017
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.264826089278953e-06,
+      "loss": 0.6055,
+      "step": 15018
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.263558528254418e-06,
+      "loss": 0.6204,
+      "step": 15019
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.262291065333324e-06,
+      "loss": 0.5872,
+      "step": 15020
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.261023700541925e-06,
+      "loss": 0.5536,
+      "step": 15021
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.259756433906466e-06,
+      "loss": 0.593,
+      "step": 15022
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.258489265453197e-06,
+      "loss": 0.5654,
+      "step": 15023
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2572221952083705e-06,
+      "loss": 0.595,
+      "step": 15024
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.255955223198221e-06,
+      "loss": 0.5825,
+      "step": 15025
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2546883494489984e-06,
+      "loss": 0.5659,
+      "step": 15026
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.253421573986936e-06,
+      "loss": 0.563,
+      "step": 15027
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.252154896838281e-06,
+      "loss": 0.574,
+      "step": 15028
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.250888318029259e-06,
+      "loss": 0.4902,
+      "step": 15029
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.249621837586112e-06,
+      "loss": 0.5435,
+      "step": 15030
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.24835545553507e-06,
+      "loss": 0.5969,
+      "step": 15031
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.247089171902358e-06,
+      "loss": 0.5867,
+      "step": 15032
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.24582298671421e-06,
+      "loss": 0.5806,
+      "step": 15033
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.244556899996845e-06,
+      "loss": 0.52,
+      "step": 15034
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.243290911776497e-06,
+      "loss": 0.5413,
+      "step": 15035
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2420250220793775e-06,
+      "loss": 0.6016,
+      "step": 15036
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.240759230931715e-06,
+      "loss": 0.6055,
+      "step": 15037
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2394935383597176e-06,
+      "loss": 0.6108,
+      "step": 15038
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.238227944389611e-06,
+      "loss": 0.4902,
+      "step": 15039
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2369624490476e-06,
+      "loss": 0.5571,
+      "step": 15040
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.235697052359905e-06,
+      "loss": 0.5574,
+      "step": 15041
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.234431754352725e-06,
+      "loss": 0.5708,
+      "step": 15042
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.233166555052278e-06,
+      "loss": 0.6848,
+      "step": 15043
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.231901454484761e-06,
+      "loss": 0.6377,
+      "step": 15044
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.230636452676383e-06,
+      "loss": 0.5449,
+      "step": 15045
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2293715496533405e-06,
+      "loss": 0.5598,
+      "step": 15046
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2281067454418354e-06,
+      "loss": 0.5195,
+      "step": 15047
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.226842040068069e-06,
+      "loss": 0.5642,
+      "step": 15048
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.225577433558229e-06,
+      "loss": 0.5481,
+      "step": 15049
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.224312925938515e-06,
+      "loss": 0.5354,
+      "step": 15050
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2230485172351104e-06,
+      "loss": 0.5332,
+      "step": 15051
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2217842074742145e-06,
+      "loss": 0.6128,
+      "step": 15052
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.220519996682004e-06,
+      "loss": 0.5715,
+      "step": 15053
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.219255884884673e-06,
+      "loss": 0.5898,
+      "step": 15054
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.217991872108395e-06,
+      "loss": 0.6304,
+      "step": 15055
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.216727958379361e-06,
+      "loss": 0.6272,
+      "step": 15056
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.21546414372374e-06,
+      "loss": 0.6589,
+      "step": 15057
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.214200428167717e-06,
+      "loss": 0.6082,
+      "step": 15058
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.212936811737459e-06,
+      "loss": 0.6211,
+      "step": 15059
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.211673294459147e-06,
+      "loss": 0.6106,
+      "step": 15060
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.210409876358946e-06,
+      "loss": 0.6191,
+      "step": 15061
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.209146557463023e-06,
+      "loss": 0.5188,
+      "step": 15062
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2078833377975516e-06,
+      "loss": 0.575,
+      "step": 15063
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.206620217388687e-06,
+      "loss": 0.5432,
+      "step": 15064
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2053571962626e-06,
+      "loss": 0.5198,
+      "step": 15065
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.204094274445444e-06,
+      "loss": 0.5852,
+      "step": 15066
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.202831451963384e-06,
+      "loss": 0.5659,
+      "step": 15067
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.201568728842569e-06,
+      "loss": 0.4983,
+      "step": 15068
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.200306105109161e-06,
+      "loss": 0.5925,
+      "step": 15069
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.199043580789302e-06,
+      "loss": 0.5132,
+      "step": 15070
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.197781155909149e-06,
+      "loss": 0.531,
+      "step": 15071
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.196518830494854e-06,
+      "loss": 0.5649,
+      "step": 15072
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.195256604572552e-06,
+      "loss": 0.6072,
+      "step": 15073
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.193994478168394e-06,
+      "loss": 0.5452,
+      "step": 15074
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.192732451308522e-06,
+      "loss": 0.5642,
+      "step": 15075
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.191470524019076e-06,
+      "loss": 0.5508,
+      "step": 15076
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1902086963261865e-06,
+      "loss": 0.5857,
+      "step": 15077
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.188946968255999e-06,
+      "loss": 0.5615,
+      "step": 15078
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.187685339834638e-06,
+      "loss": 0.6108,
+      "step": 15079
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.186423811088244e-06,
+      "loss": 0.5046,
+      "step": 15080
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.185162382042936e-06,
+      "loss": 0.5522,
+      "step": 15081
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1839010527248515e-06,
+      "loss": 0.5542,
+      "step": 15082
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.182639823160107e-06,
+      "loss": 0.5889,
+      "step": 15083
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.181378693374835e-06,
+      "loss": 0.5481,
+      "step": 15084
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.180117663395146e-06,
+      "loss": 0.5701,
+      "step": 15085
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.178856733247169e-06,
+      "loss": 0.6145,
+      "step": 15086
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.177595902957011e-06,
+      "loss": 0.6104,
+      "step": 15087
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.176335172550799e-06,
+      "loss": 0.5708,
+      "step": 15088
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1750745420546345e-06,
+      "loss": 0.6155,
+      "step": 15089
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1738140114946355e-06,
+      "loss": 0.5981,
+      "step": 15090
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.17255358089691e-06,
+      "loss": 0.5686,
+      "step": 15091
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.171293250287558e-06,
+      "loss": 0.5813,
+      "step": 15092
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.170033019692693e-06,
+      "loss": 0.5645,
+      "step": 15093
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.168772889138409e-06,
+      "loss": 0.5586,
+      "step": 15094
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.167512858650816e-06,
+      "loss": 0.6475,
+      "step": 15095
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1662529282560025e-06,
+      "loss": 0.571,
+      "step": 15096
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.16499309798007e-06,
+      "loss": 0.6545,
+      "step": 15097
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.163733367849116e-06,
+      "loss": 0.6248,
+      "step": 15098
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1624737378892245e-06,
+      "loss": 0.5742,
+      "step": 15099
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1612142081264905e-06,
+      "loss": 0.5208,
+      "step": 15100
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1599547785870055e-06,
+      "loss": 0.5498,
+      "step": 15101
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.158695449296846e-06,
+      "loss": 0.5808,
+      "step": 15102
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.157436220282107e-06,
+      "loss": 0.5908,
+      "step": 15103
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.156177091568859e-06,
+      "loss": 0.5342,
+      "step": 15104
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.154918063183192e-06,
+      "loss": 0.6401,
+      "step": 15105
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.153659135151179e-06,
+      "loss": 0.6143,
+      "step": 15106
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.15240030749889e-06,
+      "loss": 0.592,
+      "step": 15107
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1511415802524075e-06,
+      "loss": 0.5461,
+      "step": 15108
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.149882953437796e-06,
+      "loss": 0.5249,
+      "step": 15109
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1486244270811304e-06,
+      "loss": 0.5398,
+      "step": 15110
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.147366001208473e-06,
+      "loss": 0.6208,
+      "step": 15111
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.146107675845895e-06,
+      "loss": 0.5867,
+      "step": 15112
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.144849451019451e-06,
+      "loss": 0.5371,
+      "step": 15113
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.14359132675521e-06,
+      "loss": 0.5222,
+      "step": 15114
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.142333303079223e-06,
+      "loss": 0.571,
+      "step": 15115
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.141075380017557e-06,
+      "loss": 0.5994,
+      "step": 15116
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1398175575962565e-06,
+      "loss": 0.5386,
+      "step": 15117
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.138559835841381e-06,
+      "loss": 0.6348,
+      "step": 15118
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.137302214778976e-06,
+      "loss": 0.6133,
+      "step": 15119
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.136044694435097e-06,
+      "loss": 0.5447,
+      "step": 15120
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1347872748357815e-06,
+      "loss": 0.5876,
+      "step": 15121
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.133529956007082e-06,
+      "loss": 0.5078,
+      "step": 15122
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.132272737975033e-06,
+      "loss": 0.5513,
+      "step": 15123
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.131015620765678e-06,
+      "loss": 0.5552,
+      "step": 15124
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.129758604405059e-06,
+      "loss": 0.4854,
+      "step": 15125
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1285016889192055e-06,
+      "loss": 0.6011,
+      "step": 15126
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1272448743341585e-06,
+      "loss": 0.5967,
+      "step": 15127
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.125988160675941e-06,
+      "loss": 0.5823,
+      "step": 15128
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.124731547970591e-06,
+      "loss": 0.5288,
+      "step": 15129
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.123475036244127e-06,
+      "loss": 0.6062,
+      "step": 15130
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.122218625522586e-06,
+      "loss": 0.55,
+      "step": 15131
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.120962315831979e-06,
+      "loss": 0.6062,
+      "step": 15132
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.119706107198339e-06,
+      "loss": 0.5918,
+      "step": 15133
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.118449999647674e-06,
+      "loss": 0.5068,
+      "step": 15134
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.117193993206012e-06,
+      "loss": 0.5623,
+      "step": 15135
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.115938087899361e-06,
+      "loss": 0.5508,
+      "step": 15136
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.114682283753734e-06,
+      "loss": 0.6208,
+      "step": 15137
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.113426580795146e-06,
+      "loss": 0.6504,
+      "step": 15138
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.112170979049599e-06,
+      "loss": 0.5825,
+      "step": 15139
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.110915478543108e-06,
+      "loss": 0.5386,
+      "step": 15140
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.109660079301668e-06,
+      "loss": 0.5027,
+      "step": 15141
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.108404781351293e-06,
+      "loss": 0.6062,
+      "step": 15142
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.107149584717971e-06,
+      "loss": 0.575,
+      "step": 15143
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1058944894277105e-06,
+      "loss": 0.6243,
+      "step": 15144
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1046394955065e-06,
+      "loss": 0.5398,
+      "step": 15145
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.1033846029803415e-06,
+      "loss": 0.6519,
+      "step": 15146
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.102129811875216e-06,
+      "loss": 0.583,
+      "step": 15147
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.1008751222171216e-06,
+      "loss": 0.5227,
+      "step": 15148
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.099620534032048e-06,
+      "loss": 0.6401,
+      "step": 15149
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.098366047345973e-06,
+      "loss": 0.6433,
+      "step": 15150
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.097111662184888e-06,
+      "loss": 0.5256,
+      "step": 15151
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.095857378574765e-06,
+      "loss": 0.5059,
+      "step": 15152
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.094603196541594e-06,
+      "loss": 0.5776,
+      "step": 15153
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.093349116111343e-06,
+      "loss": 0.5859,
+      "step": 15154
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.092095137309996e-06,
+      "loss": 0.6133,
+      "step": 15155
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.090841260163516e-06,
+      "loss": 0.5442,
+      "step": 15156
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0895874846978834e-06,
+      "loss": 0.542,
+      "step": 15157
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.088333810939059e-06,
+      "loss": 0.5789,
+      "step": 15158
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.087080238913016e-06,
+      "loss": 0.5793,
+      "step": 15159
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.085826768645712e-06,
+      "loss": 0.563,
+      "step": 15160
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0845734001631195e-06,
+      "loss": 0.6147,
+      "step": 15161
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.083320133491188e-06,
+      "loss": 0.6045,
+      "step": 15162
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0820669686558835e-06,
+      "loss": 0.6758,
+      "step": 15163
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.080813905683157e-06,
+      "loss": 0.5981,
+      "step": 15164
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.079560944598968e-06,
+      "loss": 0.5867,
+      "step": 15165
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.078308085429266e-06,
+      "loss": 0.5872,
+      "step": 15166
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.077055328199995e-06,
+      "loss": 0.5627,
+      "step": 15167
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.075802672937113e-06,
+      "loss": 0.5842,
+      "step": 15168
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.074550119666557e-06,
+      "loss": 0.6182,
+      "step": 15169
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.073297668414277e-06,
+      "loss": 0.5635,
+      "step": 15170
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.072045319206207e-06,
+      "loss": 0.6221,
+      "step": 15171
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.070793072068294e-06,
+      "loss": 0.5659,
+      "step": 15172
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.069540927026469e-06,
+      "loss": 0.5669,
+      "step": 15173
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.068288884106669e-06,
+      "loss": 0.522,
+      "step": 15174
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.067036943334834e-06,
+      "loss": 0.6023,
+      "step": 15175
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.065785104736881e-06,
+      "loss": 0.6326,
+      "step": 15176
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.064533368338749e-06,
+      "loss": 0.5701,
+      "step": 15177
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0632817341663635e-06,
+      "loss": 0.4905,
+      "step": 15178
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.062030202245644e-06,
+      "loss": 0.5955,
+      "step": 15179
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0607787726025195e-06,
+      "loss": 0.55,
+      "step": 15180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.059527445262906e-06,
+      "loss": 0.5591,
+      "step": 15181
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.05827622025272e-06,
+      "loss": 0.5928,
+      "step": 15182
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.057025097597883e-06,
+      "loss": 0.6018,
+      "step": 15183
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.055774077324302e-06,
+      "loss": 0.6079,
+      "step": 15184
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.054523159457896e-06,
+      "loss": 0.5706,
+      "step": 15185
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0532723440245665e-06,
+      "loss": 0.5442,
+      "step": 15186
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.052021631050229e-06,
+      "loss": 0.5522,
+      "step": 15187
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.050771020560783e-06,
+      "loss": 0.6113,
+      "step": 15188
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.049520512582136e-06,
+      "loss": 0.5417,
+      "step": 15189
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0482701071401845e-06,
+      "loss": 0.5752,
+      "step": 15190
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.047019804260833e-06,
+      "loss": 0.5647,
+      "step": 15191
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.045769603969972e-06,
+      "loss": 0.5696,
+      "step": 15192
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.044519506293504e-06,
+      "loss": 0.5647,
+      "step": 15193
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.043269511257318e-06,
+      "loss": 0.6558,
+      "step": 15194
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.042019618887299e-06,
+      "loss": 0.5642,
+      "step": 15195
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.040769829209344e-06,
+      "loss": 0.5549,
+      "step": 15196
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.039520142249332e-06,
+      "loss": 0.5603,
+      "step": 15197
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.03827055803315e-06,
+      "loss": 0.5835,
+      "step": 15198
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0370210765866855e-06,
+      "loss": 0.583,
+      "step": 15199
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.035771697935808e-06,
+      "loss": 0.512,
+      "step": 15200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.034522422106403e-06,
+      "loss": 0.5254,
+      "step": 15201
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.033273249124345e-06,
+      "loss": 0.5996,
+      "step": 15202
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.032024179015504e-06,
+      "loss": 0.5161,
+      "step": 15203
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.030775211805756e-06,
+      "loss": 0.634,
+      "step": 15204
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.029526347520966e-06,
+      "loss": 0.6162,
+      "step": 15205
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.028277586187005e-06,
+      "loss": 0.5605,
+      "step": 15206
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.027028927829734e-06,
+      "loss": 0.5386,
+      "step": 15207
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.025780372475018e-06,
+      "loss": 0.5569,
+      "step": 15208
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.02453192014872e-06,
+      "loss": 0.582,
+      "step": 15209
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.023283570876691e-06,
+      "loss": 0.5791,
+      "step": 15210
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.022035324684797e-06,
+      "loss": 0.5798,
+      "step": 15211
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.020787181598883e-06,
+      "loss": 0.5645,
+      "step": 15212
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.019539141644809e-06,
+      "loss": 0.5679,
+      "step": 15213
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.018291204848418e-06,
+      "loss": 0.6055,
+      "step": 15214
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.017043371235565e-06,
+      "loss": 0.5042,
+      "step": 15215
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0157956408320885e-06,
+      "loss": 0.5872,
+      "step": 15216
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0145480136638405e-06,
+      "loss": 0.5945,
+      "step": 15217
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.013300489756653e-06,
+      "loss": 0.6064,
+      "step": 15218
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.012053069136374e-06,
+      "loss": 0.5708,
+      "step": 15219
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.010805751828833e-06,
+      "loss": 0.5925,
+      "step": 15220
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.009558537859873e-06,
+      "loss": 0.5376,
+      "step": 15221
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.008311427255318e-06,
+      "loss": 0.5737,
+      "step": 15222
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.007064420041008e-06,
+      "loss": 0.5864,
+      "step": 15223
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.005817516242762e-06,
+      "loss": 0.6011,
+      "step": 15224
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.004570715886413e-06,
+      "loss": 0.5513,
+      "step": 15225
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0033240189977876e-06,
+      "loss": 0.5503,
+      "step": 15226
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.002077425602699e-06,
+      "loss": 0.5764,
+      "step": 15227
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.000830935726977e-06,
+      "loss": 0.5774,
+      "step": 15228
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.999584549396431e-06,
+      "loss": 0.6089,
+      "step": 15229
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9983382666368855e-06,
+      "loss": 0.5081,
+      "step": 15230
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.997092087474144e-06,
+      "loss": 0.6118,
+      "step": 15231
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.995846011934028e-06,
+      "loss": 0.5186,
+      "step": 15232
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.994600040042338e-06,
+      "loss": 0.6191,
+      "step": 15233
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.993354171824889e-06,
+      "loss": 0.5698,
+      "step": 15234
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.992108407307478e-06,
+      "loss": 0.5955,
+      "step": 15235
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.990862746515917e-06,
+      "loss": 0.5505,
+      "step": 15236
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.989617189475996e-06,
+      "loss": 0.5747,
+      "step": 15237
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9883717362135255e-06,
+      "loss": 0.6008,
+      "step": 15238
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.987126386754294e-06,
+      "loss": 0.6428,
+      "step": 15239
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.985881141124095e-06,
+      "loss": 0.6343,
+      "step": 15240
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9846359993487255e-06,
+      "loss": 0.6062,
+      "step": 15241
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.983390961453971e-06,
+      "loss": 0.5493,
+      "step": 15242
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.982146027465625e-06,
+      "loss": 0.509,
+      "step": 15243
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.980901197409466e-06,
+      "loss": 0.5718,
+      "step": 15244
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.979656471311285e-06,
+      "loss": 0.5522,
+      "step": 15245
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9784118491968544e-06,
+      "loss": 0.4697,
+      "step": 15246
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.977167331091963e-06,
+      "loss": 0.6228,
+      "step": 15247
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.975922917022381e-06,
+      "loss": 0.5732,
+      "step": 15248
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.974678607013889e-06,
+      "loss": 0.5435,
+      "step": 15249
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.973434401092251e-06,
+      "loss": 0.6194,
+      "step": 15250
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9721902992832445e-06,
+      "loss": 0.6494,
+      "step": 15251
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9709463016126406e-06,
+      "loss": 0.5635,
+      "step": 15252
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.969702408106196e-06,
+      "loss": 0.5859,
+      "step": 15253
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.968458618789686e-06,
+      "loss": 0.5269,
+      "step": 15254
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.967214933688862e-06,
+      "loss": 0.6365,
+      "step": 15255
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9659713528294916e-06,
+      "loss": 0.6362,
+      "step": 15256
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.964727876237328e-06,
+      "loss": 0.5225,
+      "step": 15257
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.96348450393813e-06,
+      "loss": 0.6072,
+      "step": 15258
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.962241235957646e-06,
+      "loss": 0.5321,
+      "step": 15259
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.960998072321633e-06,
+      "loss": 0.5872,
+      "step": 15260
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.959755013055834e-06,
+      "loss": 0.5793,
+      "step": 15261
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9585120581860046e-06,
+      "loss": 0.5964,
+      "step": 15262
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.957269207737878e-06,
+      "loss": 0.5549,
+      "step": 15263
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.956026461737208e-06,
+      "loss": 0.593,
+      "step": 15264
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.954783820209724e-06,
+      "loss": 0.5908,
+      "step": 15265
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9535412831811744e-06,
+      "loss": 0.5876,
+      "step": 15266
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.952298850677287e-06,
+      "loss": 0.6523,
+      "step": 15267
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.951056522723804e-06,
+      "loss": 0.4963,
+      "step": 15268
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.949814299346452e-06,
+      "loss": 0.5483,
+      "step": 15269
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.948572180570956e-06,
+      "loss": 0.4941,
+      "step": 15270
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.947330166423052e-06,
+      "loss": 0.5176,
+      "step": 15271
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.946088256928459e-06,
+      "loss": 0.5452,
+      "step": 15272
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9448464521129044e-06,
+      "loss": 0.5334,
+      "step": 15273
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.943604752002106e-06,
+      "loss": 0.5505,
+      "step": 15274
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.942363156621781e-06,
+      "loss": 0.5435,
+      "step": 15275
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.941121665997653e-06,
+      "loss": 0.5586,
+      "step": 15276
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.939880280155429e-06,
+      "loss": 0.4893,
+      "step": 15277
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.938638999120824e-06,
+      "loss": 0.584,
+      "step": 15278
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.937397822919551e-06,
+      "loss": 0.572,
+      "step": 15279
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.936156751577311e-06,
+      "loss": 0.561,
+      "step": 15280
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.934915785119818e-06,
+      "loss": 0.6165,
+      "step": 15281
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.933674923572767e-06,
+      "loss": 0.5706,
+      "step": 15282
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.932434166961867e-06,
+      "loss": 0.5908,
+      "step": 15283
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.931193515312813e-06,
+      "loss": 0.5859,
+      "step": 15284
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9299529686513e-06,
+      "loss": 0.613,
+      "step": 15285
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9287125270030274e-06,
+      "loss": 0.4043,
+      "step": 15286
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.927472190393684e-06,
+      "loss": 0.4851,
+      "step": 15287
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.926231958848964e-06,
+      "loss": 0.6135,
+      "step": 15288
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.92499183239455e-06,
+      "loss": 0.5603,
+      "step": 15289
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.923751811056136e-06,
+      "loss": 0.5796,
+      "step": 15290
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.922511894859397e-06,
+      "loss": 0.551,
+      "step": 15291
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.921272083830026e-06,
+      "loss": 0.5742,
+      "step": 15292
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.920032377993689e-06,
+      "loss": 0.5752,
+      "step": 15293
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.918792777376075e-06,
+      "loss": 0.5247,
+      "step": 15294
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.91755328200285e-06,
+      "loss": 0.5474,
+      "step": 15295
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.916313891899697e-06,
+      "loss": 0.541,
+      "step": 15296
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.915074607092276e-06,
+      "loss": 0.6426,
+      "step": 15297
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.913835427606265e-06,
+      "loss": 0.5217,
+      "step": 15298
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.912596353467324e-06,
+      "loss": 0.5916,
+      "step": 15299
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.911357384701122e-06,
+      "loss": 0.5576,
+      "step": 15300
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9101185213333155e-06,
+      "loss": 0.6284,
+      "step": 15301
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.908879763389568e-06,
+      "loss": 0.5671,
+      "step": 15302
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.907641110895541e-06,
+      "loss": 0.5364,
+      "step": 15303
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.906402563876882e-06,
+      "loss": 0.6011,
+      "step": 15304
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.905164122359253e-06,
+      "loss": 0.5388,
+      "step": 15305
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.903925786368297e-06,
+      "loss": 0.5881,
+      "step": 15306
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.902687555929669e-06,
+      "loss": 0.5881,
+      "step": 15307
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.901449431069011e-06,
+      "loss": 0.5588,
+      "step": 15308
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9002114118119735e-06,
+      "loss": 0.5955,
+      "step": 15309
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.898973498184192e-06,
+      "loss": 0.5754,
+      "step": 15310
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.897735690211314e-06,
+      "loss": 0.6064,
+      "step": 15311
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.89649798791897e-06,
+      "loss": 0.6384,
+      "step": 15312
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.895260391332805e-06,
+      "loss": 0.4592,
+      "step": 15313
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.894022900478445e-06,
+      "loss": 0.6013,
+      "step": 15314
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8927855153815215e-06,
+      "loss": 0.6448,
+      "step": 15315
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.89154823606767e-06,
+      "loss": 0.5769,
+      "step": 15316
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.890311062562511e-06,
+      "loss": 0.572,
+      "step": 15317
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.889073994891676e-06,
+      "loss": 0.5347,
+      "step": 15318
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.88783703308078e-06,
+      "loss": 0.4973,
+      "step": 15319
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.886600177155451e-06,
+      "loss": 0.5828,
+      "step": 15320
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.885363427141301e-06,
+      "loss": 0.6057,
+      "step": 15321
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.884126783063952e-06,
+      "loss": 0.595,
+      "step": 15322
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.882890244949012e-06,
+      "loss": 0.5864,
+      "step": 15323
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.881653812822099e-06,
+      "loss": 0.594,
+      "step": 15324
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.880417486708816e-06,
+      "loss": 0.563,
+      "step": 15325
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.879181266634779e-06,
+      "loss": 0.5591,
+      "step": 15326
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.877945152625583e-06,
+      "loss": 0.5413,
+      "step": 15327
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.876709144706836e-06,
+      "loss": 0.5271,
+      "step": 15328
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.875473242904143e-06,
+      "loss": 0.5969,
+      "step": 15329
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.874237447243094e-06,
+      "loss": 0.6304,
+      "step": 15330
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.873001757749294e-06,
+      "loss": 0.5923,
+      "step": 15331
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8717661744483295e-06,
+      "loss": 0.5405,
+      "step": 15332
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.870530697365798e-06,
+      "loss": 0.5737,
+      "step": 15333
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.869295326527286e-06,
+      "loss": 0.5601,
+      "step": 15334
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.868060061958384e-06,
+      "loss": 0.5559,
+      "step": 15335
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.866824903684674e-06,
+      "loss": 0.5764,
+      "step": 15336
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.865589851731743e-06,
+      "loss": 0.572,
+      "step": 15337
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.864354906125167e-06,
+      "loss": 0.6733,
+      "step": 15338
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.863120066890531e-06,
+      "loss": 0.5713,
+      "step": 15339
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.861885334053404e-06,
+      "loss": 0.6204,
+      "step": 15340
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.860650707639369e-06,
+      "loss": 0.5923,
+      "step": 15341
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.859416187673991e-06,
+      "loss": 0.5591,
+      "step": 15342
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.858181774182846e-06,
+      "loss": 0.5881,
+      "step": 15343
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8569474671914984e-06,
+      "loss": 0.5068,
+      "step": 15344
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.85571326672551e-06,
+      "loss": 0.5474,
+      "step": 15345
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.854479172810452e-06,
+      "loss": 0.5415,
+      "step": 15346
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.853245185471877e-06,
+      "loss": 0.5388,
+      "step": 15347
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.852011304735353e-06,
+      "loss": 0.5898,
+      "step": 15348
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.850777530626429e-06,
+      "loss": 0.5354,
+      "step": 15349
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.849543863170667e-06,
+      "loss": 0.5815,
+      "step": 15350
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.84831030239361e-06,
+      "loss": 0.5525,
+      "step": 15351
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.847076848320814e-06,
+      "loss": 0.5903,
+      "step": 15352
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.845843500977829e-06,
+      "loss": 0.6035,
+      "step": 15353
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.844610260390196e-06,
+      "loss": 0.5833,
+      "step": 15354
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.843377126583459e-06,
+      "loss": 0.561,
+      "step": 15355
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8421440995831625e-06,
+      "loss": 0.5933,
+      "step": 15356
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8409111794148425e-06,
+      "loss": 0.5845,
+      "step": 15357
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8396783661040395e-06,
+      "loss": 0.6213,
+      "step": 15358
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.838445659676285e-06,
+      "loss": 0.5361,
+      "step": 15359
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.837213060157109e-06,
+      "loss": 0.6748,
+      "step": 15360
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.835980567572047e-06,
+      "loss": 0.6257,
+      "step": 15361
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.834748181946623e-06,
+      "loss": 0.6299,
+      "step": 15362
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.833515903306366e-06,
+      "loss": 0.5933,
+      "step": 15363
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.832283731676795e-06,
+      "loss": 0.5876,
+      "step": 15364
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.831051667083438e-06,
+      "loss": 0.5842,
+      "step": 15365
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.829819709551807e-06,
+      "loss": 0.5647,
+      "step": 15366
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.828587859107426e-06,
+      "loss": 0.5916,
+      "step": 15367
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.827356115775801e-06,
+      "loss": 0.511,
+      "step": 15368
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8261244795824535e-06,
+      "loss": 0.5806,
+      "step": 15369
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.824892950552887e-06,
+      "loss": 0.5513,
+      "step": 15370
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.823661528712614e-06,
+      "loss": 0.552,
+      "step": 15371
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.822430214087136e-06,
+      "loss": 0.5701,
+      "step": 15372
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.821199006701962e-06,
+      "loss": 0.6006,
+      "step": 15373
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.819967906582591e-06,
+      "loss": 0.5579,
+      "step": 15374
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.818736913754517e-06,
+      "loss": 0.6479,
+      "step": 15375
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.817506028243242e-06,
+      "loss": 0.5847,
+      "step": 15376
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8162752500742625e-06,
+      "loss": 0.6079,
+      "step": 15377
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.815044579273066e-06,
+      "loss": 0.5698,
+      "step": 15378
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.813814015865146e-06,
+      "loss": 0.5603,
+      "step": 15379
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.812583559875992e-06,
+      "loss": 0.6196,
+      "step": 15380
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.811353211331085e-06,
+      "loss": 0.6042,
+      "step": 15381
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.810122970255914e-06,
+      "loss": 0.6504,
+      "step": 15382
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.808892836675953e-06,
+      "loss": 0.5608,
+      "step": 15383
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.807662810616691e-06,
+      "loss": 0.521,
+      "step": 15384
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.806432892103595e-06,
+      "loss": 0.5078,
+      "step": 15385
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.805203081162147e-06,
+      "loss": 0.4656,
+      "step": 15386
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.803973377817814e-06,
+      "loss": 0.572,
+      "step": 15387
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8027437820960734e-06,
+      "loss": 0.5637,
+      "step": 15388
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.801514294022387e-06,
+      "loss": 0.5889,
+      "step": 15389
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.80028491362222e-06,
+      "loss": 0.5842,
+      "step": 15390
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7990556409210394e-06,
+      "loss": 0.6294,
+      "step": 15391
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7978264759443024e-06,
+      "loss": 0.5793,
+      "step": 15392
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7965974187174756e-06,
+      "loss": 0.571,
+      "step": 15393
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.795368469266006e-06,
+      "loss": 0.5762,
+      "step": 15394
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.794139627615357e-06,
+      "loss": 0.5981,
+      "step": 15395
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.792910893790974e-06,
+      "loss": 0.5847,
+      "step": 15396
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.791682267818314e-06,
+      "loss": 0.5771,
+      "step": 15397
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.790453749722816e-06,
+      "loss": 0.5425,
+      "step": 15398
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.789225339529936e-06,
+      "loss": 0.4897,
+      "step": 15399
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7879970372651085e-06,
+      "loss": 0.626,
+      "step": 15400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.786768842953782e-06,
+      "loss": 0.5789,
+      "step": 15401
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.785540756621388e-06,
+      "loss": 0.5076,
+      "step": 15402
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7843127782933695e-06,
+      "loss": 0.594,
+      "step": 15403
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.783084907995156e-06,
+      "loss": 0.5518,
+      "step": 15404
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.781857145752182e-06,
+      "loss": 0.5903,
+      "step": 15405
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.780629491589881e-06,
+      "loss": 0.552,
+      "step": 15406
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.779401945533675e-06,
+      "loss": 0.522,
+      "step": 15407
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.778174507608995e-06,
+      "loss": 0.5769,
+      "step": 15408
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.776947177841258e-06,
+      "loss": 0.6282,
+      "step": 15409
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.775719956255891e-06,
+      "loss": 0.4734,
+      "step": 15410
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.774492842878308e-06,
+      "loss": 0.5225,
+      "step": 15411
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7732658377339305e-06,
+      "loss": 0.5557,
+      "step": 15412
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.772038940848166e-06,
+      "loss": 0.5171,
+      "step": 15413
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.770812152246436e-06,
+      "loss": 0.5134,
+      "step": 15414
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.769585471954139e-06,
+      "loss": 0.5989,
+      "step": 15415
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7683588999966935e-06,
+      "loss": 0.5823,
+      "step": 15416
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7671324363995e-06,
+      "loss": 0.6353,
+      "step": 15417
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.765906081187957e-06,
+      "loss": 0.5337,
+      "step": 15418
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.764679834387474e-06,
+      "loss": 0.5569,
+      "step": 15419
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7634536960234424e-06,
+      "loss": 0.6528,
+      "step": 15420
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.762227666121264e-06,
+      "loss": 0.5906,
+      "step": 15421
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7610017447063274e-06,
+      "loss": 0.5356,
+      "step": 15422
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.759775931804031e-06,
+      "loss": 0.5151,
+      "step": 15423
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.758550227439759e-06,
+      "loss": 0.6406,
+      "step": 15424
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.757324631638902e-06,
+      "loss": 0.5537,
+      "step": 15425
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.756099144426842e-06,
+      "loss": 0.5242,
+      "step": 15426
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.754873765828967e-06,
+      "loss": 0.5603,
+      "step": 15427
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.753648495870651e-06,
+      "loss": 0.5869,
+      "step": 15428
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7524233345772756e-06,
+      "loss": 0.6331,
+      "step": 15429
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7511982819742195e-06,
+      "loss": 0.5422,
+      "step": 15430
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.749973338086852e-06,
+      "loss": 0.6802,
+      "step": 15431
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.74874850294055e-06,
+      "loss": 0.5586,
+      "step": 15432
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.747523776560676e-06,
+      "loss": 0.6128,
+      "step": 15433
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.746299158972606e-06,
+      "loss": 0.5535,
+      "step": 15434
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.745074650201695e-06,
+      "loss": 0.6069,
+      "step": 15435
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.743850250273314e-06,
+      "loss": 0.6111,
+      "step": 15436
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.742625959212816e-06,
+      "loss": 0.5583,
+      "step": 15437
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.741401777045567e-06,
+      "loss": 0.6089,
+      "step": 15438
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.740177703796914e-06,
+      "loss": 0.5417,
+      "step": 15439
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.73895373949222e-06,
+      "loss": 0.5994,
+      "step": 15440
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.737729884156829e-06,
+      "loss": 0.5776,
+      "step": 15441
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.736506137816095e-06,
+      "loss": 0.5767,
+      "step": 15442
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.735282500495358e-06,
+      "loss": 0.6606,
+      "step": 15443
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.734058972219972e-06,
+      "loss": 0.5903,
+      "step": 15444
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.732835553015271e-06,
+      "loss": 0.5686,
+      "step": 15445
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.731612242906601e-06,
+      "loss": 0.5928,
+      "step": 15446
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7303890419192976e-06,
+      "loss": 0.5625,
+      "step": 15447
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.729165950078693e-06,
+      "loss": 0.5715,
+      "step": 15448
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.727942967410126e-06,
+      "loss": 0.5349,
+      "step": 15449
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7267200939389216e-06,
+      "loss": 0.5793,
+      "step": 15450
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.725497329690416e-06,
+      "loss": 0.5457,
+      "step": 15451
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.724274674689927e-06,
+      "loss": 0.5889,
+      "step": 15452
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.723052128962784e-06,
+      "loss": 0.5486,
+      "step": 15453
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7218296925343134e-06,
+      "loss": 0.5002,
+      "step": 15454
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.720607365429825e-06,
+      "loss": 0.603,
+      "step": 15455
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.71938514767464e-06,
+      "loss": 0.5586,
+      "step": 15456
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.71816303929408e-06,
+      "loss": 0.5381,
+      "step": 15457
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.716941040313448e-06,
+      "loss": 0.5449,
+      "step": 15458
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.715719150758064e-06,
+      "loss": 0.5142,
+      "step": 15459
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.714497370653226e-06,
+      "loss": 0.5286,
+      "step": 15460
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.71327570002425e-06,
+      "loss": 0.5967,
+      "step": 15461
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.712054138896436e-06,
+      "loss": 0.6262,
+      "step": 15462
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.710832687295081e-06,
+      "loss": 0.5398,
+      "step": 15463
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.709611345245493e-06,
+      "loss": 0.6328,
+      "step": 15464
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.708390112772959e-06,
+      "loss": 0.5845,
+      "step": 15465
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.707168989902784e-06,
+      "loss": 0.5063,
+      "step": 15466
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.705947976660251e-06,
+      "loss": 0.6218,
+      "step": 15467
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.704727073070658e-06,
+      "loss": 0.5085,
+      "step": 15468
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7035062791592875e-06,
+      "loss": 0.5576,
+      "step": 15469
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7022855949514305e-06,
+      "loss": 0.479,
+      "step": 15470
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.701065020472363e-06,
+      "loss": 0.6213,
+      "step": 15471
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.699844555747375e-06,
+      "loss": 0.563,
+      "step": 15472
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.698624200801737e-06,
+      "loss": 0.5647,
+      "step": 15473
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6974039556607325e-06,
+      "loss": 0.5732,
+      "step": 15474
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.696183820349629e-06,
+      "loss": 0.5391,
+      "step": 15475
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.694963794893706e-06,
+      "loss": 0.5908,
+      "step": 15476
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6937438793182244e-06,
+      "loss": 0.574,
+      "step": 15477
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.692524073648462e-06,
+      "loss": 0.4895,
+      "step": 15478
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.691304377909675e-06,
+      "loss": 0.5527,
+      "step": 15479
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.690084792127132e-06,
+      "loss": 0.5486,
+      "step": 15480
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.68886531632609e-06,
+      "loss": 0.5713,
+      "step": 15481
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.687645950531807e-06,
+      "loss": 0.5859,
+      "step": 15482
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.686426694769546e-06,
+      "loss": 0.5564,
+      "step": 15483
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.685207549064552e-06,
+      "loss": 0.5369,
+      "step": 15484
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.683988513442085e-06,
+      "loss": 0.583,
+      "step": 15485
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.682769587927385e-06,
+      "loss": 0.5955,
+      "step": 15486
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.681550772545707e-06,
+      "loss": 0.5352,
+      "step": 15487
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.680332067322289e-06,
+      "loss": 0.5642,
+      "step": 15488
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.679113472282381e-06,
+      "loss": 0.5444,
+      "step": 15489
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.677894987451214e-06,
+      "loss": 0.6184,
+      "step": 15490
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.676676612854035e-06,
+      "loss": 0.4844,
+      "step": 15491
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.675458348516076e-06,
+      "loss": 0.6162,
+      "step": 15492
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.674240194462565e-06,
+      "loss": 0.5369,
+      "step": 15493
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.67302215071874e-06,
+      "loss": 0.5637,
+      "step": 15494
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6718042173098254e-06,
+      "loss": 0.5955,
+      "step": 15495
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.670586394261053e-06,
+      "loss": 0.6125,
+      "step": 15496
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.669368681597639e-06,
+      "loss": 0.5249,
+      "step": 15497
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.668151079344814e-06,
+      "loss": 0.6592,
+      "step": 15498
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.666933587527788e-06,
+      "loss": 0.5503,
+      "step": 15499
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.665716206171789e-06,
+      "loss": 0.5503,
+      "step": 15500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.664498935302021e-06,
+      "loss": 0.5474,
+      "step": 15501
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6632817749437066e-06,
+      "loss": 0.5742,
+      "step": 15502
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.662064725122049e-06,
+      "loss": 0.5493,
+      "step": 15503
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.660847785862261e-06,
+      "loss": 0.5732,
+      "step": 15504
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.659630957189544e-06,
+      "loss": 0.6299,
+      "step": 15505
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.658414239129103e-06,
+      "loss": 0.606,
+      "step": 15506
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.657197631706145e-06,
+      "loss": 0.5212,
+      "step": 15507
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.655981134945861e-06,
+      "loss": 0.5449,
+      "step": 15508
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.654764748873452e-06,
+      "loss": 0.5312,
+      "step": 15509
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6535484735141104e-06,
+      "loss": 0.5042,
+      "step": 15510
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.652332308893031e-06,
+      "loss": 0.6616,
+      "step": 15511
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6511162550354e-06,
+      "loss": 0.5764,
+      "step": 15512
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.649900311966408e-06,
+      "loss": 0.5652,
+      "step": 15513
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.648684479711236e-06,
+      "loss": 0.6458,
+      "step": 15514
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.647468758295074e-06,
+      "loss": 0.6294,
+      "step": 15515
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.646253147743094e-06,
+      "loss": 0.5774,
+      "step": 15516
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.645037648080482e-06,
+      "loss": 0.5161,
+      "step": 15517
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.643822259332408e-06,
+      "loss": 0.5906,
+      "step": 15518
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.64260698152405e-06,
+      "loss": 0.6074,
+      "step": 15519
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.641391814680576e-06,
+      "loss": 0.5437,
+      "step": 15520
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6401767588271605e-06,
+      "loss": 0.5398,
+      "step": 15521
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6389618139889654e-06,
+      "loss": 0.6699,
+      "step": 15522
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6377469801911536e-06,
+      "loss": 0.5583,
+      "step": 15523
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6365322574588945e-06,
+      "loss": 0.6047,
+      "step": 15524
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.635317645817338e-06,
+      "loss": 0.6038,
+      "step": 15525
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.634103145291653e-06,
+      "loss": 0.5664,
+      "step": 15526
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.632888755906986e-06,
+      "loss": 0.5222,
+      "step": 15527
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.631674477688496e-06,
+      "loss": 0.5439,
+      "step": 15528
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.630460310661326e-06,
+      "loss": 0.4934,
+      "step": 15529
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.629246254850631e-06,
+      "loss": 0.5708,
+      "step": 15530
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.62803231028156e-06,
+      "loss": 0.5723,
+      "step": 15531
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.626818476979247e-06,
+      "loss": 0.4993,
+      "step": 15532
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.625604754968839e-06,
+      "loss": 0.5862,
+      "step": 15533
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.624391144275478e-06,
+      "loss": 0.5437,
+      "step": 15534
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.623177644924296e-06,
+      "loss": 0.5508,
+      "step": 15535
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.621964256940432e-06,
+      "loss": 0.6445,
+      "step": 15536
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.620750980349017e-06,
+      "loss": 0.5547,
+      "step": 15537
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.619537815175175e-06,
+      "loss": 0.5508,
+      "step": 15538
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.618324761444042e-06,
+      "loss": 0.5129,
+      "step": 15539
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.617111819180735e-06,
+      "loss": 0.5925,
+      "step": 15540
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.615898988410387e-06,
+      "loss": 0.5269,
+      "step": 15541
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.61468626915811e-06,
+      "loss": 0.6167,
+      "step": 15542
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.613473661449028e-06,
+      "loss": 0.5393,
+      "step": 15543
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6122611653082526e-06,
+      "loss": 0.6177,
+      "step": 15544
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.611048780760903e-06,
+      "loss": 0.5894,
+      "step": 15545
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6098365078320835e-06,
+      "loss": 0.5767,
+      "step": 15546
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6086243465469125e-06,
+      "loss": 0.5852,
+      "step": 15547
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.607412296930487e-06,
+      "loss": 0.5684,
+      "step": 15548
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.606200359007921e-06,
+      "loss": 0.5723,
+      "step": 15549
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.604988532804307e-06,
+      "loss": 0.5435,
+      "step": 15550
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.603776818344755e-06,
+      "loss": 0.5598,
+      "step": 15551
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.602565215654358e-06,
+      "loss": 0.5752,
+      "step": 15552
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.601353724758206e-06,
+      "loss": 0.5745,
+      "step": 15553
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.600142345681397e-06,
+      "loss": 0.5735,
+      "step": 15554
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.598931078449027e-06,
+      "loss": 0.5596,
+      "step": 15555
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.597719923086173e-06,
+      "loss": 0.5769,
+      "step": 15556
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.596508879617932e-06,
+      "loss": 0.6016,
+      "step": 15557
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.595297948069379e-06,
+      "loss": 0.5659,
+      "step": 15558
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5940871284655984e-06,
+      "loss": 0.5051,
+      "step": 15559
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.592876420831675e-06,
+      "loss": 0.5935,
+      "step": 15560
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.591665825192675e-06,
+      "loss": 0.5322,
+      "step": 15561
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.590455341573683e-06,
+      "loss": 0.6299,
+      "step": 15562
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.589244969999762e-06,
+      "loss": 0.5986,
+      "step": 15563
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.588034710495991e-06,
+      "loss": 0.6033,
+      "step": 15564
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.586824563087428e-06,
+      "loss": 0.6208,
+      "step": 15565
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.585614527799147e-06,
+      "loss": 0.5464,
+      "step": 15566
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.584404604656207e-06,
+      "loss": 0.6267,
+      "step": 15567
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.583194793683664e-06,
+      "loss": 0.5544,
+      "step": 15568
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.581985094906585e-06,
+      "loss": 0.5469,
+      "step": 15569
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.580775508350016e-06,
+      "loss": 0.6011,
+      "step": 15570
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.57956603403902e-06,
+      "loss": 0.6567,
+      "step": 15571
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.578356671998642e-06,
+      "loss": 0.5288,
+      "step": 15572
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.577147422253935e-06,
+      "loss": 0.603,
+      "step": 15573
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.575938284829939e-06,
+      "loss": 0.4951,
+      "step": 15574
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.574729259751708e-06,
+      "loss": 0.6204,
+      "step": 15575
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.573520347044273e-06,
+      "loss": 0.5632,
+      "step": 15576
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.572311546732685e-06,
+      "loss": 0.5864,
+      "step": 15577
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.57110285884197e-06,
+      "loss": 0.5605,
+      "step": 15578
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.569894283397173e-06,
+      "loss": 0.5798,
+      "step": 15579
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.568685820423317e-06,
+      "loss": 0.5439,
+      "step": 15580
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.567477469945442e-06,
+      "loss": 0.5566,
+      "step": 15581
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5662692319885684e-06,
+      "loss": 0.615,
+      "step": 15582
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.565061106577723e-06,
+      "loss": 0.5618,
+      "step": 15583
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.563853093737935e-06,
+      "loss": 0.5771,
+      "step": 15584
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.562645193494216e-06,
+      "loss": 0.5425,
+      "step": 15585
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.561437405871594e-06,
+      "loss": 0.5801,
+      "step": 15586
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5602297308950775e-06,
+      "loss": 0.5986,
+      "step": 15587
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.559022168589686e-06,
+      "loss": 0.5779,
+      "step": 15588
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.557814718980427e-06,
+      "loss": 0.4954,
+      "step": 15589
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.556607382092314e-06,
+      "loss": 0.5405,
+      "step": 15590
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.555400157950349e-06,
+      "loss": 0.5708,
+      "step": 15591
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.554193046579542e-06,
+      "loss": 0.5146,
+      "step": 15592
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5529860480048896e-06,
+      "loss": 0.5669,
+      "step": 15593
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5517791622513975e-06,
+      "loss": 0.5552,
+      "step": 15594
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.550572389344057e-06,
+      "loss": 0.5085,
+      "step": 15595
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.54936572930787e-06,
+      "loss": 0.5229,
+      "step": 15596
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.548159182167826e-06,
+      "loss": 0.6418,
+      "step": 15597
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.546952747948913e-06,
+      "loss": 0.5889,
+      "step": 15598
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.545746426676125e-06,
+      "loss": 0.582,
+      "step": 15599
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5445402183744405e-06,
+      "loss": 0.5525,
+      "step": 15600
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.543334123068852e-06,
+      "loss": 0.572,
+      "step": 15601
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.542128140784332e-06,
+      "loss": 0.627,
+      "step": 15602
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.540922271545868e-06,
+      "loss": 0.6045,
+      "step": 15603
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.539716515378427e-06,
+      "loss": 0.5803,
+      "step": 15604
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.538510872306994e-06,
+      "loss": 0.5835,
+      "step": 15605
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5373053423565296e-06,
+      "loss": 0.5957,
+      "step": 15606
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.53609992555201e-06,
+      "loss": 0.6914,
+      "step": 15607
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.534894621918404e-06,
+      "loss": 0.5598,
+      "step": 15608
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5336894314806715e-06,
+      "loss": 0.6045,
+      "step": 15609
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.532484354263775e-06,
+      "loss": 0.604,
+      "step": 15610
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.531279390292682e-06,
+      "loss": 0.5247,
+      "step": 15611
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.530074539592344e-06,
+      "loss": 0.4988,
+      "step": 15612
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.528869802187714e-06,
+      "loss": 0.5786,
+      "step": 15613
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.527665178103753e-06,
+      "loss": 0.5728,
+      "step": 15614
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.526460667365404e-06,
+      "loss": 0.5823,
+      "step": 15615
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.525256269997621e-06,
+      "loss": 0.5481,
+      "step": 15616
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.524051986025344e-06,
+      "loss": 0.6108,
+      "step": 15617
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.522847815473526e-06,
+      "loss": 0.5454,
+      "step": 15618
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.521643758367097e-06,
+      "loss": 0.541,
+      "step": 15619
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.520439814731007e-06,
+      "loss": 0.5676,
+      "step": 15620
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.519235984590184e-06,
+      "loss": 0.6411,
+      "step": 15621
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.518032267969567e-06,
+      "loss": 0.5833,
+      "step": 15622
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.516828664894086e-06,
+      "loss": 0.5713,
+      "step": 15623
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.515625175388674e-06,
+      "loss": 0.5894,
+      "step": 15624
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5144217994782555e-06,
+      "loss": 0.4773,
+      "step": 15625
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5132185371877525e-06,
+      "loss": 0.5054,
+      "step": 15626
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.512015388542094e-06,
+      "loss": 0.5688,
+      "step": 15627
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.510812353566193e-06,
+      "loss": 0.5537,
+      "step": 15628
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5096094322849735e-06,
+      "loss": 0.6367,
+      "step": 15629
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.508406624723346e-06,
+      "loss": 0.6067,
+      "step": 15630
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5072039309062265e-06,
+      "loss": 0.5266,
+      "step": 15631
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.506001350858529e-06,
+      "loss": 0.6582,
+      "step": 15632
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.504798884605155e-06,
+      "loss": 0.5674,
+      "step": 15633
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.503596532171016e-06,
+      "loss": 0.6282,
+      "step": 15634
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.502394293581012e-06,
+      "loss": 0.5786,
+      "step": 15635
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.501192168860045e-06,
+      "loss": 0.6257,
+      "step": 15636
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.499990158033018e-06,
+      "loss": 0.5427,
+      "step": 15637
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.498788261124823e-06,
+      "loss": 0.5503,
+      "step": 15638
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.497586478160357e-06,
+      "loss": 0.6377,
+      "step": 15639
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4963848091645125e-06,
+      "loss": 0.5876,
+      "step": 15640
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.495183254162173e-06,
+      "loss": 0.5242,
+      "step": 15641
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.493981813178233e-06,
+      "loss": 0.564,
+      "step": 15642
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.492780486237571e-06,
+      "loss": 0.5854,
+      "step": 15643
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.491579273365078e-06,
+      "loss": 0.5254,
+      "step": 15644
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.490378174585623e-06,
+      "loss": 0.4875,
+      "step": 15645
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.489177189924095e-06,
+      "loss": 0.4897,
+      "step": 15646
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.487976319405358e-06,
+      "loss": 0.5291,
+      "step": 15647
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.486775563054296e-06,
+      "loss": 0.5605,
+      "step": 15648
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.485574920895771e-06,
+      "loss": 0.574,
+      "step": 15649
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.484374392954659e-06,
+      "loss": 0.5857,
+      "step": 15650
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.483173979255817e-06,
+      "loss": 0.5242,
+      "step": 15651
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.481973679824118e-06,
+      "loss": 0.5874,
+      "step": 15652
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.480773494684414e-06,
+      "loss": 0.5706,
+      "step": 15653
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.479573423861573e-06,
+      "loss": 0.5728,
+      "step": 15654
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4783734673804455e-06,
+      "loss": 0.5544,
+      "step": 15655
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.477173625265884e-06,
+      "loss": 0.5688,
+      "step": 15656
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.475973897542743e-06,
+      "loss": 0.5452,
+      "step": 15657
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.474774284235874e-06,
+      "loss": 0.5012,
+      "step": 15658
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.47357478537012e-06,
+      "loss": 0.5579,
+      "step": 15659
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4723754009703265e-06,
+      "loss": 0.5918,
+      "step": 15660
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.471176131061341e-06,
+      "loss": 0.6343,
+      "step": 15661
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.469976975667994e-06,
+      "loss": 0.5908,
+      "step": 15662
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.468777934815133e-06,
+      "loss": 0.5518,
+      "step": 15663
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.467579008527584e-06,
+      "loss": 0.5945,
+      "step": 15664
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.466380196830187e-06,
+      "loss": 0.6128,
+      "step": 15665
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.465181499747766e-06,
+      "loss": 0.5415,
+      "step": 15666
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.463982917305155e-06,
+      "loss": 0.5833,
+      "step": 15667
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.462784449527174e-06,
+      "loss": 0.5652,
+      "step": 15668
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.461586096438652e-06,
+      "loss": 0.584,
+      "step": 15669
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.460387858064408e-06,
+      "loss": 0.5496,
+      "step": 15670
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.459189734429256e-06,
+      "loss": 0.5793,
+      "step": 15671
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4579917255580176e-06,
+      "loss": 0.5681,
+      "step": 15672
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.456793831475501e-06,
+      "loss": 0.5667,
+      "step": 15673
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.455596052206524e-06,
+      "loss": 0.5962,
+      "step": 15674
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.45439838777589e-06,
+      "loss": 0.5601,
+      "step": 15675
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.453200838208412e-06,
+      "loss": 0.6204,
+      "step": 15676
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.452003403528886e-06,
+      "loss": 0.6272,
+      "step": 15677
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.450806083762123e-06,
+      "loss": 0.5159,
+      "step": 15678
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.449608878932913e-06,
+      "loss": 0.6025,
+      "step": 15679
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.448411789066061e-06,
+      "loss": 0.5352,
+      "step": 15680
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.447214814186355e-06,
+      "loss": 0.5283,
+      "step": 15681
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4460179543185945e-06,
+      "loss": 0.5015,
+      "step": 15682
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.444821209487562e-06,
+      "loss": 0.5059,
+      "step": 15683
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.443624579718049e-06,
+      "loss": 0.5698,
+      "step": 15684
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.442428065034844e-06,
+      "loss": 0.5159,
+      "step": 15685
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.441231665462722e-06,
+      "loss": 0.5266,
+      "step": 15686
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.440035381026472e-06,
+      "loss": 0.5317,
+      "step": 15687
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.438839211750863e-06,
+      "loss": 0.5576,
+      "step": 15688
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.437643157660679e-06,
+      "loss": 0.5525,
+      "step": 15689
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.436447218780686e-06,
+      "loss": 0.6116,
+      "step": 15690
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.435251395135662e-06,
+      "loss": 0.5242,
+      "step": 15691
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4340556867503696e-06,
+      "loss": 0.6111,
+      "step": 15692
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.432860093649578e-06,
+      "loss": 0.5623,
+      "step": 15693
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.431664615858048e-06,
+      "loss": 0.4824,
+      "step": 15694
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.430469253400547e-06,
+      "loss": 0.4978,
+      "step": 15695
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.429274006301825e-06,
+      "loss": 0.5762,
+      "step": 15696
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.428078874586648e-06,
+      "loss": 0.6533,
+      "step": 15697
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.42688385827976e-06,
+      "loss": 0.6438,
+      "step": 15698
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.425688957405924e-06,
+      "loss": 0.5979,
+      "step": 15699
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4244941719898815e-06,
+      "loss": 0.5681,
+      "step": 15700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.42329950205638e-06,
+      "loss": 0.6062,
+      "step": 15701
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.422104947630168e-06,
+      "loss": 0.5232,
+      "step": 15702
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.420910508735982e-06,
+      "loss": 0.5439,
+      "step": 15703
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.419716185398568e-06,
+      "loss": 0.5925,
+      "step": 15704
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.418521977642658e-06,
+      "loss": 0.6067,
+      "step": 15705
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.417327885492992e-06,
+      "loss": 0.5649,
+      "step": 15706
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.416133908974297e-06,
+      "loss": 0.6311,
+      "step": 15707
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.414940048111307e-06,
+      "loss": 0.5355,
+      "step": 15708
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4137463029287515e-06,
+      "loss": 0.5608,
+      "step": 15709
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.412552673451351e-06,
+      "loss": 0.6001,
+      "step": 15710
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.411359159703834e-06,
+      "loss": 0.5427,
+      "step": 15711
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.410165761710915e-06,
+      "loss": 0.5615,
+      "step": 15712
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.408972479497315e-06,
+      "loss": 0.5354,
+      "step": 15713
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4077793130877545e-06,
+      "loss": 0.5405,
+      "step": 15714
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.406586262506942e-06,
+      "loss": 0.5212,
+      "step": 15715
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.405393327779586e-06,
+      "loss": 0.6311,
+      "step": 15716
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.404200508930402e-06,
+      "loss": 0.5686,
+      "step": 15717
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.403007805984088e-06,
+      "loss": 0.5537,
+      "step": 15718
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.401815218965357e-06,
+      "loss": 0.5625,
+      "step": 15719
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.400622747898903e-06,
+      "loss": 0.5791,
+      "step": 15720
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.39943039280943e-06,
+      "loss": 0.5051,
+      "step": 15721
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.398238153721629e-06,
+      "loss": 0.5901,
+      "step": 15722
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.397046030660203e-06,
+      "loss": 0.6189,
+      "step": 15723
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.395854023649833e-06,
+      "loss": 0.5769,
+      "step": 15724
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.394662132715218e-06,
+      "loss": 0.5933,
+      "step": 15725
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.393470357881038e-06,
+      "loss": 0.4797,
+      "step": 15726
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.392278699171984e-06,
+      "loss": 0.5432,
+      "step": 15727
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.391087156612731e-06,
+      "loss": 0.5913,
+      "step": 15728
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.389895730227967e-06,
+      "loss": 0.6169,
+      "step": 15729
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.388704420042365e-06,
+      "loss": 0.5474,
+      "step": 15730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3875132260805954e-06,
+      "loss": 0.53,
+      "step": 15731
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.386322148367339e-06,
+      "loss": 0.5796,
+      "step": 15732
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.38513118692726e-06,
+      "loss": 0.5469,
+      "step": 15733
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.383940341785027e-06,
+      "loss": 0.6401,
+      "step": 15734
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.382749612965311e-06,
+      "loss": 0.5974,
+      "step": 15735
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.381559000492767e-06,
+      "loss": 0.4836,
+      "step": 15736
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.38036850439206e-06,
+      "loss": 0.582,
+      "step": 15737
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.37917812468785e-06,
+      "loss": 0.6545,
+      "step": 15738
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.377987861404788e-06,
+      "loss": 0.5186,
+      "step": 15739
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3767977145675325e-06,
+      "loss": 0.5911,
+      "step": 15740
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.375607684200728e-06,
+      "loss": 0.5527,
+      "step": 15741
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.374417770329029e-06,
+      "loss": 0.5991,
+      "step": 15742
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.373227972977076e-06,
+      "loss": 0.562,
+      "step": 15743
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3720382921695195e-06,
+      "loss": 0.6333,
+      "step": 15744
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.370848727930997e-06,
+      "loss": 0.5105,
+      "step": 15745
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.369659280286142e-06,
+      "loss": 0.5632,
+      "step": 15746
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3684699492596e-06,
+      "loss": 0.5081,
+      "step": 15747
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.367280734875998e-06,
+      "loss": 0.5913,
+      "step": 15748
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3660916371599745e-06,
+      "loss": 0.5405,
+      "step": 15749
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.364902656136151e-06,
+      "loss": 0.6047,
+      "step": 15750
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.363713791829161e-06,
+      "loss": 0.6204,
+      "step": 15751
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.362525044263624e-06,
+      "loss": 0.5999,
+      "step": 15752
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3613364134641655e-06,
+      "loss": 0.6104,
+      "step": 15753
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.360147899455399e-06,
+      "loss": 0.5208,
+      "step": 15754
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.358959502261952e-06,
+      "loss": 0.5483,
+      "step": 15755
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3577712219084264e-06,
+      "loss": 0.5396,
+      "step": 15756
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3565830584194466e-06,
+      "loss": 0.531,
+      "step": 15757
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.355395011819613e-06,
+      "loss": 0.5337,
+      "step": 15758
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3542070821335405e-06,
+      "loss": 0.6191,
+      "step": 15759
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.353019269385825e-06,
+      "loss": 0.5427,
+      "step": 15760
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.351831573601076e-06,
+      "loss": 0.5825,
+      "step": 15761
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.350643994803896e-06,
+      "loss": 0.5557,
+      "step": 15762
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.349456533018875e-06,
+      "loss": 0.5903,
+      "step": 15763
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.348269188270614e-06,
+      "loss": 0.4949,
+      "step": 15764
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3470819605837015e-06,
+      "loss": 0.4773,
+      "step": 15765
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.345894849982734e-06,
+      "loss": 0.5439,
+      "step": 15766
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.344707856492292e-06,
+      "loss": 0.5168,
+      "step": 15767
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.343520980136967e-06,
+      "loss": 0.5471,
+      "step": 15768
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.342334220941337e-06,
+      "loss": 0.5737,
+      "step": 15769
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.341147578929992e-06,
+      "loss": 0.6211,
+      "step": 15770
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3399610541274974e-06,
+      "loss": 0.5134,
+      "step": 15771
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.33877464655844e-06,
+      "loss": 0.5249,
+      "step": 15772
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3375883562473864e-06,
+      "loss": 0.511,
+      "step": 15773
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.336402183218913e-06,
+      "loss": 0.4651,
+      "step": 15774
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3352161274975854e-06,
+      "loss": 0.5708,
+      "step": 15775
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.334030189107966e-06,
+      "loss": 0.5608,
+      "step": 15776
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.332844368074626e-06,
+      "loss": 0.5833,
+      "step": 15777
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.33165866442212e-06,
+      "loss": 0.5205,
+      "step": 15778
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.330473078175015e-06,
+      "loss": 0.5334,
+      "step": 15779
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.329287609357856e-06,
+      "loss": 0.5337,
+      "step": 15780
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.328102257995209e-06,
+      "loss": 0.5588,
+      "step": 15781
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.326917024111615e-06,
+      "loss": 0.644,
+      "step": 15782
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.325731907731633e-06,
+      "loss": 0.5498,
+      "step": 15783
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3245469088798e-06,
+      "loss": 0.571,
+      "step": 15784
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.323362027580665e-06,
+      "loss": 0.5645,
+      "step": 15785
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.322177263858775e-06,
+      "loss": 0.6086,
+      "step": 15786
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.32099261773866e-06,
+      "loss": 0.6001,
+      "step": 15787
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.319808089244863e-06,
+      "loss": 0.6548,
+      "step": 15788
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.318623678401916e-06,
+      "loss": 0.4705,
+      "step": 15789
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.317439385234353e-06,
+      "loss": 0.5618,
+      "step": 15790
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3162552097667e-06,
+      "loss": 0.606,
+      "step": 15791
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.31507115202349e-06,
+      "loss": 0.521,
+      "step": 15792
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3138872120292405e-06,
+      "loss": 0.6157,
+      "step": 15793
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3127033898084825e-06,
+      "loss": 0.5515,
+      "step": 15794
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.311519685385726e-06,
+      "loss": 0.6023,
+      "step": 15795
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.310336098785499e-06,
+      "loss": 0.5896,
+      "step": 15796
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.309152630032307e-06,
+      "loss": 0.6416,
+      "step": 15797
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.30796927915067e-06,
+      "loss": 0.5925,
+      "step": 15798
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.306786046165091e-06,
+      "loss": 0.5344,
+      "step": 15799
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.305602931100086e-06,
+      "loss": 0.5857,
+      "step": 15800
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.304419933980151e-06,
+      "loss": 0.5334,
+      "step": 15801
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.303237054829797e-06,
+      "loss": 0.593,
+      "step": 15802
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.302054293673519e-06,
+      "loss": 0.5901,
+      "step": 15803
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3008716505358195e-06,
+      "loss": 0.5696,
+      "step": 15804
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.299689125441191e-06,
+      "loss": 0.5371,
+      "step": 15805
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2985067184141225e-06,
+      "loss": 0.604,
+      "step": 15806
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.297324429479114e-06,
+      "loss": 0.5513,
+      "step": 15807
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.296142258660645e-06,
+      "loss": 0.6294,
+      "step": 15808
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.294960205983208e-06,
+      "loss": 0.554,
+      "step": 15809
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.293778271471278e-06,
+      "loss": 0.6572,
+      "step": 15810
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.292596455149343e-06,
+      "loss": 0.5671,
+      "step": 15811
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.291414757041881e-06,
+      "loss": 0.5679,
+      "step": 15812
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.290233177173363e-06,
+      "loss": 0.5476,
+      "step": 15813
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2890517155682655e-06,
+      "loss": 0.5464,
+      "step": 15814
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.287870372251063e-06,
+      "loss": 0.5974,
+      "step": 15815
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.286689147246217e-06,
+      "loss": 0.5793,
+      "step": 15816
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.285508040578201e-06,
+      "loss": 0.5269,
+      "step": 15817
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.284327052271471e-06,
+      "loss": 0.5784,
+      "step": 15818
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.283146182350495e-06,
+      "loss": 0.6365,
+      "step": 15819
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.28196543083973e-06,
+      "loss": 0.4854,
+      "step": 15820
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.280784797763625e-06,
+      "loss": 0.5525,
+      "step": 15821
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.279604283146645e-06,
+      "loss": 0.5455,
+      "step": 15822
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2784238870132335e-06,
+      "loss": 0.5652,
+      "step": 15823
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.277243609387844e-06,
+      "loss": 0.5876,
+      "step": 15824
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.276063450294918e-06,
+      "loss": 0.562,
+      "step": 15825
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2748834097589066e-06,
+      "loss": 0.5366,
+      "step": 15826
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.273703487804244e-06,
+      "loss": 0.5095,
+      "step": 15827
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.272523684455375e-06,
+      "loss": 0.5498,
+      "step": 15828
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.271343999736732e-06,
+      "loss": 0.5527,
+      "step": 15829
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.270164433672754e-06,
+      "loss": 0.5298,
+      "step": 15830
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2689849862878655e-06,
+      "loss": 0.5256,
+      "step": 15831
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.267805657606504e-06,
+      "loss": 0.5425,
+      "step": 15832
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.266626447653093e-06,
+      "loss": 0.6475,
+      "step": 15833
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.265447356452052e-06,
+      "loss": 0.5854,
+      "step": 15834
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.264268384027809e-06,
+      "loss": 0.5696,
+      "step": 15835
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.263089530404783e-06,
+      "loss": 0.5476,
+      "step": 15836
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.261910795607388e-06,
+      "loss": 0.6062,
+      "step": 15837
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.260732179660039e-06,
+      "loss": 0.5374,
+      "step": 15838
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.259553682587153e-06,
+      "loss": 0.5281,
+      "step": 15839
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.258375304413133e-06,
+      "loss": 0.6704,
+      "step": 15840
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.257197045162391e-06,
+      "loss": 0.5996,
+      "step": 15841
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.256018904859327e-06,
+      "loss": 0.5876,
+      "step": 15842
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.254840883528349e-06,
+      "loss": 0.6333,
+      "step": 15843
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.253662981193851e-06,
+      "loss": 0.5247,
+      "step": 15844
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2524851978802366e-06,
+      "loss": 0.5352,
+      "step": 15845
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.251307533611891e-06,
+      "loss": 0.5349,
+      "step": 15846
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.250129988413218e-06,
+      "loss": 0.5938,
+      "step": 15847
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.248952562308601e-06,
+      "loss": 0.5083,
+      "step": 15848
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.247775255322425e-06,
+      "loss": 0.615,
+      "step": 15849
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.246598067479082e-06,
+      "loss": 0.5835,
+      "step": 15850
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.245420998802947e-06,
+      "loss": 0.5801,
+      "step": 15851
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2442440493184064e-06,
+      "loss": 0.616,
+      "step": 15852
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.243067219049833e-06,
+      "loss": 0.5515,
+      "step": 15853
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.241890508021607e-06,
+      "loss": 0.6401,
+      "step": 15854
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.240713916258094e-06,
+      "loss": 0.5391,
+      "step": 15855
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2395374437836726e-06,
+      "loss": 0.5559,
+      "step": 15856
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.238361090622701e-06,
+      "loss": 0.5959,
+      "step": 15857
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.237184856799553e-06,
+      "loss": 0.6646,
+      "step": 15858
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.236008742338584e-06,
+      "loss": 0.6248,
+      "step": 15859
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2348327472641625e-06,
+      "loss": 0.5742,
+      "step": 15860
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2336568716006376e-06,
+      "loss": 0.51,
+      "step": 15861
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.232481115372369e-06,
+      "loss": 0.5459,
+      "step": 15862
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.231305478603712e-06,
+      "loss": 0.553,
+      "step": 15863
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2301299613190105e-06,
+      "loss": 0.5676,
+      "step": 15864
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.228954563542621e-06,
+      "loss": 0.5935,
+      "step": 15865
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.227779285298879e-06,
+      "loss": 0.562,
+      "step": 15866
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.226604126612136e-06,
+      "loss": 0.5205,
+      "step": 15867
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.225429087506725e-06,
+      "loss": 0.5852,
+      "step": 15868
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.224254168006992e-06,
+      "loss": 0.54,
+      "step": 15869
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.223079368137265e-06,
+      "loss": 0.561,
+      "step": 15870
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.221904687921883e-06,
+      "loss": 0.4941,
+      "step": 15871
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.220730127385169e-06,
+      "loss": 0.519,
+      "step": 15872
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.219555686551462e-06,
+      "loss": 0.5667,
+      "step": 15873
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.218381365445077e-06,
+      "loss": 0.5618,
+      "step": 15874
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.217207164090345e-06,
+      "loss": 0.5757,
+      "step": 15875
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.216033082511579e-06,
+      "loss": 0.5637,
+      "step": 15876
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.214859120733106e-06,
+      "loss": 0.6687,
+      "step": 15877
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.213685278779236e-06,
+      "loss": 0.5764,
+      "step": 15878
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.212511556674279e-06,
+      "loss": 0.5498,
+      "step": 15879
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2113379544425544e-06,
+      "loss": 0.5527,
+      "step": 15880
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2101644721083624e-06,
+      "loss": 0.6138,
+      "step": 15881
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.208991109696016e-06,
+      "loss": 0.5813,
+      "step": 15882
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.20781786722981e-06,
+      "loss": 0.5999,
+      "step": 15883
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.206644744734053e-06,
+      "loss": 0.5093,
+      "step": 15884
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.205471742233038e-06,
+      "loss": 0.5598,
+      "step": 15885
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.204298859751066e-06,
+      "loss": 0.4985,
+      "step": 15886
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.203126097312422e-06,
+      "loss": 0.5986,
+      "step": 15887
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.201953454941403e-06,
+      "loss": 0.519,
+      "step": 15888
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2007809326623005e-06,
+      "loss": 0.5693,
+      "step": 15889
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.199608530499392e-06,
+      "loss": 0.645,
+      "step": 15890
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.198436248476965e-06,
+      "loss": 0.5754,
+      "step": 15891
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.197264086619304e-06,
+      "loss": 0.5244,
+      "step": 15892
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.196092044950683e-06,
+      "loss": 0.5715,
+      "step": 15893
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.194920123495376e-06,
+      "loss": 0.5547,
+      "step": 15894
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.193748322277662e-06,
+      "loss": 0.5557,
+      "step": 15895
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.192576641321806e-06,
+      "loss": 0.5649,
+      "step": 15896
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.191405080652082e-06,
+      "loss": 0.4395,
+      "step": 15897
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.19023364029275e-06,
+      "loss": 0.6042,
+      "step": 15898
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.189062320268082e-06,
+      "loss": 0.5134,
+      "step": 15899
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.187891120602328e-06,
+      "loss": 0.5608,
+      "step": 15900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.186720041319755e-06,
+      "loss": 0.5117,
+      "step": 15901
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1855490824446135e-06,
+      "loss": 0.5186,
+      "step": 15902
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.184378244001163e-06,
+      "loss": 0.5247,
+      "step": 15903
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.183207526013647e-06,
+      "loss": 0.6567,
+      "step": 15904
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1820369285063225e-06,
+      "loss": 0.5601,
+      "step": 15905
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.180866451503427e-06,
+      "loss": 0.5635,
+      "step": 15906
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.179696095029211e-06,
+      "loss": 0.5603,
+      "step": 15907
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.178525859107913e-06,
+      "loss": 0.6575,
+      "step": 15908
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1773557437637655e-06,
+      "loss": 0.4973,
+      "step": 15909
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1761857490210155e-06,
+      "loss": 0.5254,
+      "step": 15910
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.175015874903886e-06,
+      "loss": 0.6113,
+      "step": 15911
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.173846121436613e-06,
+      "loss": 0.5767,
+      "step": 15912
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.172676488643429e-06,
+      "loss": 0.5803,
+      "step": 15913
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.171506976548552e-06,
+      "loss": 0.5366,
+      "step": 15914
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.17033758517621e-06,
+      "loss": 0.6396,
+      "step": 15915
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1691683145506255e-06,
+      "loss": 0.5903,
+      "step": 15916
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.167999164696012e-06,
+      "loss": 0.552,
+      "step": 15917
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.16683013563659e-06,
+      "loss": 0.5488,
+      "step": 15918
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1656612273965705e-06,
+      "loss": 0.6179,
+      "step": 15919
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.164492440000167e-06,
+      "loss": 0.5356,
+      "step": 15920
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.163323773471584e-06,
+      "loss": 0.6597,
+      "step": 15921
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.162155227835032e-06,
+      "loss": 0.5151,
+      "step": 15922
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.160986803114713e-06,
+      "loss": 0.6189,
+      "step": 15923
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.159818499334823e-06,
+      "loss": 0.5574,
+      "step": 15924
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.158650316519568e-06,
+      "loss": 0.5918,
+      "step": 15925
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.157482254693137e-06,
+      "loss": 0.4907,
+      "step": 15926
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.156314313879731e-06,
+      "loss": 0.5356,
+      "step": 15927
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.155146494103533e-06,
+      "loss": 0.5596,
+      "step": 15928
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.15397879538874e-06,
+      "loss": 0.5442,
+      "step": 15929
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.152811217759529e-06,
+      "loss": 0.5942,
+      "step": 15930
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1516437612400915e-06,
+      "loss": 0.5623,
+      "step": 15931
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.150476425854602e-06,
+      "loss": 0.6086,
+      "step": 15932
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.149309211627245e-06,
+      "loss": 0.5649,
+      "step": 15933
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.148142118582188e-06,
+      "loss": 0.6013,
+      "step": 15934
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.146975146743613e-06,
+      "loss": 0.5249,
+      "step": 15935
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.145808296135686e-06,
+      "loss": 0.5688,
+      "step": 15936
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.144641566782578e-06,
+      "loss": 0.5078,
+      "step": 15937
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.143474958708451e-06,
+      "loss": 0.501,
+      "step": 15938
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.142308471937471e-06,
+      "loss": 0.551,
+      "step": 15939
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.141142106493802e-06,
+      "loss": 0.5496,
+      "step": 15940
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.139975862401594e-06,
+      "loss": 0.5178,
+      "step": 15941
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.138809739685013e-06,
+      "loss": 0.5427,
+      "step": 15942
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1376437383682036e-06,
+      "loss": 0.5422,
+      "step": 15943
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.136477858475323e-06,
+      "loss": 0.5718,
+      "step": 15944
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.135312100030513e-06,
+      "loss": 0.5474,
+      "step": 15945
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1341464630579265e-06,
+      "loss": 0.54,
+      "step": 15946
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1329809475816995e-06,
+      "loss": 0.5498,
+      "step": 15947
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.131815553625979e-06,
+      "loss": 0.5403,
+      "step": 15948
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.130650281214899e-06,
+      "loss": 0.6165,
+      "step": 15949
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.129485130372599e-06,
+      "loss": 0.5266,
+      "step": 15950
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.128320101123207e-06,
+      "loss": 0.5554,
+      "step": 15951
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1271551934908594e-06,
+      "loss": 0.616,
+      "step": 15952
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.125990407499681e-06,
+      "loss": 0.5371,
+      "step": 15953
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.124825743173796e-06,
+      "loss": 0.585,
+      "step": 15954
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.123661200537331e-06,
+      "loss": 0.6165,
+      "step": 15955
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.122496779614402e-06,
+      "loss": 0.5933,
+      "step": 15956
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.121332480429135e-06,
+      "loss": 0.5752,
+      "step": 15957
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1201683030056335e-06,
+      "loss": 0.5923,
+      "step": 15958
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.119004247368024e-06,
+      "loss": 0.6079,
+      "step": 15959
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.117840313540405e-06,
+      "loss": 0.5732,
+      "step": 15960
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1166765015468946e-06,
+      "loss": 0.5322,
+      "step": 15961
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.115512811411589e-06,
+      "loss": 0.6091,
+      "step": 15962
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.114349243158601e-06,
+      "loss": 0.5847,
+      "step": 15963
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.11318579681202e-06,
+      "loss": 0.6177,
+      "step": 15964
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.11202247239595e-06,
+      "loss": 0.5928,
+      "step": 15965
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.110859269934491e-06,
+      "loss": 0.5344,
+      "step": 15966
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1096961894517255e-06,
+      "loss": 0.5532,
+      "step": 15967
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.108533230971752e-06,
+      "loss": 0.5557,
+      "step": 15968
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.107370394518652e-06,
+      "loss": 0.5076,
+      "step": 15969
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1062076801165175e-06,
+      "loss": 0.5137,
+      "step": 15970
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.105045087789424e-06,
+      "loss": 0.6208,
+      "step": 15971
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.103882617561459e-06,
+      "loss": 0.5349,
+      "step": 15972
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1027202694566925e-06,
+      "loss": 0.6218,
+      "step": 15973
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.101558043499207e-06,
+      "loss": 0.5259,
+      "step": 15974
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.100395939713067e-06,
+      "loss": 0.5281,
+      "step": 15975
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0992339581223524e-06,
+      "loss": 0.5852,
+      "step": 15976
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.098072098751122e-06,
+      "loss": 0.5864,
+      "step": 15977
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.096910361623446e-06,
+      "loss": 0.5764,
+      "step": 15978
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0957487467633825e-06,
+      "loss": 0.5134,
+      "step": 15979
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0945872541949974e-06,
+      "loss": 0.4792,
+      "step": 15980
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0934258839423416e-06,
+      "loss": 0.5603,
+      "step": 15981
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.092264636029476e-06,
+      "loss": 0.4563,
+      "step": 15982
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.09110351048045e-06,
+      "loss": 0.5393,
+      "step": 15983
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.08994250731931e-06,
+      "loss": 0.4719,
+      "step": 15984
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.08878162657011e-06,
+      "loss": 0.5239,
+      "step": 15985
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.087620868256888e-06,
+      "loss": 0.5461,
+      "step": 15986
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.086460232403692e-06,
+      "loss": 0.6558,
+      "step": 15987
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0852997190345556e-06,
+      "loss": 0.5439,
+      "step": 15988
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0841393281735206e-06,
+      "loss": 0.5725,
+      "step": 15989
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.082979059844624e-06,
+      "loss": 0.6494,
+      "step": 15990
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.08181891407189e-06,
+      "loss": 0.5618,
+      "step": 15991
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.080658890879352e-06,
+      "loss": 0.5706,
+      "step": 15992
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.079498990291041e-06,
+      "loss": 0.5703,
+      "step": 15993
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.078339212330975e-06,
+      "loss": 0.5261,
+      "step": 15994
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.077179557023181e-06,
+      "loss": 0.5618,
+      "step": 15995
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0760200243916725e-06,
+      "loss": 0.5693,
+      "step": 15996
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0748606144604744e-06,
+      "loss": 0.5337,
+      "step": 15997
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.073701327253596e-06,
+      "loss": 0.5837,
+      "step": 15998
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.072542162795044e-06,
+      "loss": 0.562,
+      "step": 15999
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.071383121108838e-06,
+      "loss": 0.4719,
+      "step": 16000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.070224202218975e-06,
+      "loss": 0.5637,
+      "step": 16001
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.069065406149468e-06,
+      "loss": 0.5762,
+      "step": 16002
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.06790673292431e-06,
+      "loss": 0.5605,
+      "step": 16003
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.066748182567508e-06,
+      "loss": 0.5627,
+      "step": 16004
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.065589755103051e-06,
+      "loss": 0.4792,
+      "step": 16005
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0644314505549395e-06,
+      "loss": 0.6475,
+      "step": 16006
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.063273268947159e-06,
+      "loss": 0.5515,
+      "step": 16007
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.062115210303703e-06,
+      "loss": 0.5188,
+      "step": 16008
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.060957274648554e-06,
+      "loss": 0.6057,
+      "step": 16009
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0597994620057e-06,
+      "loss": 0.6157,
+      "step": 16010
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.058641772399117e-06,
+      "loss": 0.5776,
+      "step": 16011
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.057484205852789e-06,
+      "loss": 0.4998,
+      "step": 16012
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.056326762390686e-06,
+      "loss": 0.5879,
+      "step": 16013
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0551694420367905e-06,
+      "loss": 0.5872,
+      "step": 16014
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.054012244815062e-06,
+      "loss": 0.5071,
+      "step": 16015
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0528551707494755e-06,
+      "loss": 0.5654,
+      "step": 16016
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.051698219864001e-06,
+      "loss": 0.5889,
+      "step": 16017
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.050541392182592e-06,
+      "loss": 0.6077,
+      "step": 16018
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.049384687729219e-06,
+      "loss": 0.5129,
+      "step": 16019
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.048228106527831e-06,
+      "loss": 0.5891,
+      "step": 16020
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.047071648602392e-06,
+      "loss": 0.5515,
+      "step": 16021
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.045915313976847e-06,
+      "loss": 0.5439,
+      "step": 16022
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0447591026751554e-06,
+      "loss": 0.5383,
+      "step": 16023
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.043603014721257e-06,
+      "loss": 0.533,
+      "step": 16024
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.042447050139105e-06,
+      "loss": 0.5806,
+      "step": 16025
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0412912089526334e-06,
+      "loss": 0.521,
+      "step": 16026
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.040135491185791e-06,
+      "loss": 0.564,
+      "step": 16027
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.038979896862514e-06,
+      "loss": 0.5918,
+      "step": 16028
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.03782442600673e-06,
+      "loss": 0.5959,
+      "step": 16029
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.036669078642381e-06,
+      "loss": 0.5769,
+      "step": 16030
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.035513854793389e-06,
+      "loss": 0.584,
+      "step": 16031
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.034358754483691e-06,
+      "loss": 0.5001,
+      "step": 16032
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.033203777737203e-06,
+      "loss": 0.585,
+      "step": 16033
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.032048924577854e-06,
+      "loss": 0.5781,
+      "step": 16034
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.030894195029558e-06,
+      "loss": 0.6248,
+      "step": 16035
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.029739589116239e-06,
+      "loss": 0.5593,
+      "step": 16036
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.028585106861804e-06,
+      "loss": 0.5679,
+      "step": 16037
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.027430748290174e-06,
+      "loss": 0.5479,
+      "step": 16038
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.02627651342525e-06,
+      "loss": 0.5737,
+      "step": 16039
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.025122402290948e-06,
+      "loss": 0.6223,
+      "step": 16040
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0239684149111626e-06,
+      "loss": 0.5481,
+      "step": 16041
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0228145513098e-06,
+      "loss": 0.5803,
+      "step": 16042
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0216608115107645e-06,
+      "loss": 0.6025,
+      "step": 16043
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.020507195537946e-06,
+      "loss": 0.6135,
+      "step": 16044
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.019353703415244e-06,
+      "loss": 0.5562,
+      "step": 16045
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0182003351665444e-06,
+      "loss": 0.6045,
+      "step": 16046
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.017047090815743e-06,
+      "loss": 0.6118,
+      "step": 16047
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0158939703867196e-06,
+      "loss": 0.5747,
+      "step": 16048
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.014740973903365e-06,
+      "loss": 0.4717,
+      "step": 16049
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.013588101389554e-06,
+      "loss": 0.5505,
+      "step": 16050
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.012435352869172e-06,
+      "loss": 0.5879,
+      "step": 16051
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.011282728366087e-06,
+      "loss": 0.6318,
+      "step": 16052
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.010130227904182e-06,
+      "loss": 0.5403,
+      "step": 16053
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.00897785150732e-06,
+      "loss": 0.562,
+      "step": 16054
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0078255991993765e-06,
+      "loss": 0.5125,
+      "step": 16055
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.006673471004216e-06,
+      "loss": 0.519,
+      "step": 16056
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.005521466945694e-06,
+      "loss": 0.571,
+      "step": 16057
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.004369587047683e-06,
+      "loss": 0.5618,
+      "step": 16058
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.003217831334031e-06,
+      "loss": 0.5466,
+      "step": 16059
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0020661998286024e-06,
+      "loss": 0.5693,
+      "step": 16060
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.000914692555243e-06,
+      "loss": 0.6104,
+      "step": 16061
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.999763309537811e-06,
+      "loss": 0.5056,
+      "step": 16062
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.998612050800146e-06,
+      "loss": 0.5779,
+      "step": 16063
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9974609163661016e-06,
+      "loss": 0.5811,
+      "step": 16064
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.996309906259513e-06,
+      "loss": 0.6187,
+      "step": 16065
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.995159020504225e-06,
+      "loss": 0.5872,
+      "step": 16066
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.994008259124076e-06,
+      "loss": 0.5925,
+      "step": 16067
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.992857622142897e-06,
+      "loss": 0.5859,
+      "step": 16068
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9917071095845236e-06,
+      "loss": 0.6125,
+      "step": 16069
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.990556721472789e-06,
+      "loss": 0.562,
+      "step": 16070
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9894064578315165e-06,
+      "loss": 0.5085,
+      "step": 16071
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.988256318684528e-06,
+      "loss": 0.6001,
+      "step": 16072
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.987106304055652e-06,
+      "loss": 0.6091,
+      "step": 16073
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.985956413968702e-06,
+      "loss": 0.554,
+      "step": 16074
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.984806648447503e-06,
+      "loss": 0.4973,
+      "step": 16075
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.983657007515858e-06,
+      "loss": 0.5105,
+      "step": 16076
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.982507491197592e-06,
+      "loss": 0.6216,
+      "step": 16077
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.981358099516503e-06,
+      "loss": 0.5195,
+      "step": 16078
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.980208832496407e-06,
+      "loss": 0.5481,
+      "step": 16079
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.979059690161101e-06,
+      "loss": 0.5703,
+      "step": 16080
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.977910672534393e-06,
+      "loss": 0.5344,
+      "step": 16081
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.976761779640073e-06,
+      "loss": 0.5806,
+      "step": 16082
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.975613011501949e-06,
+      "loss": 0.5491,
+      "step": 16083
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9744643681438035e-06,
+      "loss": 0.5403,
+      "step": 16084
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9733158495894375e-06,
+      "loss": 0.533,
+      "step": 16085
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9721674558626345e-06,
+      "loss": 0.6118,
+      "step": 16086
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9710191869871785e-06,
+      "loss": 0.5166,
+      "step": 16087
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.969871042986858e-06,
+      "loss": 0.4912,
+      "step": 16088
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.968723023885449e-06,
+      "loss": 0.6248,
+      "step": 16089
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.967575129706732e-06,
+      "loss": 0.5032,
+      "step": 16090
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9664273604744875e-06,
+      "loss": 0.5938,
+      "step": 16091
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.96527971621248e-06,
+      "loss": 0.552,
+      "step": 16092
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.964132196944486e-06,
+      "loss": 0.5051,
+      "step": 16093
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9629848026942745e-06,
+      "loss": 0.6001,
+      "step": 16094
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.961837533485604e-06,
+      "loss": 0.5352,
+      "step": 16095
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.960690389342245e-06,
+      "loss": 0.5259,
+      "step": 16096
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.959543370287952e-06,
+      "loss": 0.5696,
+      "step": 16097
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.958396476346487e-06,
+      "loss": 0.5796,
+      "step": 16098
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9572497075416005e-06,
+      "loss": 0.5178,
+      "step": 16099
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.95610306389705e-06,
+      "loss": 0.5479,
+      "step": 16100
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9549565454365824e-06,
+      "loss": 0.6616,
+      "step": 16101
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.953810152183942e-06,
+      "loss": 0.5168,
+      "step": 16102
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.95266388416288e-06,
+      "loss": 0.6025,
+      "step": 16103
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.951517741397133e-06,
+      "loss": 0.5403,
+      "step": 16104
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.950371723910445e-06,
+      "loss": 0.5454,
+      "step": 16105
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.949225831726546e-06,
+      "loss": 0.6318,
+      "step": 16106
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.948080064869179e-06,
+      "loss": 0.5225,
+      "step": 16107
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.946934423362068e-06,
+      "loss": 0.6013,
+      "step": 16108
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.945788907228948e-06,
+      "loss": 0.5793,
+      "step": 16109
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.94464351649354e-06,
+      "loss": 0.5889,
+      "step": 16110
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.943498251179575e-06,
+      "loss": 0.5549,
+      "step": 16111
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.942353111310765e-06,
+      "loss": 0.5601,
+      "step": 16112
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.941208096910838e-06,
+      "loss": 0.5139,
+      "step": 16113
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.940063208003503e-06,
+      "loss": 0.5317,
+      "step": 16114
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.938918444612479e-06,
+      "loss": 0.5566,
+      "step": 16115
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.93777380676147e-06,
+      "loss": 0.6133,
+      "step": 16116
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.936629294474192e-06,
+      "loss": 0.5745,
+      "step": 16117
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.935484907774345e-06,
+      "loss": 0.5129,
+      "step": 16118
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.934340646685633e-06,
+      "loss": 0.5552,
+      "step": 16119
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.933196511231763e-06,
+      "loss": 0.5769,
+      "step": 16120
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.932052501436422e-06,
+      "loss": 0.5828,
+      "step": 16121
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.930908617323316e-06,
+      "loss": 0.5781,
+      "step": 16122
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.929764858916127e-06,
+      "loss": 0.54,
+      "step": 16123
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9286212262385555e-06,
+      "loss": 0.5938,
+      "step": 16124
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.927477719314281e-06,
+      "loss": 0.5664,
+      "step": 16125
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.926334338166994e-06,
+      "loss": 0.5413,
+      "step": 16126
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.925191082820371e-06,
+      "loss": 0.6162,
+      "step": 16127
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.924047953298099e-06,
+      "loss": 0.6111,
+      "step": 16128
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.922904949623847e-06,
+      "loss": 0.5549,
+      "step": 16129
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.921762071821297e-06,
+      "loss": 0.5723,
+      "step": 16130
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.920619319914116e-06,
+      "loss": 0.5891,
+      "step": 16131
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.919476693925972e-06,
+      "loss": 0.6162,
+      "step": 16132
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.918334193880537e-06,
+      "loss": 0.5369,
+      "step": 16133
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.917191819801469e-06,
+      "loss": 0.5432,
+      "step": 16134
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.916049571712436e-06,
+      "loss": 0.4417,
+      "step": 16135
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.914907449637089e-06,
+      "loss": 0.6252,
+      "step": 16136
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9137654535990925e-06,
+      "loss": 0.5996,
+      "step": 16137
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.912623583622092e-06,
+      "loss": 0.5315,
+      "step": 16138
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9114818397297464e-06,
+      "loss": 0.5212,
+      "step": 16139
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.910340221945695e-06,
+      "loss": 0.5798,
+      "step": 16140
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.909198730293594e-06,
+      "loss": 0.6104,
+      "step": 16141
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.908057364797077e-06,
+      "loss": 0.5696,
+      "step": 16142
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.906916125479788e-06,
+      "loss": 0.4805,
+      "step": 16143
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9057750123653706e-06,
+      "loss": 0.5786,
+      "step": 16144
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.90463402547745e-06,
+      "loss": 0.5181,
+      "step": 16145
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.903493164839668e-06,
+      "loss": 0.5742,
+      "step": 16146
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.902352430475647e-06,
+      "loss": 0.5764,
+      "step": 16147
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.901211822409021e-06,
+      "loss": 0.593,
+      "step": 16148
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.900071340663408e-06,
+      "loss": 0.5825,
+      "step": 16149
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.898930985262439e-06,
+      "loss": 0.6064,
+      "step": 16150
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.897790756229723e-06,
+      "loss": 0.6069,
+      "step": 16151
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8966506535888886e-06,
+      "loss": 0.4944,
+      "step": 16152
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.895510677363539e-06,
+      "loss": 0.4978,
+      "step": 16153
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.894370827577294e-06,
+      "loss": 0.6111,
+      "step": 16154
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.893231104253756e-06,
+      "loss": 0.509,
+      "step": 16155
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.89209150741654e-06,
+      "loss": 0.5623,
+      "step": 16156
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.890952037089241e-06,
+      "loss": 0.5708,
+      "step": 16157
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.889812693295469e-06,
+      "loss": 0.5574,
+      "step": 16158
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.888673476058812e-06,
+      "loss": 0.5095,
+      "step": 16159
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.887534385402877e-06,
+      "loss": 0.5063,
+      "step": 16160
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.886395421351252e-06,
+      "loss": 0.5872,
+      "step": 16161
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.885256583927526e-06,
+      "loss": 0.531,
+      "step": 16162
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8841178731552905e-06,
+      "loss": 0.6099,
+      "step": 16163
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.882979289058127e-06,
+      "loss": 0.5571,
+      "step": 16164
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8818408316596255e-06,
+      "loss": 0.5391,
+      "step": 16165
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.880702500983359e-06,
+      "loss": 0.5413,
+      "step": 16166
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.879564297052907e-06,
+      "loss": 0.5798,
+      "step": 16167
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.87842621989185e-06,
+      "loss": 0.5566,
+      "step": 16168
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.877288269523752e-06,
+      "loss": 0.5547,
+      "step": 16169
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.876150445972186e-06,
+      "loss": 0.5586,
+      "step": 16170
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.875012749260725e-06,
+      "loss": 0.5559,
+      "step": 16171
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.873875179412924e-06,
+      "loss": 0.5413,
+      "step": 16172
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.872737736452354e-06,
+      "loss": 0.6139,
+      "step": 16173
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.871600420402565e-06,
+      "loss": 0.5959,
+      "step": 16174
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.870463231287122e-06,
+      "loss": 0.5417,
+      "step": 16175
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.869326169129575e-06,
+      "loss": 0.6477,
+      "step": 16176
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.868189233953472e-06,
+      "loss": 0.531,
+      "step": 16177
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.867052425782371e-06,
+      "loss": 0.5188,
+      "step": 16178
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.865915744639806e-06,
+      "loss": 0.678,
+      "step": 16179
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.864779190549333e-06,
+      "loss": 0.5454,
+      "step": 16180
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.863642763534482e-06,
+      "loss": 0.5105,
+      "step": 16181
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.862506463618799e-06,
+      "loss": 0.6333,
+      "step": 16182
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.861370290825814e-06,
+      "loss": 0.5859,
+      "step": 16183
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8602342451790665e-06,
+      "loss": 0.55,
+      "step": 16184
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8590983267020795e-06,
+      "loss": 0.5344,
+      "step": 16185
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.857962535418387e-06,
+      "loss": 0.5701,
+      "step": 16186
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8568268713515075e-06,
+      "loss": 0.6094,
+      "step": 16187
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.855691334524971e-06,
+      "loss": 0.5977,
+      "step": 16188
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.85455592496229e-06,
+      "loss": 0.5933,
+      "step": 16189
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.853420642686989e-06,
+      "loss": 0.5486,
+      "step": 16190
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.852285487722573e-06,
+      "loss": 0.542,
+      "step": 16191
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.851150460092565e-06,
+      "loss": 0.5698,
+      "step": 16192
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.850015559820465e-06,
+      "loss": 0.5869,
+      "step": 16193
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8488807869297865e-06,
+      "loss": 0.5654,
+      "step": 16194
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.847746141444026e-06,
+      "loss": 0.5786,
+      "step": 16195
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.846611623386689e-06,
+      "loss": 0.5024,
+      "step": 16196
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.845477232781278e-06,
+      "loss": 0.5554,
+      "step": 16197
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.844342969651281e-06,
+      "loss": 0.4968,
+      "step": 16198
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8432088340202004e-06,
+      "loss": 0.5588,
+      "step": 16199
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.842074825911518e-06,
+      "loss": 0.6772,
+      "step": 16200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.84094094534873e-06,
+      "loss": 0.5771,
+      "step": 16201
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.839807192355315e-06,
+      "loss": 0.6514,
+      "step": 16202
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.838673566954761e-06,
+      "loss": 0.5344,
+      "step": 16203
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.837540069170543e-06,
+      "loss": 0.5916,
+      "step": 16204
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8364066990261454e-06,
+      "loss": 0.5215,
+      "step": 16205
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.835273456545039e-06,
+      "loss": 0.5317,
+      "step": 16206
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.834140341750693e-06,
+      "loss": 0.5686,
+      "step": 16207
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.833007354666583e-06,
+      "loss": 0.5203,
+      "step": 16208
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.831874495316171e-06,
+      "loss": 0.5344,
+      "step": 16209
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.830741763722926e-06,
+      "loss": 0.6643,
+      "step": 16210
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.829609159910305e-06,
+      "loss": 0.5183,
+      "step": 16211
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.828476683901772e-06,
+      "loss": 0.5664,
+      "step": 16212
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.827344335720777e-06,
+      "loss": 0.5212,
+      "step": 16213
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.826212115390782e-06,
+      "loss": 0.5544,
+      "step": 16214
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.825080022935229e-06,
+      "loss": 0.5625,
+      "step": 16215
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8239480583775755e-06,
+      "loss": 0.5437,
+      "step": 16216
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.822816221741259e-06,
+      "loss": 0.6311,
+      "step": 16217
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8216845130497295e-06,
+      "loss": 0.5325,
+      "step": 16218
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.820552932326421e-06,
+      "loss": 0.5198,
+      "step": 16219
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8194214795947746e-06,
+      "loss": 0.543,
+      "step": 16220
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.81829015487823e-06,
+      "loss": 0.6162,
+      "step": 16221
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.817158958200211e-06,
+      "loss": 0.5652,
+      "step": 16222
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8160278895841565e-06,
+      "loss": 0.5603,
+      "step": 16223
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.814896949053484e-06,
+      "loss": 0.627,
+      "step": 16224
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.813766136631628e-06,
+      "loss": 0.584,
+      "step": 16225
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.812635452342002e-06,
+      "loss": 0.5117,
+      "step": 16226
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8115048962080315e-06,
+      "loss": 0.5266,
+      "step": 16227
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8103744682531274e-06,
+      "loss": 0.5808,
+      "step": 16228
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8092441685007096e-06,
+      "loss": 0.625,
+      "step": 16229
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8081139969741844e-06,
+      "loss": 0.6372,
+      "step": 16230
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8069839536969657e-06,
+      "loss": 0.5964,
+      "step": 16231
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8058540386924526e-06,
+      "loss": 0.5864,
+      "step": 16232
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8047242519840568e-06,
+      "loss": 0.5901,
+      "step": 16233
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.803594593595169e-06,
+      "loss": 0.5156,
+      "step": 16234
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8024650635491975e-06,
+      "loss": 0.584,
+      "step": 16235
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.801335661869533e-06,
+      "loss": 0.5669,
+      "step": 16236
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8002063885795636e-06,
+      "loss": 0.5498,
+      "step": 16237
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.799077243702688e-06,
+      "loss": 0.519,
+      "step": 16238
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7979482272622857e-06,
+      "loss": 0.6211,
+      "step": 16239
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.79681933928175e-06,
+      "loss": 0.6033,
+      "step": 16240
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.795690579784452e-06,
+      "loss": 0.6128,
+      "step": 16241
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.794561948793782e-06,
+      "loss": 0.5544,
+      "step": 16242
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7934334463331092e-06,
+      "loss": 0.6714,
+      "step": 16243
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7923050724258093e-06,
+      "loss": 0.5366,
+      "step": 16244
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7911768270952587e-06,
+      "loss": 0.6226,
+      "step": 16245
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.790048710364819e-06,
+      "loss": 0.5234,
+      "step": 16246
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7889207222578595e-06,
+      "loss": 0.5256,
+      "step": 16247
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7877928627977468e-06,
+      "loss": 0.5591,
+      "step": 16248
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7866651320078353e-06,
+      "loss": 0.5801,
+      "step": 16249
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7855375299114906e-06,
+      "loss": 0.5513,
+      "step": 16250
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.784410056532063e-06,
+      "loss": 0.5336,
+      "step": 16251
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7832827118929027e-06,
+      "loss": 0.606,
+      "step": 16252
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7821554960173666e-06,
+      "loss": 0.5513,
+      "step": 16253
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.781028408928796e-06,
+      "loss": 0.5369,
+      "step": 16254
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7799014506505416e-06,
+      "loss": 0.5854,
+      "step": 16255
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7787746212059394e-06,
+      "loss": 0.6152,
+      "step": 16256
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7776479206183347e-06,
+      "loss": 0.5747,
+      "step": 16257
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7765213489110586e-06,
+      "loss": 0.5718,
+      "step": 16258
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.77539490610745e-06,
+      "loss": 0.5972,
+      "step": 16259
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7742685922308364e-06,
+      "loss": 0.5442,
+      "step": 16260
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.773142407304551e-06,
+      "loss": 0.521,
+      "step": 16261
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.772016351351915e-06,
+      "loss": 0.5427,
+      "step": 16262
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.770890424396256e-06,
+      "loss": 0.5447,
+      "step": 16263
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.769764626460893e-06,
+      "loss": 0.6951,
+      "step": 16264
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7686389575691417e-06,
+      "loss": 0.6052,
+      "step": 16265
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.767513417744323e-06,
+      "loss": 0.5522,
+      "step": 16266
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7663880070097424e-06,
+      "loss": 0.6006,
+      "step": 16267
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.765262725388713e-06,
+      "loss": 0.5552,
+      "step": 16268
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.764137572904548e-06,
+      "loss": 0.5935,
+      "step": 16269
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7630125495805436e-06,
+      "loss": 0.6299,
+      "step": 16270
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7618876554400075e-06,
+      "loss": 0.6033,
+      "step": 16271
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7607628905062342e-06,
+      "loss": 0.5449,
+      "step": 16272
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7596382548025235e-06,
+      "loss": 0.6033,
+      "step": 16273
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7585137483521726e-06,
+      "loss": 0.4946,
+      "step": 16274
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7573893711784647e-06,
+      "loss": 0.5474,
+      "step": 16275
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.756265123304698e-06,
+      "loss": 0.5667,
+      "step": 16276
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.755141004754148e-06,
+      "loss": 0.5271,
+      "step": 16277
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.754017015550109e-06,
+      "loss": 0.5227,
+      "step": 16278
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.752893155715854e-06,
+      "loss": 0.5627,
+      "step": 16279
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7517694252746608e-06,
+      "loss": 0.6084,
+      "step": 16280
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.750645824249809e-06,
+      "loss": 0.5811,
+      "step": 16281
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7495223526645664e-06,
+      "loss": 0.5713,
+      "step": 16282
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7483990105422086e-06,
+      "loss": 0.5923,
+      "step": 16283
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.747275797905996e-06,
+      "loss": 0.5386,
+      "step": 16284
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7461527147792e-06,
+      "loss": 0.4922,
+      "step": 16285
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7450297611850763e-06,
+      "loss": 0.4905,
+      "step": 16286
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7439069371468894e-06,
+      "loss": 0.5671,
+      "step": 16287
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.74278424268789e-06,
+      "loss": 0.5298,
+      "step": 16288
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7416616778313386e-06,
+      "loss": 0.6099,
+      "step": 16289
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7405392426004806e-06,
+      "loss": 0.5979,
+      "step": 16290
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7394169370185683e-06,
+      "loss": 0.5598,
+      "step": 16291
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.738294761108844e-06,
+      "loss": 0.5325,
+      "step": 16292
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7371727148945548e-06,
+      "loss": 0.5476,
+      "step": 16293
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7360507983989357e-06,
+      "loss": 0.5007,
+      "step": 16294
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7349290116452306e-06,
+      "loss": 0.5593,
+      "step": 16295
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.733807354656669e-06,
+      "loss": 0.5781,
+      "step": 16296
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.732685827456485e-06,
+      "loss": 0.5442,
+      "step": 16297
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.731564430067912e-06,
+      "loss": 0.5574,
+      "step": 16298
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.73044316251417e-06,
+      "loss": 0.5723,
+      "step": 16299
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7293220248184903e-06,
+      "loss": 0.6667,
+      "step": 16300
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.728201017004088e-06,
+      "loss": 0.6682,
+      "step": 16301
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.727080139094188e-06,
+      "loss": 0.5764,
+      "step": 16302
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.725959391112001e-06,
+      "loss": 0.5354,
+      "step": 16303
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7248387730807458e-06,
+      "loss": 0.5232,
+      "step": 16304
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7237182850236265e-06,
+      "loss": 0.481,
+      "step": 16305
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.722597926963858e-06,
+      "loss": 0.5381,
+      "step": 16306
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7214776989246394e-06,
+      "loss": 0.5537,
+      "step": 16307
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7203576009291807e-06,
+      "loss": 0.5459,
+      "step": 16308
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7192376330006776e-06,
+      "loss": 0.509,
+      "step": 16309
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.718117795162325e-06,
+      "loss": 0.5527,
+      "step": 16310
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7169980874373224e-06,
+      "loss": 0.5159,
+      "step": 16311
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.715878509848857e-06,
+      "loss": 0.5574,
+      "step": 16312
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7147590624201236e-06,
+      "loss": 0.4746,
+      "step": 16313
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7136397451743024e-06,
+      "loss": 0.5322,
+      "step": 16314
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7125205581345836e-06,
+      "loss": 0.5093,
+      "step": 16315
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.711401501324142e-06,
+      "loss": 0.4746,
+      "step": 16316
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.710282574766163e-06,
+      "loss": 0.5623,
+      "step": 16317
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7091637784838143e-06,
+      "loss": 0.6001,
+      "step": 16318
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.708045112500278e-06,
+      "loss": 0.4988,
+      "step": 16319
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7069265768387154e-06,
+      "loss": 0.5718,
+      "step": 16320
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.705808171522298e-06,
+      "loss": 0.5691,
+      "step": 16321
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.704689896574196e-06,
+      "loss": 0.5564,
+      "step": 16322
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.703571752017563e-06,
+      "loss": 0.54,
+      "step": 16323
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7024537378755656e-06,
+      "loss": 0.5674,
+      "step": 16324
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.701335854171353e-06,
+      "loss": 0.5251,
+      "step": 16325
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.700218100928089e-06,
+      "loss": 0.604,
+      "step": 16326
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6991004781689154e-06,
+      "loss": 0.5144,
+      "step": 16327
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.697982985916989e-06,
+      "loss": 0.553,
+      "step": 16328
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.696865624195448e-06,
+      "loss": 0.5273,
+      "step": 16329
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.695748393027444e-06,
+      "loss": 0.5637,
+      "step": 16330
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6946312924361103e-06,
+      "loss": 0.5059,
+      "step": 16331
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6935143224445914e-06,
+      "loss": 0.5293,
+      "step": 16332
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6923974830760156e-06,
+      "loss": 0.5435,
+      "step": 16333
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.691280774353523e-06,
+      "loss": 0.5979,
+      "step": 16334
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6901641963002355e-06,
+      "loss": 0.5393,
+      "step": 16335
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6890477489392873e-06,
+      "loss": 0.646,
+      "step": 16336
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6879314322937963e-06,
+      "loss": 0.5276,
+      "step": 16337
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6868152463868912e-06,
+      "loss": 0.6516,
+      "step": 16338
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.685699191241687e-06,
+      "loss": 0.5833,
+      "step": 16339
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6845832668812975e-06,
+      "loss": 0.5398,
+      "step": 16340
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6834674733288424e-06,
+      "loss": 0.6118,
+      "step": 16341
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6823518106074243e-06,
+      "loss": 0.5979,
+      "step": 16342
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.68123627874016e-06,
+      "loss": 0.5427,
+      "step": 16343
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6801208777501484e-06,
+      "loss": 0.5439,
+      "step": 16344
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.679005607660495e-06,
+      "loss": 0.5532,
+      "step": 16345
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6778904684943027e-06,
+      "loss": 0.5356,
+      "step": 16346
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.676775460274662e-06,
+      "loss": 0.541,
+      "step": 16347
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.675660583024675e-06,
+      "loss": 0.5913,
+      "step": 16348
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.674545836767427e-06,
+      "loss": 0.5134,
+      "step": 16349
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6734312215260095e-06,
+      "loss": 0.5916,
+      "step": 16350
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6723167373235137e-06,
+      "loss": 0.5388,
+      "step": 16351
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6712023841830145e-06,
+      "loss": 0.6262,
+      "step": 16352
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.670088162127602e-06,
+      "loss": 0.5649,
+      "step": 16353
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.668974071180349e-06,
+      "loss": 0.5635,
+      "step": 16354
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.66786011136433e-06,
+      "loss": 0.6016,
+      "step": 16355
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6667462827026222e-06,
+      "loss": 0.5991,
+      "step": 16356
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6656325852182906e-06,
+      "loss": 0.4951,
+      "step": 16357
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.664519018934408e-06,
+      "loss": 0.7036,
+      "step": 16358
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6634055838740346e-06,
+      "loss": 0.6069,
+      "step": 16359
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6622922800602368e-06,
+      "loss": 0.5977,
+      "step": 16360
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6611791075160695e-06,
+      "loss": 0.6033,
+      "step": 16361
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.660066066264594e-06,
+      "loss": 0.5527,
+      "step": 16362
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.658953156328857e-06,
+      "loss": 0.5454,
+      "step": 16363
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6578403777319192e-06,
+      "loss": 0.5662,
+      "step": 16364
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6567277304968196e-06,
+      "loss": 0.6133,
+      "step": 16365
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6556152146466115e-06,
+      "loss": 0.5957,
+      "step": 16366
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6545028302043317e-06,
+      "loss": 0.5625,
+      "step": 16367
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6533905771930266e-06,
+      "loss": 0.5352,
+      "step": 16368
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6522784556357305e-06,
+      "loss": 0.5725,
+      "step": 16369
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6511664655554733e-06,
+      "loss": 0.554,
+      "step": 16370
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.650054606975294e-06,
+      "loss": 0.5051,
+      "step": 16371
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6489428799182215e-06,
+      "loss": 0.5686,
+      "step": 16372
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6478312844072784e-06,
+      "loss": 0.5203,
+      "step": 16373
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6467198204654896e-06,
+      "loss": 0.5527,
+      "step": 16374
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6456084881158827e-06,
+      "loss": 0.6042,
+      "step": 16375
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6444972873814664e-06,
+      "loss": 0.6243,
+      "step": 16376
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.643386218285264e-06,
+      "loss": 0.5808,
+      "step": 16377
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.642275280850284e-06,
+      "loss": 0.5974,
+      "step": 16378
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.64116447509954e-06,
+      "loss": 0.6018,
+      "step": 16379
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6400538010560347e-06,
+      "loss": 0.5588,
+      "step": 16380
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6389432587427797e-06,
+      "loss": 0.5852,
+      "step": 16381
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6378328481827697e-06,
+      "loss": 0.6272,
+      "step": 16382
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6367225693990117e-06,
+      "loss": 0.5393,
+      "step": 16383
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.635612422414497e-06,
+      "loss": 0.5273,
+      "step": 16384
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6345024072522185e-06,
+      "loss": 0.5964,
+      "step": 16385
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.633392523935173e-06,
+      "loss": 0.5449,
+      "step": 16386
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.632282772486342e-06,
+      "loss": 0.5251,
+      "step": 16387
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6311731529287188e-06,
+      "loss": 0.4993,
+      "step": 16388
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.630063665285278e-06,
+      "loss": 0.542,
+      "step": 16389
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.628954309579008e-06,
+      "loss": 0.5955,
+      "step": 16390
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6278450858328796e-06,
+      "loss": 0.6189,
+      "step": 16391
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6267359940698733e-06,
+      "loss": 0.5496,
+      "step": 16392
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.625627034312955e-06,
+      "loss": 0.6709,
+      "step": 16393
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6245182065851015e-06,
+      "loss": 0.6904,
+      "step": 16394
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6234095109092715e-06,
+      "loss": 0.5847,
+      "step": 16395
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.622300947308436e-06,
+      "loss": 0.5225,
+      "step": 16396
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6211925158055485e-06,
+      "loss": 0.5789,
+      "step": 16397
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6200842164235726e-06,
+      "loss": 0.5715,
+      "step": 16398
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6189760491854653e-06,
+      "loss": 0.6001,
+      "step": 16399
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.617868014114174e-06,
+      "loss": 0.5349,
+      "step": 16400
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6167601112326555e-06,
+      "loss": 0.572,
+      "step": 16401
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.61565234056385e-06,
+      "loss": 0.5706,
+      "step": 16402
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.614544702130709e-06,
+      "loss": 0.6392,
+      "step": 16403
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.613437195956169e-06,
+      "loss": 0.571,
+      "step": 16404
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.612329822063173e-06,
+      "loss": 0.498,
+      "step": 16405
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6112225804746535e-06,
+      "loss": 0.5928,
+      "step": 16406
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.61011547121355e-06,
+      "loss": 0.53,
+      "step": 16407
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.609008494302787e-06,
+      "loss": 0.5679,
+      "step": 16408
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6079016497652987e-06,
+      "loss": 0.5422,
+      "step": 16409
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.606794937624004e-06,
+      "loss": 0.5713,
+      "step": 16410
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.605688357901833e-06,
+      "loss": 0.593,
+      "step": 16411
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6045819106216985e-06,
+      "loss": 0.5977,
+      "step": 16412
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.603475595806525e-06,
+      "loss": 0.5198,
+      "step": 16413
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.602369413479224e-06,
+      "loss": 0.5823,
+      "step": 16414
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6012633636627024e-06,
+      "loss": 0.5278,
+      "step": 16415
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.600157446379877e-06,
+      "loss": 0.5325,
+      "step": 16416
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5990516616536463e-06,
+      "loss": 0.6406,
+      "step": 16417
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5979460095069228e-06,
+      "loss": 0.5525,
+      "step": 16418
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5968404899625986e-06,
+      "loss": 0.5413,
+      "step": 16419
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5957351030435795e-06,
+      "loss": 0.5369,
+      "step": 16420
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.594629848772754e-06,
+      "loss": 0.4949,
+      "step": 16421
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5935247271730166e-06,
+      "loss": 0.5774,
+      "step": 16422
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.592419738267262e-06,
+      "loss": 0.5491,
+      "step": 16423
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5913148820783694e-06,
+      "loss": 0.6592,
+      "step": 16424
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5902101586292313e-06,
+      "loss": 0.5901,
+      "step": 16425
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5891055679427223e-06,
+      "loss": 0.5007,
+      "step": 16426
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5880011100417234e-06,
+      "loss": 0.5532,
+      "step": 16427
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5868967849491145e-06,
+      "loss": 0.5093,
+      "step": 16428
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5857925926877656e-06,
+      "loss": 0.5349,
+      "step": 16429
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.584688533280545e-06,
+      "loss": 0.6313,
+      "step": 16430
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5835846067503244e-06,
+      "loss": 0.5076,
+      "step": 16431
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.582480813119965e-06,
+      "loss": 0.5132,
+      "step": 16432
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5813771524123355e-06,
+      "loss": 0.552,
+      "step": 16433
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5802736246502866e-06,
+      "loss": 0.6072,
+      "step": 16434
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5791702298566843e-06,
+      "loss": 0.5811,
+      "step": 16435
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5780669680543746e-06,
+      "loss": 0.5046,
+      "step": 16436
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5769638392662153e-06,
+      "loss": 0.6216,
+      "step": 16437
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.57586084351505e-06,
+      "loss": 0.5891,
+      "step": 16438
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5747579808237286e-06,
+      "loss": 0.6484,
+      "step": 16439
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.573655251215089e-06,
+      "loss": 0.4866,
+      "step": 16440
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5725526547119772e-06,
+      "loss": 0.5793,
+      "step": 16441
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5714501913372257e-06,
+      "loss": 0.6018,
+      "step": 16442
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.570347861113674e-06,
+      "loss": 0.5332,
+      "step": 16443
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5692456640641527e-06,
+      "loss": 0.5959,
+      "step": 16444
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5681436002114854e-06,
+      "loss": 0.5698,
+      "step": 16445
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.567041669578507e-06,
+      "loss": 0.5227,
+      "step": 16446
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5659398721880335e-06,
+      "loss": 0.5764,
+      "step": 16447
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5648382080628908e-06,
+      "loss": 0.6025,
+      "step": 16448
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5637366772258984e-06,
+      "loss": 0.4932,
+      "step": 16449
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5626352796998666e-06,
+      "loss": 0.5142,
+      "step": 16450
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5615340155076107e-06,
+      "loss": 0.5005,
+      "step": 16451
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5604328846719437e-06,
+      "loss": 0.6506,
+      "step": 16452
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.559331887215667e-06,
+      "loss": 0.5935,
+      "step": 16453
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5582310231615902e-06,
+      "loss": 0.6191,
+      "step": 16454
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5571302925325093e-06,
+      "loss": 0.5562,
+      "step": 16455
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5560296953512296e-06,
+      "loss": 0.6123,
+      "step": 16456
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5549292316405414e-06,
+      "loss": 0.4956,
+      "step": 16457
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.553828901423243e-06,
+      "loss": 0.5566,
+      "step": 16458
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5527287047221238e-06,
+      "loss": 0.5264,
+      "step": 16459
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5516286415599655e-06,
+      "loss": 0.5269,
+      "step": 16460
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5505287119595623e-06,
+      "loss": 0.6411,
+      "step": 16461
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5494289159436877e-06,
+      "loss": 0.5598,
+      "step": 16462
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.54832925353513e-06,
+      "loss": 0.572,
+      "step": 16463
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.547229724756658e-06,
+      "loss": 0.5481,
+      "step": 16464
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5461303296310523e-06,
+      "loss": 0.5315,
+      "step": 16465
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5450310681810774e-06,
+      "loss": 0.5117,
+      "step": 16466
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.543931940429509e-06,
+      "loss": 0.5498,
+      "step": 16467
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5428329463991063e-06,
+      "loss": 0.5625,
+      "step": 16468
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.541734086112637e-06,
+      "loss": 0.4985,
+      "step": 16469
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.540635359592858e-06,
+      "loss": 0.5803,
+      "step": 16470
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5395367668625293e-06,
+      "loss": 0.5593,
+      "step": 16471
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5384383079444007e-06,
+      "loss": 0.6228,
+      "step": 16472
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.537339982861232e-06,
+      "loss": 0.4758,
+      "step": 16473
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5362417916357637e-06,
+      "loss": 0.5149,
+      "step": 16474
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5351437342907456e-06,
+      "loss": 0.54,
+      "step": 16475
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.534045810848925e-06,
+      "loss": 0.5386,
+      "step": 16476
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5329480213330357e-06,
+      "loss": 0.5671,
+      "step": 16477
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5318503657658222e-06,
+      "loss": 0.5281,
+      "step": 16478
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5307528441700145e-06,
+      "loss": 0.5723,
+      "step": 16479
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5296554565683494e-06,
+      "loss": 0.4885,
+      "step": 16480
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5285582029835507e-06,
+      "loss": 0.5642,
+      "step": 16481
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5274610834383515e-06,
+      "loss": 0.5139,
+      "step": 16482
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5263640979554713e-06,
+      "loss": 0.5964,
+      "step": 16483
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5252672465576356e-06,
+      "loss": 0.5508,
+      "step": 16484
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5241705292675567e-06,
+      "loss": 0.6125,
+      "step": 16485
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.523073946107959e-06,
+      "loss": 0.5232,
+      "step": 16486
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5219774971015486e-06,
+      "loss": 0.4871,
+      "step": 16487
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5208811822710355e-06,
+      "loss": 0.5208,
+      "step": 16488
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5197850016391324e-06,
+      "loss": 0.562,
+      "step": 16489
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5186889552285377e-06,
+      "loss": 0.5559,
+      "step": 16490
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5175930430619586e-06,
+      "loss": 0.6096,
+      "step": 16491
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.516497265162089e-06,
+      "loss": 0.573,
+      "step": 16492
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5154016215516316e-06,
+      "loss": 0.6313,
+      "step": 16493
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5143061122532727e-06,
+      "loss": 0.4968,
+      "step": 16494
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5132107372897107e-06,
+      "loss": 0.4884,
+      "step": 16495
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5121154966836247e-06,
+      "loss": 0.5537,
+      "step": 16496
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.511020390457709e-06,
+      "loss": 0.5474,
+      "step": 16497
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5099254186346375e-06,
+      "loss": 0.5491,
+      "step": 16498
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5088305812370947e-06,
+      "loss": 0.6118,
+      "step": 16499
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.507735878287759e-06,
+      "loss": 0.564,
+      "step": 16500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5066413098093e-06,
+      "loss": 0.5769,
+      "step": 16501
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5055468758243926e-06,
+      "loss": 0.5676,
+      "step": 16502
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5044525763557013e-06,
+      "loss": 0.6262,
+      "step": 16503
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.503358411425897e-06,
+      "loss": 0.5078,
+      "step": 16504
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5022643810576353e-06,
+      "loss": 0.5479,
+      "step": 16505
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.501170485273585e-06,
+      "loss": 0.5938,
+      "step": 16506
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.500076724096395e-06,
+      "loss": 0.5422,
+      "step": 16507
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4989830975487273e-06,
+      "loss": 0.553,
+      "step": 16508
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.497889605653226e-06,
+      "loss": 0.5349,
+      "step": 16509
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.496796248432549e-06,
+      "loss": 0.5505,
+      "step": 16510
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4957030259093328e-06,
+      "loss": 0.5549,
+      "step": 16511
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.494609938106229e-06,
+      "loss": 0.5913,
+      "step": 16512
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4935169850458717e-06,
+      "loss": 0.4744,
+      "step": 16513
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4924241667509053e-06,
+      "loss": 0.5081,
+      "step": 16514
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.491331483243956e-06,
+      "loss": 0.5627,
+      "step": 16515
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.490238934547666e-06,
+      "loss": 0.5601,
+      "step": 16516
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.489146520684659e-06,
+      "loss": 0.5591,
+      "step": 16517
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4880542416775586e-06,
+      "loss": 0.5281,
+      "step": 16518
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.486962097548997e-06,
+      "loss": 0.6218,
+      "step": 16519
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4858700883215857e-06,
+      "loss": 0.6074,
+      "step": 16520
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.484778214017952e-06,
+      "loss": 0.6284,
+      "step": 16521
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4836864746607035e-06,
+      "loss": 0.5935,
+      "step": 16522
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4825948702724597e-06,
+      "loss": 0.5203,
+      "step": 16523
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4815034008758233e-06,
+      "loss": 0.5037,
+      "step": 16524
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.480412066493406e-06,
+      "loss": 0.5793,
+      "step": 16525
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.479320867147814e-06,
+      "loss": 0.5143,
+      "step": 16526
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4782298028616425e-06,
+      "loss": 0.616,
+      "step": 16527
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.477138873657494e-06,
+      "loss": 0.5774,
+      "step": 16528
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4760480795579676e-06,
+      "loss": 0.5005,
+      "step": 16529
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4749574205856486e-06,
+      "loss": 0.571,
+      "step": 16530
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.473866896763135e-06,
+      "loss": 0.5654,
+      "step": 16531
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4727765081130106e-06,
+      "loss": 0.5576,
+      "step": 16532
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4716862546578567e-06,
+      "loss": 0.4868,
+      "step": 16533
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.470596136420261e-06,
+      "loss": 0.5757,
+      "step": 16534
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.469506153422798e-06,
+      "loss": 0.5605,
+      "step": 16535
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4684163056880494e-06,
+      "loss": 0.5417,
+      "step": 16536
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4673265932385804e-06,
+      "loss": 0.5693,
+      "step": 16537
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4662370160969717e-06,
+      "loss": 0.5693,
+      "step": 16538
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.465147574285781e-06,
+      "loss": 0.5164,
+      "step": 16539
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4640582678275824e-06,
+      "loss": 0.5332,
+      "step": 16540
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.46296909674493e-06,
+      "loss": 0.5862,
+      "step": 16541
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4618800610603907e-06,
+      "loss": 0.5059,
+      "step": 16542
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.460791160796515e-06,
+      "loss": 0.5479,
+      "step": 16543
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4597023959758624e-06,
+      "loss": 0.5806,
+      "step": 16544
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.458613766620977e-06,
+      "loss": 0.5344,
+      "step": 16545
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4575252727544164e-06,
+      "loss": 0.543,
+      "step": 16546
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4564369143987187e-06,
+      "loss": 0.5762,
+      "step": 16547
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4553486915764244e-06,
+      "loss": 0.5332,
+      "step": 16548
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.454260604310078e-06,
+      "loss": 0.4712,
+      "step": 16549
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.45317265262222e-06,
+      "loss": 0.5576,
+      "step": 16550
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4520848365353752e-06,
+      "loss": 0.5667,
+      "step": 16551
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.450997156072081e-06,
+      "loss": 0.6479,
+      "step": 16552
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.449909611254868e-06,
+      "loss": 0.6084,
+      "step": 16553
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.448822202106256e-06,
+      "loss": 0.5674,
+      "step": 16554
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.447734928648774e-06,
+      "loss": 0.6421,
+      "step": 16555
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.446647790904936e-06,
+      "loss": 0.4224,
+      "step": 16556
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4455607888972643e-06,
+      "loss": 0.603,
+      "step": 16557
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4444739226482693e-06,
+      "loss": 0.5374,
+      "step": 16558
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4433871921804685e-06,
+      "loss": 0.6543,
+      "step": 16559
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4423005975163627e-06,
+      "loss": 0.509,
+      "step": 16560
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4412141386784658e-06,
+      "loss": 0.5693,
+      "step": 16561
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4401278156892782e-06,
+      "loss": 0.6111,
+      "step": 16562
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.439041628571297e-06,
+      "loss": 0.5911,
+      "step": 16563
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4379555773470252e-06,
+      "loss": 0.6128,
+      "step": 16564
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4368696620389507e-06,
+      "loss": 0.4895,
+      "step": 16565
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4357838826695744e-06,
+      "loss": 0.5132,
+      "step": 16566
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.434698239261377e-06,
+      "loss": 0.5537,
+      "step": 16567
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.433612731836853e-06,
+      "loss": 0.5757,
+      "step": 16568
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4325273604184774e-06,
+      "loss": 0.5803,
+      "step": 16569
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4314421250287387e-06,
+      "loss": 0.5381,
+      "step": 16570
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4303570256901087e-06,
+      "loss": 0.5793,
+      "step": 16571
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4292720624250674e-06,
+      "loss": 0.634,
+      "step": 16572
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.428187235256082e-06,
+      "loss": 0.6052,
+      "step": 16573
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4271025442056284e-06,
+      "loss": 0.647,
+      "step": 16574
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4260179892961666e-06,
+      "loss": 0.5991,
+      "step": 16575
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.424933570550162e-06,
+      "loss": 0.5454,
+      "step": 16576
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.423849287990082e-06,
+      "loss": 0.5408,
+      "step": 16577
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4227651416383754e-06,
+      "loss": 0.5994,
+      "step": 16578
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.421681131517506e-06,
+      "loss": 0.5864,
+      "step": 16579
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.420597257649918e-06,
+      "loss": 0.6184,
+      "step": 16580
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.41951352005807e-06,
+      "loss": 0.5696,
+      "step": 16581
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4184299187644e-06,
+      "loss": 0.5874,
+      "step": 16582
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.41734645379136e-06,
+      "loss": 0.5811,
+      "step": 16583
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.416263125161384e-06,
+      "loss": 0.5361,
+      "step": 16584
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.415179932896918e-06,
+      "loss": 0.5959,
+      "step": 16585
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.41409687702039e-06,
+      "loss": 0.5134,
+      "step": 16586
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.41301395755424e-06,
+      "loss": 0.6111,
+      "step": 16587
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4119311745208906e-06,
+      "loss": 0.5737,
+      "step": 16588
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.410848527942776e-06,
+      "loss": 0.606,
+      "step": 16589
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4097660178423144e-06,
+      "loss": 0.5002,
+      "step": 16590
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4086836442419334e-06,
+      "loss": 0.5183,
+      "step": 16591
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4076014071640473e-06,
+      "loss": 0.5579,
+      "step": 16592
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4065193066310707e-06,
+      "loss": 0.4814,
+      "step": 16593
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.405437342665421e-06,
+      "loss": 0.6033,
+      "step": 16594
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4043555152895036e-06,
+      "loss": 0.5552,
+      "step": 16595
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4032738245257313e-06,
+      "loss": 0.5957,
+      "step": 16596
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.402192270396503e-06,
+      "loss": 0.5918,
+      "step": 16597
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4011108529242255e-06,
+      "loss": 0.5479,
+      "step": 16598
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4000295721312924e-06,
+      "loss": 0.5408,
+      "step": 16599
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3989484280401054e-06,
+      "loss": 0.5486,
+      "step": 16600
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.397867420673052e-06,
+      "loss": 0.5645,
+      "step": 16601
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3967865500525245e-06,
+      "loss": 0.5081,
+      "step": 16602
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.395705816200915e-06,
+      "loss": 0.5234,
+      "step": 16603
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3946252191406005e-06,
+      "loss": 0.5515,
+      "step": 16604
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.393544758893967e-06,
+      "loss": 0.5432,
+      "step": 16605
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.392464435483397e-06,
+      "loss": 0.5664,
+      "step": 16606
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.391384248931262e-06,
+      "loss": 0.6196,
+      "step": 16607
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3903041992599337e-06,
+      "loss": 0.5337,
+      "step": 16608
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.389224286491789e-06,
+      "loss": 0.5254,
+      "step": 16609
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3881445106491873e-06,
+      "loss": 0.6621,
+      "step": 16610
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.387064871754501e-06,
+      "loss": 0.5752,
+      "step": 16611
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.385985369830086e-06,
+      "loss": 0.5669,
+      "step": 16612
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3849060048983082e-06,
+      "loss": 0.4934,
+      "step": 16613
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3838267769815158e-06,
+      "loss": 0.584,
+      "step": 16614
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3827476861020702e-06,
+      "loss": 0.6074,
+      "step": 16615
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.381668732282316e-06,
+      "loss": 0.5439,
+      "step": 16616
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3805899155446064e-06,
+      "loss": 0.5811,
+      "step": 16617
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3795112359112793e-06,
+      "loss": 0.5642,
+      "step": 16618
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3784326934046853e-06,
+      "loss": 0.5168,
+      "step": 16619
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3773542880471554e-06,
+      "loss": 0.5303,
+      "step": 16620
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3762760198610333e-06,
+      "loss": 0.5745,
+      "step": 16621
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.37519788886865e-06,
+      "loss": 0.6091,
+      "step": 16622
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3741198950923315e-06,
+      "loss": 0.5703,
+      "step": 16623
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3730420385544127e-06,
+      "loss": 0.5298,
+      "step": 16624
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3719643192772135e-06,
+      "loss": 0.6499,
+      "step": 16625
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.370886737283058e-06,
+      "loss": 0.6199,
+      "step": 16626
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3698092925942696e-06,
+      "loss": 0.6411,
+      "step": 16627
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3687319852331567e-06,
+      "loss": 0.6184,
+      "step": 16628
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.367654815222039e-06,
+      "loss": 0.5159,
+      "step": 16629
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.366577782583228e-06,
+      "loss": 0.5874,
+      "step": 16630
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.365500887339027e-06,
+      "loss": 0.6104,
+      "step": 16631
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3644241295117473e-06,
+      "loss": 0.5859,
+      "step": 16632
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3633475091236845e-06,
+      "loss": 0.5027,
+      "step": 16633
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3622710261971436e-06,
+      "loss": 0.5923,
+      "step": 16634
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.361194680754416e-06,
+      "loss": 0.5864,
+      "step": 16635
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3601184728178026e-06,
+      "loss": 0.5515,
+      "step": 16636
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.359042402409589e-06,
+      "loss": 0.6182,
+      "step": 16637
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.357966469552062e-06,
+      "loss": 0.5901,
+      "step": 16638
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3568906742675123e-06,
+      "loss": 0.6235,
+      "step": 16639
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3558150165782156e-06,
+      "loss": 0.6108,
+      "step": 16640
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3547394965064583e-06,
+      "loss": 0.5354,
+      "step": 16641
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3536641140745108e-06,
+      "loss": 0.6108,
+      "step": 16642
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3525888693046537e-06,
+      "loss": 0.5388,
+      "step": 16643
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.351513762219151e-06,
+      "loss": 0.5117,
+      "step": 16644
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.350438792840276e-06,
+      "loss": 0.5513,
+      "step": 16645
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3493639611902894e-06,
+      "loss": 0.5857,
+      "step": 16646
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.34828926729146e-06,
+      "loss": 0.606,
+      "step": 16647
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3472147111660413e-06,
+      "loss": 0.5466,
+      "step": 16648
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.346140292836294e-06,
+      "loss": 0.5393,
+      "step": 16649
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.345066012324468e-06,
+      "loss": 0.5493,
+      "step": 16650
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3439918696528207e-06,
+      "loss": 0.6174,
+      "step": 16651
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3429178648435912e-06,
+      "loss": 0.5437,
+      "step": 16652
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.341843997919031e-06,
+      "loss": 0.5508,
+      "step": 16653
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3407702689013845e-06,
+      "loss": 0.5645,
+      "step": 16654
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.339696677812885e-06,
+      "loss": 0.6067,
+      "step": 16655
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3386232246757757e-06,
+      "loss": 0.5701,
+      "step": 16656
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3375499095122833e-06,
+      "loss": 0.6094,
+      "step": 16657
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3364767323446467e-06,
+      "loss": 0.686,
+      "step": 16658
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.335403693195086e-06,
+      "loss": 0.5188,
+      "step": 16659
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.334330792085835e-06,
+      "loss": 0.511,
+      "step": 16660
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.333258029039108e-06,
+      "loss": 0.5696,
+      "step": 16661
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.332185404077131e-06,
+      "loss": 0.5371,
+      "step": 16662
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3311129172221157e-06,
+      "loss": 0.5867,
+      "step": 16663
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.33004056849628e-06,
+      "loss": 0.5315,
+      "step": 16664
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.328968357921831e-06,
+      "loss": 0.5239,
+      "step": 16665
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3278962855209828e-06,
+      "loss": 0.5371,
+      "step": 16666
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.326824351315936e-06,
+      "loss": 0.541,
+      "step": 16667
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3257525553288905e-06,
+      "loss": 0.5334,
+      "step": 16668
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3246808975820533e-06,
+      "loss": 0.6038,
+      "step": 16669
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.323609378097613e-06,
+      "loss": 0.5845,
+      "step": 16670
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.322537996897771e-06,
+      "loss": 0.5701,
+      "step": 16671
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.321466754004712e-06,
+      "loss": 0.5723,
+      "step": 16672
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3203956494406284e-06,
+      "loss": 0.5461,
+      "step": 16673
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3193246832277028e-06,
+      "loss": 0.5476,
+      "step": 16674
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3182538553881203e-06,
+      "loss": 0.5891,
+      "step": 16675
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3171831659440555e-06,
+      "loss": 0.5276,
+      "step": 16676
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3161126149176926e-06,
+      "loss": 0.5837,
+      "step": 16677
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3150422023311967e-06,
+      "loss": 0.5862,
+      "step": 16678
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.313971928206744e-06,
+      "loss": 0.519,
+      "step": 16679
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3129017925665043e-06,
+      "loss": 0.5928,
+      "step": 16680
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.311831795432636e-06,
+      "loss": 0.5669,
+      "step": 16681
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.310761936827309e-06,
+      "loss": 0.5681,
+      "step": 16682
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3096922167726766e-06,
+      "loss": 0.5681,
+      "step": 16683
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3086226352909e-06,
+      "loss": 0.5969,
+      "step": 16684
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3075531924041282e-06,
+      "loss": 0.5608,
+      "step": 16685
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3064838881345173e-06,
+      "loss": 0.4692,
+      "step": 16686
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.30541472250421e-06,
+      "loss": 0.4778,
+      "step": 16687
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.304345695535357e-06,
+      "loss": 0.4922,
+      "step": 16688
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.303276807250094e-06,
+      "loss": 0.5588,
+      "step": 16689
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.302208057670566e-06,
+      "loss": 0.562,
+      "step": 16690
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.301139446818905e-06,
+      "loss": 0.4734,
+      "step": 16691
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.30007097471725e-06,
+      "loss": 0.5266,
+      "step": 16692
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.299002641387724e-06,
+      "loss": 0.5825,
+      "step": 16693
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.297934446852464e-06,
+      "loss": 0.6179,
+      "step": 16694
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2968663911335897e-06,
+      "loss": 0.5522,
+      "step": 16695
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.29579847425322e-06,
+      "loss": 0.562,
+      "step": 16696
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.294730696233481e-06,
+      "loss": 0.5422,
+      "step": 16697
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2936630570964835e-06,
+      "loss": 0.6145,
+      "step": 16698
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2925955568643455e-06,
+      "loss": 0.571,
+      "step": 16699
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2915281955591727e-06,
+      "loss": 0.5894,
+      "step": 16700
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.290460973203079e-06,
+      "loss": 0.5396,
+      "step": 16701
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.289393889818161e-06,
+      "loss": 0.6357,
+      "step": 16702
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2883269454265266e-06,
+      "loss": 0.5774,
+      "step": 16703
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.287260140050277e-06,
+      "loss": 0.5532,
+      "step": 16704
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.286193473711501e-06,
+      "loss": 0.5811,
+      "step": 16705
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2851269464322954e-06,
+      "loss": 0.5984,
+      "step": 16706
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.284060558234755e-06,
+      "loss": 0.6633,
+      "step": 16707
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.28299430914096e-06,
+      "loss": 0.5779,
+      "step": 16708
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2819281991730024e-06,
+      "loss": 0.5979,
+      "step": 16709
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.280862228352959e-06,
+      "loss": 0.5002,
+      "step": 16710
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2797963967029077e-06,
+      "loss": 0.5432,
+      "step": 16711
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.278730704244929e-06,
+      "loss": 0.4829,
+      "step": 16712
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2776651510010904e-06,
+      "loss": 0.6255,
+      "step": 16713
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.276599736993469e-06,
+      "loss": 0.5828,
+      "step": 16714
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2755344622441256e-06,
+      "loss": 0.5742,
+      "step": 16715
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2744693267751292e-06,
+      "loss": 0.5437,
+      "step": 16716
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2734043306085382e-06,
+      "loss": 0.582,
+      "step": 16717
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.272339473766415e-06,
+      "loss": 0.6016,
+      "step": 16718
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2712747562708115e-06,
+      "loss": 0.5083,
+      "step": 16719
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2702101781437846e-06,
+      "loss": 0.5334,
+      "step": 16720
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.269145739407379e-06,
+      "loss": 0.6152,
+      "step": 16721
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2680814400836493e-06,
+      "loss": 0.5791,
+      "step": 16722
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.267017280194631e-06,
+      "loss": 0.5081,
+      "step": 16723
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2659532597623734e-06,
+      "loss": 0.5603,
+      "step": 16724
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.264889378808912e-06,
+      "loss": 0.6487,
+      "step": 16725
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.263825637356279e-06,
+      "loss": 0.5098,
+      "step": 16726
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.262762035426511e-06,
+      "loss": 0.6072,
+      "step": 16727
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2616985730416383e-06,
+      "loss": 0.645,
+      "step": 16728
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.260635250223685e-06,
+      "loss": 0.5444,
+      "step": 16729
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.259572066994676e-06,
+      "loss": 0.6086,
+      "step": 16730
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.258509023376637e-06,
+      "loss": 0.5317,
+      "step": 16731
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2574461193915787e-06,
+      "loss": 0.6326,
+      "step": 16732
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.256383355061523e-06,
+      "loss": 0.5984,
+      "step": 16733
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2553207304084766e-06,
+      "loss": 0.5657,
+      "step": 16734
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2542582454544556e-06,
+      "loss": 0.5461,
+      "step": 16735
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2531959002214597e-06,
+      "loss": 0.5764,
+      "step": 16736
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2521336947314996e-06,
+      "loss": 0.6108,
+      "step": 16737
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.251071629006569e-06,
+      "loss": 0.5139,
+      "step": 16738
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.250009703068674e-06,
+      "loss": 0.5859,
+      "step": 16739
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.248947916939805e-06,
+      "loss": 0.5684,
+      "step": 16740
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2478862706419513e-06,
+      "loss": 0.551,
+      "step": 16741
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2468247641971084e-06,
+      "loss": 0.5828,
+      "step": 16742
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.245763397627257e-06,
+      "loss": 0.5515,
+      "step": 16743
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.244702170954387e-06,
+      "loss": 0.5515,
+      "step": 16744
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2436410842004716e-06,
+      "loss": 0.6221,
+      "step": 16745
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2425801373874968e-06,
+      "loss": 0.6091,
+      "step": 16746
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.241519330537429e-06,
+      "loss": 0.4753,
+      "step": 16747
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.240458663672248e-06,
+      "loss": 0.635,
+      "step": 16748
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.239398136813916e-06,
+      "loss": 0.6062,
+      "step": 16749
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2383377499844048e-06,
+      "loss": 0.5261,
+      "step": 16750
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2372775032056724e-06,
+      "loss": 0.5481,
+      "step": 16751
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2362173964996847e-06,
+      "loss": 0.5664,
+      "step": 16752
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2351574298883936e-06,
+      "loss": 0.5251,
+      "step": 16753
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.234097603393759e-06,
+      "loss": 0.552,
+      "step": 16754
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.233037917037727e-06,
+      "loss": 0.5659,
+      "step": 16755
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.231978370842249e-06,
+      "loss": 0.5813,
+      "step": 16756
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2309189648292737e-06,
+      "loss": 0.5757,
+      "step": 16757
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2298596990207386e-06,
+      "loss": 0.5017,
+      "step": 16758
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2288005734385896e-06,
+      "loss": 0.605,
+      "step": 16759
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.227741588104758e-06,
+      "loss": 0.5284,
+      "step": 16760
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.226682743041183e-06,
+      "loss": 0.5496,
+      "step": 16761
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2256240382697913e-06,
+      "loss": 0.5049,
+      "step": 16762
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2245654738125155e-06,
+      "loss": 0.6436,
+      "step": 16763
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.223507049691277e-06,
+      "loss": 0.5054,
+      "step": 16764
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2224487659280034e-06,
+      "loss": 0.5847,
+      "step": 16765
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2213906225446077e-06,
+      "loss": 0.6135,
+      "step": 16766
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2203326195630145e-06,
+      "loss": 0.5247,
+      "step": 16767
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.219274757005131e-06,
+      "loss": 0.5667,
+      "step": 16768
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2182170348928732e-06,
+      "loss": 0.6492,
+      "step": 16769
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2171594532481464e-06,
+      "loss": 0.5674,
+      "step": 16770
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.216102012092853e-06,
+      "loss": 0.498,
+      "step": 16771
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2150447114489027e-06,
+      "loss": 0.5747,
+      "step": 16772
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.213987551338186e-06,
+      "loss": 0.6108,
+      "step": 16773
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2129305317826077e-06,
+      "loss": 0.5359,
+      "step": 16774
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2118736528040526e-06,
+      "loss": 0.5081,
+      "step": 16775
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2108169144244207e-06,
+      "loss": 0.5488,
+      "step": 16776
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2097603166655923e-06,
+      "loss": 0.5505,
+      "step": 16777
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2087038595494567e-06,
+      "loss": 0.5769,
+      "step": 16778
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2076475430978914e-06,
+      "loss": 0.6445,
+      "step": 16779
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2065913673327777e-06,
+      "loss": 0.5198,
+      "step": 16780
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2055353322759954e-06,
+      "loss": 0.6191,
+      "step": 16781
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2044794379494094e-06,
+      "loss": 0.5251,
+      "step": 16782
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2034236843748956e-06,
+      "loss": 0.5652,
+      "step": 16783
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2023680715743234e-06,
+      "loss": 0.5591,
+      "step": 16784
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.201312599569553e-06,
+      "loss": 0.5701,
+      "step": 16785
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.200257268382444e-06,
+      "loss": 0.5938,
+      "step": 16786
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.199202078034861e-06,
+      "loss": 0.5525,
+      "step": 16787
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1981470285486527e-06,
+      "loss": 0.491,
+      "step": 16788
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1970921199456786e-06,
+      "loss": 0.5095,
+      "step": 16789
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1960373522477816e-06,
+      "loss": 0.5859,
+      "step": 16790
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1949827254768163e-06,
+      "loss": 0.6079,
+      "step": 16791
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.193928239654619e-06,
+      "loss": 0.4871,
+      "step": 16792
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1928738948030367e-06,
+      "loss": 0.6372,
+      "step": 16793
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1918196909439027e-06,
+      "loss": 0.5872,
+      "step": 16794
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1907656280990574e-06,
+      "loss": 0.5276,
+      "step": 16795
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.189711706290327e-06,
+      "loss": 0.4939,
+      "step": 16796
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1886579255395467e-06,
+      "loss": 0.6318,
+      "step": 16797
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1876042858685374e-06,
+      "loss": 0.5635,
+      "step": 16798
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1865507872991284e-06,
+      "loss": 0.4814,
+      "step": 16799
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1854974298531372e-06,
+      "loss": 0.6008,
+      "step": 16800
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.184444213552378e-06,
+      "loss": 0.5803,
+      "step": 16801
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1833911384186733e-06,
+      "loss": 0.542,
+      "step": 16802
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.182338204473826e-06,
+      "loss": 0.5996,
+      "step": 16803
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1812854117396496e-06,
+      "loss": 0.561,
+      "step": 16804
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1802327602379547e-06,
+      "loss": 0.5723,
+      "step": 16805
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1791802499905354e-06,
+      "loss": 0.5227,
+      "step": 16806
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1781278810191962e-06,
+      "loss": 0.6321,
+      "step": 16807
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.177075653345737e-06,
+      "loss": 0.5369,
+      "step": 16808
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1760235669919458e-06,
+      "loss": 0.6326,
+      "step": 16809
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1749716219796202e-06,
+      "loss": 0.5596,
+      "step": 16810
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.173919818330543e-06,
+      "loss": 0.5303,
+      "step": 16811
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1728681560665055e-06,
+      "loss": 0.5688,
+      "step": 16812
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.171816635209284e-06,
+      "loss": 0.5491,
+      "step": 16813
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1707652557806646e-06,
+      "loss": 0.5088,
+      "step": 16814
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.16971401780242e-06,
+      "loss": 0.5479,
+      "step": 16815
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1686629212963204e-06,
+      "loss": 0.5859,
+      "step": 16816
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.167611966284145e-06,
+      "loss": 0.5188,
+      "step": 16817
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.166561152787654e-06,
+      "loss": 0.5762,
+      "step": 16818
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1655104808286184e-06,
+      "loss": 0.6069,
+      "step": 16819
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1644599504287945e-06,
+      "loss": 0.4747,
+      "step": 16820
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.163409561609948e-06,
+      "loss": 0.4941,
+      "step": 16821
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.162359314393828e-06,
+      "loss": 0.5757,
+      "step": 16822
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1613092088021945e-06,
+      "loss": 0.4922,
+      "step": 16823
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1602592448567904e-06,
+      "loss": 0.5315,
+      "step": 16824
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1592094225793712e-06,
+      "loss": 0.561,
+      "step": 16825
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1581597419916745e-06,
+      "loss": 0.5803,
+      "step": 16826
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1571102031154464e-06,
+      "loss": 0.533,
+      "step": 16827
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.156060805972422e-06,
+      "loss": 0.5557,
+      "step": 16828
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1550115505843405e-06,
+      "loss": 0.5312,
+      "step": 16829
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1539624369729295e-06,
+      "loss": 0.585,
+      "step": 16830
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.152913465159925e-06,
+      "loss": 0.5327,
+      "step": 16831
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.151864635167048e-06,
+      "loss": 0.574,
+      "step": 16832
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1508159470160236e-06,
+      "loss": 0.51,
+      "step": 16833
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.149767400728578e-06,
+      "loss": 0.5046,
+      "step": 16834
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.148718996326423e-06,
+      "loss": 0.5474,
+      "step": 16835
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1476707338312785e-06,
+      "loss": 0.5686,
+      "step": 16836
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1466226132648514e-06,
+      "loss": 0.6401,
+      "step": 16837
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1455746346488557e-06,
+      "loss": 0.5972,
+      "step": 16838
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.144526798004993e-06,
+      "loss": 0.6084,
+      "step": 16839
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1434791033549726e-06,
+      "loss": 0.636,
+      "step": 16840
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.142431550720487e-06,
+      "loss": 0.5532,
+      "step": 16841
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1413841401232426e-06,
+      "loss": 0.5259,
+      "step": 16842
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.140336871584925e-06,
+      "loss": 0.5737,
+      "step": 16843
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.139289745127233e-06,
+      "loss": 0.5476,
+      "step": 16844
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1382427607718524e-06,
+      "loss": 0.5205,
+      "step": 16845
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1371959185404655e-06,
+      "loss": 0.473,
+      "step": 16846
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1361492184547614e-06,
+      "loss": 0.6135,
+      "step": 16847
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.135102660536413e-06,
+      "loss": 0.6553,
+      "step": 16848
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1340562448071045e-06,
+      "loss": 0.6082,
+      "step": 16849
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.133009971288502e-06,
+      "loss": 0.5171,
+      "step": 16850
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1319638400022835e-06,
+      "loss": 0.5796,
+      "step": 16851
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.130917850970111e-06,
+      "loss": 0.5898,
+      "step": 16852
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1298720042136556e-06,
+      "loss": 0.625,
+      "step": 16853
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1288262997545728e-06,
+      "loss": 0.5227,
+      "step": 16854
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1277807376145273e-06,
+      "loss": 0.5134,
+      "step": 16855
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1267353178151703e-06,
+      "loss": 0.5068,
+      "step": 16856
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1256900403781577e-06,
+      "loss": 0.5459,
+      "step": 16857
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1246449053251436e-06,
+      "loss": 0.5298,
+      "step": 16858
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1235999126777684e-06,
+      "loss": 0.5657,
+      "step": 16859
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1225550624576818e-06,
+      "loss": 0.6357,
+      "step": 16860
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.12151035468652e-06,
+      "loss": 0.5237,
+      "step": 16861
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1204657893859293e-06,
+      "loss": 0.5396,
+      "step": 16862
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1194213665775354e-06,
+      "loss": 0.5752,
+      "step": 16863
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.11837708628298e-06,
+      "loss": 0.635,
+      "step": 16864
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1173329485238855e-06,
+      "loss": 0.5088,
+      "step": 16865
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1162889533218854e-06,
+      "loss": 0.5825,
+      "step": 16866
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1152451006985963e-06,
+      "loss": 0.541,
+      "step": 16867
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.114201390675645e-06,
+      "loss": 0.4954,
+      "step": 16868
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.113157823274644e-06,
+      "loss": 0.5557,
+      "step": 16869
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1121143985172154e-06,
+      "loss": 0.6299,
+      "step": 16870
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.111071116424962e-06,
+      "loss": 0.5637,
+      "step": 16871
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.110027977019501e-06,
+      "loss": 0.5144,
+      "step": 16872
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1089849803224315e-06,
+      "loss": 0.5911,
+      "step": 16873
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1079421263553634e-06,
+      "loss": 0.5986,
+      "step": 16874
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1068994151398934e-06,
+      "loss": 0.5024,
+      "step": 16875
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1058568466976156e-06,
+      "loss": 0.5818,
+      "step": 16876
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1048144210501295e-06,
+      "loss": 0.5183,
+      "step": 16877
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1037721382190213e-06,
+      "loss": 0.5557,
+      "step": 16878
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.102729998225885e-06,
+      "loss": 0.6389,
+      "step": 16879
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1016880010923e-06,
+      "loss": 0.4917,
+      "step": 16880
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.100646146839851e-06,
+      "loss": 0.5698,
+      "step": 16881
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.09960443549012e-06,
+      "loss": 0.5735,
+      "step": 16882
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0985628670646794e-06,
+      "loss": 0.5845,
+      "step": 16883
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.097521441585104e-06,
+      "loss": 0.5857,
+      "step": 16884
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0964801590729685e-06,
+      "loss": 0.637,
+      "step": 16885
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.095439019549833e-06,
+      "loss": 0.5864,
+      "step": 16886
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0943980230372696e-06,
+      "loss": 0.644,
+      "step": 16887
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.093357169556832e-06,
+      "loss": 0.5618,
+      "step": 16888
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.092316459130087e-06,
+      "loss": 0.5774,
+      "step": 16889
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0912758917785858e-06,
+      "loss": 0.6501,
+      "step": 16890
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.090235467523879e-06,
+      "loss": 0.5327,
+      "step": 16891
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0891951863875215e-06,
+      "loss": 0.5515,
+      "step": 16892
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.088155048391054e-06,
+      "loss": 0.613,
+      "step": 16893
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.087115053556028e-06,
+      "loss": 0.4929,
+      "step": 16894
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0860752019039766e-06,
+      "loss": 0.5938,
+      "step": 16895
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0850354934564453e-06,
+      "loss": 0.5393,
+      "step": 16896
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0839959282349607e-06,
+      "loss": 0.5969,
+      "step": 16897
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0829565062610635e-06,
+      "loss": 0.5588,
+      "step": 16898
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.081917227556275e-06,
+      "loss": 0.5017,
+      "step": 16899
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0808780921421276e-06,
+      "loss": 0.561,
+      "step": 16900
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0798391000401384e-06,
+      "loss": 0.6233,
+      "step": 16901
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.078800251271833e-06,
+      "loss": 0.5349,
+      "step": 16902
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0777615458587272e-06,
+      "loss": 0.5537,
+      "step": 16903
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0767229838223302e-06,
+      "loss": 0.5286,
+      "step": 16904
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0756845651841605e-06,
+      "loss": 0.6306,
+      "step": 16905
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.074646289965719e-06,
+      "loss": 0.5842,
+      "step": 16906
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0736081581885146e-06,
+      "loss": 0.5178,
+      "step": 16907
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.072570169874054e-06,
+      "loss": 0.5154,
+      "step": 16908
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.071532325043829e-06,
+      "loss": 0.5918,
+      "step": 16909
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.070494623719338e-06,
+      "loss": 0.5637,
+      "step": 16910
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.06945706592208e-06,
+      "loss": 0.613,
+      "step": 16911
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0684196516735377e-06,
+      "loss": 0.5925,
+      "step": 16912
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.067382380995204e-06,
+      "loss": 0.623,
+      "step": 16913
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.066345253908558e-06,
+      "loss": 0.5742,
+      "step": 16914
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.065308270435089e-06,
+      "loss": 0.6082,
+      "step": 16915
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0642714305962663e-06,
+      "loss": 0.5242,
+      "step": 16916
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0632347344135737e-06,
+      "loss": 0.6309,
+      "step": 16917
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.062198181908479e-06,
+      "loss": 0.533,
+      "step": 16918
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0611617731024513e-06,
+      "loss": 0.5161,
+      "step": 16919
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.06012550801696e-06,
+      "loss": 0.5757,
+      "step": 16920
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.059089386673465e-06,
+      "loss": 0.5493,
+      "step": 16921
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0580534090934323e-06,
+      "loss": 0.6104,
+      "step": 16922
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0570175752983144e-06,
+      "loss": 0.6177,
+      "step": 16923
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.05598188530957e-06,
+      "loss": 0.6074,
+      "step": 16924
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.054946339148647e-06,
+      "loss": 0.5493,
+      "step": 16925
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.053910936836998e-06,
+      "loss": 0.5645,
+      "step": 16926
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0528756783960634e-06,
+      "loss": 0.5916,
+      "step": 16927
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.051840563847294e-06,
+      "loss": 0.6604,
+      "step": 16928
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0508055932121207e-06,
+      "loss": 0.587,
+      "step": 16929
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.049770766511988e-06,
+      "loss": 0.5361,
+      "step": 16930
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.048736083768323e-06,
+      "loss": 0.5347,
+      "step": 16931
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0477015450025636e-06,
+      "loss": 0.4834,
+      "step": 16932
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0466671502361312e-06,
+      "loss": 0.6262,
+      "step": 16933
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0456328994904525e-06,
+      "loss": 0.595,
+      "step": 16934
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0445987927869536e-06,
+      "loss": 0.5386,
+      "step": 16935
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0435648301470467e-06,
+      "loss": 0.5703,
+      "step": 16936
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.042531011592155e-06,
+      "loss": 0.6484,
+      "step": 16937
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.041497337143684e-06,
+      "loss": 0.5042,
+      "step": 16938
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0404638068230498e-06,
+      "loss": 0.5664,
+      "step": 16939
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0394304206516545e-06,
+      "loss": 0.6113,
+      "step": 16940
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.038397178650907e-06,
+      "loss": 0.5676,
+      "step": 16941
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0373640808422032e-06,
+      "loss": 0.6497,
+      "step": 16942
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0363311272469465e-06,
+      "loss": 0.5237,
+      "step": 16943
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.035298317886526e-06,
+      "loss": 0.5691,
+      "step": 16944
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.034265652782339e-06,
+      "loss": 0.5227,
+      "step": 16945
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0332331319557704e-06,
+      "loss": 0.5732,
+      "step": 16946
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0322007554282106e-06,
+      "loss": 0.5183,
+      "step": 16947
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.031168523221041e-06,
+      "loss": 0.6082,
+      "step": 16948
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.030136435355637e-06,
+      "loss": 0.5449,
+      "step": 16949
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.029104491853383e-06,
+      "loss": 0.6191,
+      "step": 16950
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0280726927356476e-06,
+      "loss": 0.563,
+      "step": 16951
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.027041038023806e-06,
+      "loss": 0.5569,
+      "step": 16952
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.026009527739222e-06,
+      "loss": 0.563,
+      "step": 16953
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.024978161903266e-06,
+      "loss": 0.6243,
+      "step": 16954
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0239469405372933e-06,
+      "loss": 0.5676,
+      "step": 16955
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.022915863662672e-06,
+      "loss": 0.5996,
+      "step": 16956
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.021884931300749e-06,
+      "loss": 0.562,
+      "step": 16957
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0208541434728823e-06,
+      "loss": 0.5728,
+      "step": 16958
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.019823500200424e-06,
+      "loss": 0.5781,
+      "step": 16959
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.018793001504716e-06,
+      "loss": 0.5566,
+      "step": 16960
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0177626474071055e-06,
+      "loss": 0.4834,
+      "step": 16961
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0167324379289365e-06,
+      "loss": 0.6177,
+      "step": 16962
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0157023730915437e-06,
+      "loss": 0.5625,
+      "step": 16963
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.01467245291626e-06,
+      "loss": 0.5613,
+      "step": 16964
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.013642677424423e-06,
+      "loss": 0.5327,
+      "step": 16965
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.012613046637357e-06,
+      "loss": 0.5535,
+      "step": 16966
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0115835605763934e-06,
+      "loss": 0.6194,
+      "step": 16967
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.010554219262849e-06,
+      "loss": 0.5815,
+      "step": 16968
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0095250227180515e-06,
+      "loss": 0.6016,
+      "step": 16969
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.00849597096331e-06,
+      "loss": 0.4702,
+      "step": 16970
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0074670640199467e-06,
+      "loss": 0.5342,
+      "step": 16971
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0064383019092645e-06,
+      "loss": 0.5696,
+      "step": 16972
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0054096846525805e-06,
+      "loss": 0.5789,
+      "step": 16973
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.004381212271191e-06,
+      "loss": 0.5989,
+      "step": 16974
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0033528847864067e-06,
+      "loss": 0.5122,
+      "step": 16975
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.00232470221952e-06,
+      "loss": 0.5332,
+      "step": 16976
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.001296664591833e-06,
+      "loss": 0.5239,
+      "step": 16977
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.000268771924635e-06,
+      "loss": 0.658,
+      "step": 16978
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9992410242392146e-06,
+      "loss": 0.5364,
+      "step": 16979
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.998213421556864e-06,
+      "loss": 0.5398,
+      "step": 16980
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.997185963898862e-06,
+      "loss": 0.5439,
+      "step": 16981
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9961586512864947e-06,
+      "loss": 0.5576,
+      "step": 16982
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.995131483741036e-06,
+      "loss": 0.5928,
+      "step": 16983
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9941044612837635e-06,
+      "loss": 0.5288,
+      "step": 16984
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.993077583935952e-06,
+      "loss": 0.5608,
+      "step": 16985
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9920508517188652e-06,
+      "loss": 0.5293,
+      "step": 16986
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.991024264653771e-06,
+      "loss": 0.5576,
+      "step": 16987
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9899978227619363e-06,
+      "loss": 0.5645,
+      "step": 16988
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.988971526064616e-06,
+      "loss": 0.499,
+      "step": 16989
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.987945374583073e-06,
+      "loss": 0.5667,
+      "step": 16990
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9869193683385545e-06,
+      "loss": 0.5811,
+      "step": 16991
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9858935073523186e-06,
+      "loss": 0.5935,
+      "step": 16992
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.984867791645609e-06,
+      "loss": 0.5361,
+      "step": 16993
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.98384222123967e-06,
+      "loss": 0.5327,
+      "step": 16994
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9828167961557486e-06,
+      "loss": 0.5549,
+      "step": 16995
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9817915164150768e-06,
+      "loss": 0.6035,
+      "step": 16996
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.980766382038899e-06,
+      "loss": 0.5898,
+      "step": 16997
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9797413930484407e-06,
+      "loss": 0.5432,
+      "step": 16998
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9787165494649374e-06,
+      "loss": 0.5881,
+      "step": 16999
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9776918513096122e-06,
+      "loss": 0.5249,
+      "step": 17000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.976667298603694e-06,
+      "loss": 0.5779,
+      "step": 17001
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9756428913683964e-06,
+      "loss": 0.5906,
+      "step": 17002
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9746186296249456e-06,
+      "loss": 0.5022,
+      "step": 17003
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9735945133945497e-06,
+      "loss": 0.592,
+      "step": 17004
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9725705426984276e-06,
+      "loss": 0.5857,
+      "step": 17005
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9715467175577807e-06,
+      "loss": 0.5105,
+      "step": 17006
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.970523037993821e-06,
+      "loss": 0.4839,
+      "step": 17007
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9694995040277475e-06,
+      "loss": 0.4915,
+      "step": 17008
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.968476115680764e-06,
+      "loss": 0.6162,
+      "step": 17009
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9674528729740626e-06,
+      "loss": 0.6567,
+      "step": 17010
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9664297759288396e-06,
+      "loss": 0.5308,
+      "step": 17011
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.965406824566289e-06,
+      "loss": 0.5203,
+      "step": 17012
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.964384018907592e-06,
+      "loss": 0.5972,
+      "step": 17013
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.963361358973942e-06,
+      "loss": 0.5217,
+      "step": 17014
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.962338844786511e-06,
+      "loss": 0.5801,
+      "step": 17015
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.961316476366487e-06,
+      "loss": 0.562,
+      "step": 17016
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9602942537350375e-06,
+      "loss": 0.6401,
+      "step": 17017
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.959272176913344e-06,
+      "loss": 0.5239,
+      "step": 17018
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.958250245922567e-06,
+      "loss": 0.4927,
+      "step": 17019
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.957228460783882e-06,
+      "loss": 0.5381,
+      "step": 17020
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9562068215184445e-06,
+      "loss": 0.521,
+      "step": 17021
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.955185328147423e-06,
+      "loss": 0.6057,
+      "step": 17022
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9541639806919707e-06,
+      "loss": 0.5779,
+      "step": 17023
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9531427791732403e-06,
+      "loss": 0.5615,
+      "step": 17024
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.952121723612389e-06,
+      "loss": 0.54,
+      "step": 17025
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9511008140305586e-06,
+      "loss": 0.5149,
+      "step": 17026
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9500800504489022e-06,
+      "loss": 0.5417,
+      "step": 17027
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.949059432888556e-06,
+      "loss": 0.5247,
+      "step": 17028
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.948038961370664e-06,
+      "loss": 0.5742,
+      "step": 17029
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9470186359163586e-06,
+      "loss": 0.4719,
+      "step": 17030
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.945998456546778e-06,
+      "loss": 0.5669,
+      "step": 17031
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.944978423283047e-06,
+      "loss": 0.5759,
+      "step": 17032
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9439585361462996e-06,
+      "loss": 0.6143,
+      "step": 17033
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9429387951576536e-06,
+      "loss": 0.4958,
+      "step": 17034
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9419192003382337e-06,
+      "loss": 0.6377,
+      "step": 17035
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9408997517091607e-06,
+      "loss": 0.5281,
+      "step": 17036
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9398804492915434e-06,
+      "loss": 0.5129,
+      "step": 17037
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9388612931065017e-06,
+      "loss": 0.5991,
+      "step": 17038
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9378422831751387e-06,
+      "loss": 0.5732,
+      "step": 17039
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9368234195185653e-06,
+      "loss": 0.5627,
+      "step": 17040
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9358047021578796e-06,
+      "loss": 0.5862,
+      "step": 17041
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9347861311141877e-06,
+      "loss": 0.5471,
+      "step": 17042
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.93376770640858e-06,
+      "loss": 0.5549,
+      "step": 17043
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.932749428062157e-06,
+      "loss": 0.5366,
+      "step": 17044
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9317312960960042e-06,
+      "loss": 0.5378,
+      "step": 17045
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.930713310531216e-06,
+      "loss": 0.4807,
+      "step": 17046
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.92969547138887e-06,
+      "loss": 0.4949,
+      "step": 17047
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9286777786900554e-06,
+      "loss": 0.5798,
+      "step": 17048
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9276602324558444e-06,
+      "loss": 0.5532,
+      "step": 17049
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.926642832707319e-06,
+      "loss": 0.5396,
+      "step": 17050
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9256255794655464e-06,
+      "loss": 0.5796,
+      "step": 17051
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9246084727516034e-06,
+      "loss": 0.6292,
+      "step": 17052
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9235915125865512e-06,
+      "loss": 0.551,
+      "step": 17053
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9225746989914526e-06,
+      "loss": 0.572,
+      "step": 17054
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9215580319873736e-06,
+      "loss": 0.5237,
+      "step": 17055
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.920541511595366e-06,
+      "loss": 0.5371,
+      "step": 17056
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.919525137836491e-06,
+      "loss": 0.5649,
+      "step": 17057
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9185089107317933e-06,
+      "loss": 0.4634,
+      "step": 17058
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9174928303023286e-06,
+      "loss": 0.5676,
+      "step": 17059
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.916476896569135e-06,
+      "loss": 0.5605,
+      "step": 17060
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.915461109553259e-06,
+      "loss": 0.5818,
+      "step": 17061
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9144454692757428e-06,
+      "loss": 0.5261,
+      "step": 17062
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9134299757576167e-06,
+      "loss": 0.5247,
+      "step": 17063
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9124146290199174e-06,
+      "loss": 0.512,
+      "step": 17064
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9113994290836777e-06,
+      "loss": 0.5981,
+      "step": 17065
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9103843759699204e-06,
+      "loss": 0.5408,
+      "step": 17066
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9093694696996744e-06,
+      "loss": 0.6042,
+      "step": 17067
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.908354710293958e-06,
+      "loss": 0.5818,
+      "step": 17068
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9073400977737864e-06,
+      "loss": 0.605,
+      "step": 17069
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9063256321601805e-06,
+      "loss": 0.5637,
+      "step": 17070
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9053113134741472e-06,
+      "loss": 0.5505,
+      "step": 17071
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.904297141736702e-06,
+      "loss": 0.4758,
+      "step": 17072
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9032831169688426e-06,
+      "loss": 0.5857,
+      "step": 17073
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9022692391915807e-06,
+      "loss": 0.5793,
+      "step": 17074
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9012555084259085e-06,
+      "loss": 0.615,
+      "step": 17075
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.900241924692829e-06,
+      "loss": 0.488,
+      "step": 17076
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8992284880133316e-06,
+      "loss": 0.4951,
+      "step": 17077
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8982151984084104e-06,
+      "loss": 0.5698,
+      "step": 17078
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8972020558990487e-06,
+      "loss": 0.5703,
+      "step": 17079
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.896189060506238e-06,
+      "loss": 0.5688,
+      "step": 17080
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.895176212250953e-06,
+      "loss": 0.5991,
+      "step": 17081
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.894163511154178e-06,
+      "loss": 0.4988,
+      "step": 17082
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8931509572368854e-06,
+      "loss": 0.5325,
+      "step": 17083
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.892138550520046e-06,
+      "loss": 0.5732,
+      "step": 17084
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8911262910246308e-06,
+      "loss": 0.5742,
+      "step": 17085
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8901141787716102e-06,
+      "loss": 0.5693,
+      "step": 17086
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8891022137819413e-06,
+      "loss": 0.5686,
+      "step": 17087
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8880903960765873e-06,
+      "loss": 0.615,
+      "step": 17088
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8870787256765075e-06,
+      "loss": 0.5513,
+      "step": 17089
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8860672026026525e-06,
+      "loss": 0.5535,
+      "step": 17090
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8850558268759767e-06,
+      "loss": 0.6021,
+      "step": 17091
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.884044598517425e-06,
+      "loss": 0.5728,
+      "step": 17092
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8830335175479453e-06,
+      "loss": 0.5601,
+      "step": 17093
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8820225839884764e-06,
+      "loss": 0.5879,
+      "step": 17094
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8810117978599616e-06,
+      "loss": 0.5688,
+      "step": 17095
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.880001159183331e-06,
+      "loss": 0.51,
+      "step": 17096
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.878990667979524e-06,
+      "loss": 0.5708,
+      "step": 17097
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.877980324269467e-06,
+      "loss": 0.594,
+      "step": 17098
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8769701280740834e-06,
+      "loss": 0.625,
+      "step": 17099
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.875960079414304e-06,
+      "loss": 0.5151,
+      "step": 17100
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8749501783110423e-06,
+      "loss": 0.5452,
+      "step": 17101
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.873940424785223e-06,
+      "loss": 0.5381,
+      "step": 17102
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8729308188577532e-06,
+      "loss": 0.5532,
+      "step": 17103
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.871921360549551e-06,
+      "loss": 0.5454,
+      "step": 17104
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.870912049881518e-06,
+      "loss": 0.4983,
+      "step": 17105
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.869902886874567e-06,
+      "loss": 0.6555,
+      "step": 17106
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.868893871549594e-06,
+      "loss": 0.5994,
+      "step": 17107
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8678850039275032e-06,
+      "loss": 0.5774,
+      "step": 17108
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.866876284029184e-06,
+      "loss": 0.5337,
+      "step": 17109
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8658677118755384e-06,
+      "loss": 0.563,
+      "step": 17110
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8648592874874494e-06,
+      "loss": 0.668,
+      "step": 17111
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8638510108858052e-06,
+      "loss": 0.5481,
+      "step": 17112
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8628428820914934e-06,
+      "loss": 0.5408,
+      "step": 17113
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.86183490112539e-06,
+      "loss": 0.4967,
+      "step": 17114
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.860827068008378e-06,
+      "loss": 0.4976,
+      "step": 17115
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8598193827613275e-06,
+      "loss": 0.4705,
+      "step": 17116
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8588118454051127e-06,
+      "loss": 0.5068,
+      "step": 17117
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8578044559606e-06,
+      "loss": 0.5166,
+      "step": 17118
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8567972144486587e-06,
+      "loss": 0.5859,
+      "step": 17119
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.855790120890146e-06,
+      "loss": 0.5444,
+      "step": 17120
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8547831753059274e-06,
+      "loss": 0.564,
+      "step": 17121
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8537763777168527e-06,
+      "loss": 0.4871,
+      "step": 17122
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8527697281437816e-06,
+      "loss": 0.5674,
+      "step": 17123
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.851763226607558e-06,
+      "loss": 0.5586,
+      "step": 17124
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8507568731290358e-06,
+      "loss": 0.5752,
+      "step": 17125
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8497506677290553e-06,
+      "loss": 0.5593,
+      "step": 17126
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8487446104284544e-06,
+      "loss": 0.5532,
+      "step": 17127
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8477387012480785e-06,
+      "loss": 0.5286,
+      "step": 17128
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.846732940208754e-06,
+      "loss": 0.5369,
+      "step": 17129
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.84572732733132e-06,
+      "loss": 0.5281,
+      "step": 17130
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8447218626366003e-06,
+      "loss": 0.5496,
+      "step": 17131
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.843716546145425e-06,
+      "loss": 0.5427,
+      "step": 17132
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8427113778786107e-06,
+      "loss": 0.5825,
+      "step": 17133
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8417063578569847e-06,
+      "loss": 0.5801,
+      "step": 17134
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.840701486101355e-06,
+      "loss": 0.6711,
+      "step": 17135
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8396967626325435e-06,
+      "loss": 0.5591,
+      "step": 17136
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8386921874713524e-06,
+      "loss": 0.5671,
+      "step": 17137
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.837687760638592e-06,
+      "loss": 0.5913,
+      "step": 17138
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8366834821550713e-06,
+      "loss": 0.5178,
+      "step": 17139
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.835679352041585e-06,
+      "loss": 0.5012,
+      "step": 17140
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8346753703189354e-06,
+      "loss": 0.4878,
+      "step": 17141
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8336715370079126e-06,
+      "loss": 0.5327,
+      "step": 17142
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.832667852129315e-06,
+      "loss": 0.5706,
+      "step": 17143
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8316643157039235e-06,
+      "loss": 0.4712,
+      "step": 17144
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.830660927752532e-06,
+      "loss": 0.5525,
+      "step": 17145
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.829657688295916e-06,
+      "loss": 0.6042,
+      "step": 17146
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.828654597354862e-06,
+      "loss": 0.571,
+      "step": 17147
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8276516549501387e-06,
+      "loss": 0.55,
+      "step": 17148
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8266488611025277e-06,
+      "loss": 0.6021,
+      "step": 17149
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8256462158327915e-06,
+      "loss": 0.5435,
+      "step": 17150
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.824643719161705e-06,
+      "loss": 0.5637,
+      "step": 17151
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8236413711100243e-06,
+      "loss": 0.5952,
+      "step": 17152
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.822639171698519e-06,
+      "loss": 0.5142,
+      "step": 17153
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8216371209479398e-06,
+      "loss": 0.5129,
+      "step": 17154
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8206352188790463e-06,
+      "loss": 0.6323,
+      "step": 17155
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8196334655125903e-06,
+      "loss": 0.5173,
+      "step": 17156
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8186318608693162e-06,
+      "loss": 0.5139,
+      "step": 17157
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8176304049699752e-06,
+      "loss": 0.5398,
+      "step": 17158
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.816629097835304e-06,
+      "loss": 0.5396,
+      "step": 17159
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8156279394860487e-06,
+      "loss": 0.6069,
+      "step": 17160
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.81462692994294e-06,
+      "loss": 0.5383,
+      "step": 17161
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8136260692267135e-06,
+      "loss": 0.5271,
+      "step": 17162
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.812625357358103e-06,
+      "loss": 0.5098,
+      "step": 17163
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.81162479435783e-06,
+      "loss": 0.6152,
+      "step": 17164
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8106243802466206e-06,
+      "loss": 0.5969,
+      "step": 17165
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8096241150452e-06,
+      "loss": 0.5901,
+      "step": 17166
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8086239987742793e-06,
+      "loss": 0.5596,
+      "step": 17167
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8076240314545798e-06,
+      "loss": 0.5801,
+      "step": 17168
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.806624213106808e-06,
+      "loss": 0.501,
+      "step": 17169
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.805624543751677e-06,
+      "loss": 0.6204,
+      "step": 17170
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8046250234098916e-06,
+      "loss": 0.564,
+      "step": 17171
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.803625652102149e-06,
+      "loss": 0.5774,
+      "step": 17172
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8026264298491557e-06,
+      "loss": 0.5081,
+      "step": 17173
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.801627356671601e-06,
+      "loss": 0.5249,
+      "step": 17174
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.800628432590187e-06,
+      "loss": 0.5457,
+      "step": 17175
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7996296576255954e-06,
+      "loss": 0.5605,
+      "step": 17176
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.798631031798521e-06,
+      "loss": 0.5383,
+      "step": 17177
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7976325551296402e-06,
+      "loss": 0.5107,
+      "step": 17178
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7966342276396397e-06,
+      "loss": 0.6445,
+      "step": 17179
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.795636049349193e-06,
+      "loss": 0.6123,
+      "step": 17180
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.79463802027898e-06,
+      "loss": 0.6689,
+      "step": 17181
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.793640140449667e-06,
+      "loss": 0.5225,
+      "step": 17182
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7926424098819283e-06,
+      "loss": 0.5449,
+      "step": 17183
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7916448285964214e-06,
+      "loss": 0.5417,
+      "step": 17184
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.790647396613818e-06,
+      "loss": 0.5715,
+      "step": 17185
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7896501139547695e-06,
+      "loss": 0.5312,
+      "step": 17186
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.788652980639938e-06,
+      "loss": 0.4895,
+      "step": 17187
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.78765599668997e-06,
+      "loss": 0.5229,
+      "step": 17188
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.78665916212552e-06,
+      "loss": 0.511,
+      "step": 17189
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.785662476967238e-06,
+      "loss": 0.5171,
+      "step": 17190
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7846659412357613e-06,
+      "loss": 0.5471,
+      "step": 17191
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7836695549517356e-06,
+      "loss": 0.5488,
+      "step": 17192
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7826733181357934e-06,
+      "loss": 0.5439,
+      "step": 17193
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.781677230808576e-06,
+      "loss": 0.5708,
+      "step": 17194
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7806812929907078e-06,
+      "loss": 0.5547,
+      "step": 17195
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7796855047028237e-06,
+      "loss": 0.5347,
+      "step": 17196
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7786898659655428e-06,
+      "loss": 0.6172,
+      "step": 17197
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7776943767994924e-06,
+      "loss": 0.5764,
+      "step": 17198
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.776699037225287e-06,
+      "loss": 0.5261,
+      "step": 17199
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.775703847263548e-06,
+      "loss": 0.5696,
+      "step": 17200
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.774708806934885e-06,
+      "loss": 0.541,
+      "step": 17201
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7737139162599035e-06,
+      "loss": 0.6006,
+      "step": 17202
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7727191752592187e-06,
+      "loss": 0.5825,
+      "step": 17203
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7717245839534266e-06,
+      "loss": 0.5281,
+      "step": 17204
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.770730142363134e-06,
+      "loss": 0.5793,
+      "step": 17205
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7697358505089323e-06,
+      "loss": 0.5903,
+      "step": 17206
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.768741708411422e-06,
+      "loss": 0.5806,
+      "step": 17207
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7677477160911882e-06,
+      "loss": 0.6072,
+      "step": 17208
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.766753873568824e-06,
+      "loss": 0.5764,
+      "step": 17209
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7657601808649094e-06,
+      "loss": 0.5999,
+      "step": 17210
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7647666380000314e-06,
+      "loss": 0.6235,
+      "step": 17211
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.763773244994763e-06,
+      "loss": 0.5552,
+      "step": 17212
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.762780001869686e-06,
+      "loss": 0.5657,
+      "step": 17213
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7617869086453674e-06,
+      "loss": 0.6123,
+      "step": 17214
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.76079396534238e-06,
+      "loss": 0.5789,
+      "step": 17215
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.759801171981291e-06,
+      "loss": 0.5432,
+      "step": 17216
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7588085285826593e-06,
+      "loss": 0.6116,
+      "step": 17217
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.75781603516705e-06,
+      "loss": 0.5261,
+      "step": 17218
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.756823691755016e-06,
+      "loss": 0.4695,
+      "step": 17219
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.755831498367115e-06,
+      "loss": 0.5271,
+      "step": 17220
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7548394550238935e-06,
+      "loss": 0.6072,
+      "step": 17221
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.753847561745904e-06,
+      "loss": 0.6094,
+      "step": 17222
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7528558185536847e-06,
+      "loss": 0.5935,
+      "step": 17223
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.751864225467784e-06,
+      "loss": 0.5427,
+      "step": 17224
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7508727825087343e-06,
+      "loss": 0.5415,
+      "step": 17225
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.749881489697076e-06,
+      "loss": 0.5542,
+      "step": 17226
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7488903470533367e-06,
+      "loss": 0.5923,
+      "step": 17227
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7478993545980505e-06,
+      "loss": 0.5518,
+      "step": 17228
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.746908512351737e-06,
+      "loss": 0.5168,
+      "step": 17229
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7459178203349246e-06,
+      "loss": 0.6011,
+      "step": 17230
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7449272785681323e-06,
+      "loss": 0.5488,
+      "step": 17231
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.743936887071871e-06,
+      "loss": 0.605,
+      "step": 17232
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.742946645866661e-06,
+      "loss": 0.6255,
+      "step": 17233
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7419565549730074e-06,
+      "loss": 0.5842,
+      "step": 17234
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7409666144114233e-06,
+      "loss": 0.5869,
+      "step": 17235
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.739976824202406e-06,
+      "loss": 0.585,
+      "step": 17236
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.738987184366463e-06,
+      "loss": 0.5532,
+      "step": 17237
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.737997694924086e-06,
+      "loss": 0.5566,
+      "step": 17238
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.737008355895774e-06,
+      "loss": 0.5637,
+      "step": 17239
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.736019167302019e-06,
+      "loss": 0.5352,
+      "step": 17240
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7350301291633062e-06,
+      "loss": 0.5833,
+      "step": 17241
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7340412415001225e-06,
+      "loss": 0.5901,
+      "step": 17242
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7330525043329536e-06,
+      "loss": 0.5454,
+      "step": 17243
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7320639176822727e-06,
+      "loss": 0.5681,
+      "step": 17244
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7310754815685627e-06,
+      "loss": 0.5769,
+      "step": 17245
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.730087196012293e-06,
+      "loss": 0.5552,
+      "step": 17246
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7290990610339297e-06,
+      "loss": 0.4978,
+      "step": 17247
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7281110766539456e-06,
+      "loss": 0.5703,
+      "step": 17248
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7271232428927998e-06,
+      "loss": 0.6387,
+      "step": 17249
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.726135559770957e-06,
+      "loss": 0.5356,
+      "step": 17250
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.725148027308869e-06,
+      "loss": 0.5706,
+      "step": 17251
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7241606455269976e-06,
+      "loss": 0.5969,
+      "step": 17252
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7231734144457866e-06,
+      "loss": 0.564,
+      "step": 17253
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.722186334085689e-06,
+      "loss": 0.5354,
+      "step": 17254
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7211994044671466e-06,
+      "loss": 0.5496,
+      "step": 17255
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.720212625610604e-06,
+      "loss": 0.4866,
+      "step": 17256
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7192259975364953e-06,
+      "loss": 0.5796,
+      "step": 17257
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.718239520265262e-06,
+      "loss": 0.5889,
+      "step": 17258
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7172531938173306e-06,
+      "loss": 0.5659,
+      "step": 17259
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7162670182131357e-06,
+      "loss": 0.5298,
+      "step": 17260
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.715280993473102e-06,
+      "loss": 0.5686,
+      "step": 17261
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.714295119617647e-06,
+      "loss": 0.5566,
+      "step": 17262
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7133093966671965e-06,
+      "loss": 0.5862,
+      "step": 17263
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7123238246421677e-06,
+      "loss": 0.5488,
+      "step": 17264
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7113384035629697e-06,
+      "loss": 0.5266,
+      "step": 17265
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7103531334500155e-06,
+      "loss": 0.5054,
+      "step": 17266
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7093680143237157e-06,
+      "loss": 0.5459,
+      "step": 17267
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.708383046204468e-06,
+      "loss": 0.5723,
+      "step": 17268
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.707398229112681e-06,
+      "loss": 0.6323,
+      "step": 17269
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.706413563068745e-06,
+      "loss": 0.5786,
+      "step": 17270
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7054290480930613e-06,
+      "loss": 0.5098,
+      "step": 17271
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.704444684206017e-06,
+      "loss": 0.5022,
+      "step": 17272
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7034604714280044e-06,
+      "loss": 0.6003,
+      "step": 17273
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7024764097794044e-06,
+      "loss": 0.5054,
+      "step": 17274
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.701492499280606e-06,
+      "loss": 0.5442,
+      "step": 17275
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7005087399519836e-06,
+      "loss": 0.5312,
+      "step": 17276
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6995251318139126e-06,
+      "loss": 0.6021,
+      "step": 17277
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.69854167488677e-06,
+      "loss": 0.5295,
+      "step": 17278
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6975583691909203e-06,
+      "loss": 0.5674,
+      "step": 17279
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.696575214746736e-06,
+      "loss": 0.5303,
+      "step": 17280
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.695592211574575e-06,
+      "loss": 0.5725,
+      "step": 17281
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.694609359694803e-06,
+      "loss": 0.6157,
+      "step": 17282
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6936266591277726e-06,
+      "loss": 0.6077,
+      "step": 17283
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6926441098938426e-06,
+      "loss": 0.5789,
+      "step": 17284
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6916617120133603e-06,
+      "loss": 0.5889,
+      "step": 17285
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.690679465506676e-06,
+      "loss": 0.5403,
+      "step": 17286
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.689697370394132e-06,
+      "loss": 0.5989,
+      "step": 17287
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6887154266960735e-06,
+      "loss": 0.552,
+      "step": 17288
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.687733634432834e-06,
+      "loss": 0.502,
+      "step": 17289
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6867519936247542e-06,
+      "loss": 0.5527,
+      "step": 17290
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.685770504292161e-06,
+      "loss": 0.5476,
+      "step": 17291
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.684789166455387e-06,
+      "loss": 0.5747,
+      "step": 17292
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6838079801347606e-06,
+      "loss": 0.5359,
+      "step": 17293
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6828269453505973e-06,
+      "loss": 0.6116,
+      "step": 17294
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6818460621232246e-06,
+      "loss": 0.5691,
+      "step": 17295
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6808653304729517e-06,
+      "loss": 0.5681,
+      "step": 17296
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6798847504200986e-06,
+      "loss": 0.5671,
+      "step": 17297
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6789043219849708e-06,
+      "loss": 0.5708,
+      "step": 17298
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6779240451878784e-06,
+      "loss": 0.5557,
+      "step": 17299
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6769439200491222e-06,
+      "loss": 0.5525,
+      "step": 17300
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6759639465890074e-06,
+      "loss": 0.5947,
+      "step": 17301
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6749841248278263e-06,
+      "loss": 0.6106,
+      "step": 17302
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.67400445478588e-06,
+      "loss": 0.5635,
+      "step": 17303
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6730249364834524e-06,
+      "loss": 0.5249,
+      "step": 17304
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.672045569940839e-06,
+      "loss": 0.6116,
+      "step": 17305
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.671066355178321e-06,
+      "loss": 0.6223,
+      "step": 17306
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.670087292216179e-06,
+      "loss": 0.5071,
+      "step": 17307
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6691083810746954e-06,
+      "loss": 0.5754,
+      "step": 17308
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6681296217741416e-06,
+      "loss": 0.5571,
+      "step": 17309
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6671510143347957e-06,
+      "loss": 0.542,
+      "step": 17310
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6661725587769217e-06,
+      "loss": 0.5369,
+      "step": 17311
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.66519425512079e-06,
+      "loss": 0.5547,
+      "step": 17312
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.66421610338666e-06,
+      "loss": 0.5537,
+      "step": 17313
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6632381035947963e-06,
+      "loss": 0.5337,
+      "step": 17314
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6622602557654487e-06,
+      "loss": 0.6426,
+      "step": 17315
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6612825599188764e-06,
+      "loss": 0.5635,
+      "step": 17316
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6603050160753306e-06,
+      "loss": 0.5796,
+      "step": 17317
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6593276242550547e-06,
+      "loss": 0.5325,
+      "step": 17318
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6583503844782977e-06,
+      "loss": 0.5933,
+      "step": 17319
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.657373296765294e-06,
+      "loss": 0.4995,
+      "step": 17320
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6563963611362896e-06,
+      "loss": 0.5168,
+      "step": 17321
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.655419577611511e-06,
+      "loss": 0.5854,
+      "step": 17322
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6544429462111976e-06,
+      "loss": 0.585,
+      "step": 17323
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.65346646695557e-06,
+      "loss": 0.4839,
+      "step": 17324
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.652490139864861e-06,
+      "loss": 0.5098,
+      "step": 17325
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.651513964959287e-06,
+      "loss": 0.5647,
+      "step": 17326
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6505379422590714e-06,
+      "loss": 0.5964,
+      "step": 17327
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.649562071784425e-06,
+      "loss": 0.5735,
+      "step": 17328
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6485863535555667e-06,
+      "loss": 0.574,
+      "step": 17329
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.647610787592699e-06,
+      "loss": 0.5808,
+      "step": 17330
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6466353739160366e-06,
+      "loss": 0.5198,
+      "step": 17331
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6456601125457747e-06,
+      "loss": 0.5684,
+      "step": 17332
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6446850035021188e-06,
+      "loss": 0.5366,
+      "step": 17333
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6437100468052658e-06,
+      "loss": 0.5869,
+      "step": 17334
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6427352424754026e-06,
+      "loss": 0.5637,
+      "step": 17335
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6417605905327293e-06,
+      "loss": 0.5217,
+      "step": 17336
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.640786090997425e-06,
+      "loss": 0.6411,
+      "step": 17337
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6398117438896807e-06,
+      "loss": 0.5391,
+      "step": 17338
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6388375492296723e-06,
+      "loss": 0.5269,
+      "step": 17339
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6378635070375804e-06,
+      "loss": 0.521,
+      "step": 17340
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6368896173335824e-06,
+      "loss": 0.6077,
+      "step": 17341
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6359158801378436e-06,
+      "loss": 0.5422,
+      "step": 17342
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.634942295470537e-06,
+      "loss": 0.5847,
+      "step": 17343
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6339688633518303e-06,
+      "loss": 0.6174,
+      "step": 17344
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6329955838018783e-06,
+      "loss": 0.5427,
+      "step": 17345
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.632022456840847e-06,
+      "loss": 0.5588,
+      "step": 17346
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6310494824888875e-06,
+      "loss": 0.5872,
+      "step": 17347
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6300766607661575e-06,
+      "loss": 0.5955,
+      "step": 17348
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6291039916928028e-06,
+      "loss": 0.6016,
+      "step": 17349
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6281314752889676e-06,
+      "loss": 0.5967,
+      "step": 17350
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.627159111574801e-06,
+      "loss": 0.5125,
+      "step": 17351
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6261869005704376e-06,
+      "loss": 0.5833,
+      "step": 17352
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6252148422960187e-06,
+      "loss": 0.5237,
+      "step": 17353
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.624242936771674e-06,
+      "loss": 0.5334,
+      "step": 17354
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6232711840175383e-06,
+      "loss": 0.5818,
+      "step": 17355
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6222995840537345e-06,
+      "loss": 0.5674,
+      "step": 17356
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.621328136900393e-06,
+      "loss": 0.553,
+      "step": 17357
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.620356842577627e-06,
+      "loss": 0.5823,
+      "step": 17358
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6193857011055623e-06,
+      "loss": 0.624,
+      "step": 17359
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.618414712504307e-06,
+      "loss": 0.4661,
+      "step": 17360
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.617443876793977e-06,
+      "loss": 0.5688,
+      "step": 17361
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6164731939946775e-06,
+      "loss": 0.6138,
+      "step": 17362
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.615502664126518e-06,
+      "loss": 0.5403,
+      "step": 17363
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.614532287209596e-06,
+      "loss": 0.5315,
+      "step": 17364
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6135620632640146e-06,
+      "loss": 0.5188,
+      "step": 17365
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.612591992309864e-06,
+      "loss": 0.4792,
+      "step": 17366
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.611622074367245e-06,
+      "loss": 0.5588,
+      "step": 17367
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6106523094562373e-06,
+      "loss": 0.5698,
+      "step": 17368
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6096826975969338e-06,
+      "loss": 0.5505,
+      "step": 17369
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.608713238809417e-06,
+      "loss": 0.5396,
+      "step": 17370
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.607743933113764e-06,
+      "loss": 0.6211,
+      "step": 17371
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.606774780530056e-06,
+      "loss": 0.5947,
+      "step": 17372
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.60580578107836e-06,
+      "loss": 0.5876,
+      "step": 17373
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6048369347787527e-06,
+      "loss": 0.5515,
+      "step": 17374
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6038682416512972e-06,
+      "loss": 0.5432,
+      "step": 17375
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6028997017160607e-06,
+      "loss": 0.5847,
+      "step": 17376
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6019313149931003e-06,
+      "loss": 0.6023,
+      "step": 17377
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6009630815024777e-06,
+      "loss": 0.5254,
+      "step": 17378
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.599995001264246e-06,
+      "loss": 0.6465,
+      "step": 17379
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5990270742984535e-06,
+      "loss": 0.5847,
+      "step": 17380
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.598059300625153e-06,
+      "loss": 0.5071,
+      "step": 17381
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5970916802643852e-06,
+      "loss": 0.5361,
+      "step": 17382
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5961242132361965e-06,
+      "loss": 0.6418,
+      "step": 17383
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5951568995606204e-06,
+      "loss": 0.5024,
+      "step": 17384
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5941897392576977e-06,
+      "loss": 0.533,
+      "step": 17385
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5932227323474558e-06,
+      "loss": 0.5769,
+      "step": 17386
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5922558788499285e-06,
+      "loss": 0.575,
+      "step": 17387
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.591289178785137e-06,
+      "loss": 0.522,
+      "step": 17388
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.590322632173109e-06,
+      "loss": 0.5806,
+      "step": 17389
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.589356239033858e-06,
+      "loss": 0.4717,
+      "step": 17390
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5883899993874074e-06,
+      "loss": 0.5278,
+      "step": 17391
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5874239132537637e-06,
+      "loss": 0.552,
+      "step": 17392
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5864579806529398e-06,
+      "loss": 0.5239,
+      "step": 17393
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5854922016049455e-06,
+      "loss": 0.4934,
+      "step": 17394
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.58452657612978e-06,
+      "loss": 0.5632,
+      "step": 17395
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5835611042474485e-06,
+      "loss": 0.5894,
+      "step": 17396
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5825957859779414e-06,
+      "loss": 0.5627,
+      "step": 17397
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5816306213412602e-06,
+      "loss": 0.5894,
+      "step": 17398
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.58066561035739e-06,
+      "loss": 0.6145,
+      "step": 17399
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5797007530463235e-06,
+      "loss": 0.5806,
+      "step": 17400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.578736049428041e-06,
+      "loss": 0.4851,
+      "step": 17401
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5777714995225276e-06,
+      "loss": 0.5596,
+      "step": 17402
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5768071033497578e-06,
+      "loss": 0.5762,
+      "step": 17403
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5758428609297106e-06,
+      "loss": 0.5083,
+      "step": 17404
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.574878772282353e-06,
+      "loss": 0.5771,
+      "step": 17405
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.57391483742766e-06,
+      "loss": 0.5906,
+      "step": 17406
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.57295105638559e-06,
+      "loss": 0.4827,
+      "step": 17407
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5719874291761114e-06,
+      "loss": 0.5281,
+      "step": 17408
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.571023955819182e-06,
+      "loss": 0.5325,
+      "step": 17409
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5700606363347526e-06,
+      "loss": 0.5474,
+      "step": 17410
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.569097470742783e-06,
+      "loss": 0.5994,
+      "step": 17411
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5681344590632164e-06,
+      "loss": 0.5413,
+      "step": 17412
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.567171601316005e-06,
+      "loss": 0.5598,
+      "step": 17413
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5662088975210863e-06,
+      "loss": 0.5649,
+      "step": 17414
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5652463476984067e-06,
+      "loss": 0.5679,
+      "step": 17415
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5642839518678976e-06,
+      "loss": 0.543,
+      "step": 17416
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5633217100494924e-06,
+      "loss": 0.5735,
+      "step": 17417
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5623596222631275e-06,
+      "loss": 0.5679,
+      "step": 17418
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.561397688528724e-06,
+      "loss": 0.5474,
+      "step": 17419
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.560435908866208e-06,
+      "loss": 0.605,
+      "step": 17420
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.559474283295502e-06,
+      "loss": 0.5681,
+      "step": 17421
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.558512811836521e-06,
+      "loss": 0.5808,
+      "step": 17422
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5575514945091827e-06,
+      "loss": 0.5247,
+      "step": 17423
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5565903313333972e-06,
+      "loss": 0.5227,
+      "step": 17424
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5556293223290685e-06,
+      "loss": 0.6384,
+      "step": 17425
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5546684675161073e-06,
+      "loss": 0.6003,
+      "step": 17426
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5537077669144096e-06,
+      "loss": 0.4861,
+      "step": 17427
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5527472205438798e-06,
+      "loss": 0.5508,
+      "step": 17428
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5517868284244075e-06,
+      "loss": 0.5732,
+      "step": 17429
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.55082659057589e-06,
+      "loss": 0.5669,
+      "step": 17430
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5498665070182117e-06,
+      "loss": 0.5383,
+      "step": 17431
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.548906577771264e-06,
+      "loss": 0.5603,
+      "step": 17432
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5479468028549213e-06,
+      "loss": 0.5466,
+      "step": 17433
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5469871822890714e-06,
+      "loss": 0.5354,
+      "step": 17434
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.546027716093584e-06,
+      "loss": 0.5591,
+      "step": 17435
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5450684042883365e-06,
+      "loss": 0.4517,
+      "step": 17436
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5441092468931948e-06,
+      "loss": 0.5413,
+      "step": 17437
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5431502439280297e-06,
+      "loss": 0.5298,
+      "step": 17438
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.542191395412702e-06,
+      "loss": 0.594,
+      "step": 17439
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.541232701367069e-06,
+      "loss": 0.5491,
+      "step": 17440
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5402741618109907e-06,
+      "loss": 0.5007,
+      "step": 17441
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5393157767643228e-06,
+      "loss": 0.521,
+      "step": 17442
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.538357546246911e-06,
+      "loss": 0.542,
+      "step": 17443
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5373994702786087e-06,
+      "loss": 0.5791,
+      "step": 17444
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5364415488792536e-06,
+      "loss": 0.5505,
+      "step": 17445
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5354837820686905e-06,
+      "loss": 0.4685,
+      "step": 17446
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.534526169866759e-06,
+      "loss": 0.501,
+      "step": 17447
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5335687122932875e-06,
+      "loss": 0.5798,
+      "step": 17448
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.532611409368114e-06,
+      "loss": 0.5779,
+      "step": 17449
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5316542611110607e-06,
+      "loss": 0.5903,
+      "step": 17450
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.530697267541957e-06,
+      "loss": 0.5625,
+      "step": 17451
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5297404286806204e-06,
+      "loss": 0.49,
+      "step": 17452
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5287837445468755e-06,
+      "loss": 0.6353,
+      "step": 17453
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5278272151605334e-06,
+      "loss": 0.5452,
+      "step": 17454
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.526870840541403e-06,
+      "loss": 0.6011,
+      "step": 17455
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.525914620709301e-06,
+      "loss": 0.533,
+      "step": 17456
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.524958555684026e-06,
+      "loss": 0.5911,
+      "step": 17457
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.524002645485386e-06,
+      "loss": 0.5293,
+      "step": 17458
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.523046890133175e-06,
+      "loss": 0.5779,
+      "step": 17459
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.522091289647195e-06,
+      "loss": 0.5488,
+      "step": 17460
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.521135844047232e-06,
+      "loss": 0.5486,
+      "step": 17461
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.520180553353082e-06,
+      "loss": 0.5232,
+      "step": 17462
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5192254175845266e-06,
+      "loss": 0.5178,
+      "step": 17463
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.518270436761353e-06,
+      "loss": 0.4905,
+      "step": 17464
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5173156109033383e-06,
+      "loss": 0.5366,
+      "step": 17465
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5163609400302615e-06,
+      "loss": 0.6348,
+      "step": 17466
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5154064241618927e-06,
+      "loss": 0.6555,
+      "step": 17467
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.514452063318008e-06,
+      "loss": 0.5337,
+      "step": 17468
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5134978575183676e-06,
+      "loss": 0.5588,
+      "step": 17469
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.51254380678274e-06,
+      "loss": 0.5647,
+      "step": 17470
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.511589911130887e-06,
+      "loss": 0.6021,
+      "step": 17471
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.510636170582562e-06,
+      "loss": 0.5571,
+      "step": 17472
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5096825851575245e-06,
+      "loss": 0.7058,
+      "step": 17473
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.50872915487552e-06,
+      "loss": 0.6099,
+      "step": 17474
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.507775879756301e-06,
+      "loss": 0.5618,
+      "step": 17475
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.506822759819608e-06,
+      "loss": 0.603,
+      "step": 17476
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.505869795085187e-06,
+      "loss": 0.5637,
+      "step": 17477
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5049169855727718e-06,
+      "loss": 0.4944,
+      "step": 17478
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5039643313021023e-06,
+      "loss": 0.5005,
+      "step": 17479
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5030118322929054e-06,
+      "loss": 0.5098,
+      "step": 17480
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.502059488564914e-06,
+      "loss": 0.6128,
+      "step": 17481
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.501107300137848e-06,
+      "loss": 0.5344,
+      "step": 17482
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.500155267031438e-06,
+      "loss": 0.4905,
+      "step": 17483
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4992033892653966e-06,
+      "loss": 0.5476,
+      "step": 17484
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4982516668594382e-06,
+      "loss": 0.564,
+      "step": 17485
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4973000998332808e-06,
+      "loss": 0.6104,
+      "step": 17486
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.496348688206629e-06,
+      "loss": 0.5071,
+      "step": 17487
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4953974319991926e-06,
+      "loss": 0.6008,
+      "step": 17488
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.494446331230671e-06,
+      "loss": 0.5476,
+      "step": 17489
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4934953859207676e-06,
+      "loss": 0.489,
+      "step": 17490
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4925445960891746e-06,
+      "loss": 0.5896,
+      "step": 17491
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4915939617555907e-06,
+      "loss": 0.5562,
+      "step": 17492
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4906434829396987e-06,
+      "loss": 0.5549,
+      "step": 17493
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4896931596611896e-06,
+      "loss": 0.574,
+      "step": 17494
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.48874299193975e-06,
+      "loss": 0.5605,
+      "step": 17495
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.487792979795054e-06,
+      "loss": 0.531,
+      "step": 17496
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4868431232467817e-06,
+      "loss": 0.5671,
+      "step": 17497
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.485893422314609e-06,
+      "loss": 0.5632,
+      "step": 17498
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4849438770182056e-06,
+      "loss": 0.5366,
+      "step": 17499
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.483994487377235e-06,
+      "loss": 0.6028,
+      "step": 17500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4830452534113657e-06,
+      "loss": 0.5974,
+      "step": 17501
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.482096175140256e-06,
+      "loss": 0.5771,
+      "step": 17502
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.481147252583567e-06,
+      "loss": 0.5649,
+      "step": 17503
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.480198485760947e-06,
+      "loss": 0.5293,
+      "step": 17504
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4792498746920566e-06,
+      "loss": 0.5347,
+      "step": 17505
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4783014193965336e-06,
+      "loss": 0.533,
+      "step": 17506
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4773531198940328e-06,
+      "loss": 0.5303,
+      "step": 17507
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4764049762041874e-06,
+      "loss": 0.5625,
+      "step": 17508
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4754569883466416e-06,
+      "loss": 0.5217,
+      "step": 17509
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.474509156341026e-06,
+      "loss": 0.6016,
+      "step": 17510
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4735614802069784e-06,
+      "loss": 0.5823,
+      "step": 17511
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4726139599641206e-06,
+      "loss": 0.5215,
+      "step": 17512
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.471666595632084e-06,
+      "loss": 0.5396,
+      "step": 17513
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.470719387230489e-06,
+      "loss": 0.6042,
+      "step": 17514
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4697723347789517e-06,
+      "loss": 0.6328,
+      "step": 17515
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4688254382970923e-06,
+      "loss": 0.5759,
+      "step": 17516
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.467878697804519e-06,
+      "loss": 0.5251,
+      "step": 17517
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4669321133208425e-06,
+      "loss": 0.584,
+      "step": 17518
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4659856848656727e-06,
+      "loss": 0.5925,
+      "step": 17519
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4650394124586073e-06,
+      "loss": 0.5896,
+      "step": 17520
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.464093296119252e-06,
+      "loss": 0.4976,
+      "step": 17521
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4631473358671953e-06,
+      "loss": 0.5374,
+      "step": 17522
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.462201531722036e-06,
+      "loss": 0.5818,
+      "step": 17523
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.461255883703364e-06,
+      "loss": 0.4612,
+      "step": 17524
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4603103918307624e-06,
+      "loss": 0.5447,
+      "step": 17525
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.45936505612382e-06,
+      "loss": 0.5969,
+      "step": 17526
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.458419876602114e-06,
+      "loss": 0.6221,
+      "step": 17527
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4574748532852188e-06,
+      "loss": 0.5581,
+      "step": 17528
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.456529986192713e-06,
+      "loss": 0.5259,
+      "step": 17529
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.455585275344162e-06,
+      "loss": 0.5208,
+      "step": 17530
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4546407207591394e-06,
+      "loss": 0.5569,
+      "step": 17531
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.453696322457203e-06,
+      "loss": 0.6284,
+      "step": 17532
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4527520804579186e-06,
+      "loss": 0.5369,
+      "step": 17533
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4518079947808394e-06,
+      "loss": 0.5647,
+      "step": 17534
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.450864065445525e-06,
+      "loss": 0.5173,
+      "step": 17535
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4499202924715203e-06,
+      "loss": 0.5696,
+      "step": 17536
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4489766758783795e-06,
+      "loss": 0.5186,
+      "step": 17537
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4480332156856414e-06,
+      "loss": 0.6084,
+      "step": 17538
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.447089911912853e-06,
+      "loss": 0.5413,
+      "step": 17539
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4461467645795467e-06,
+      "loss": 0.6201,
+      "step": 17540
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4452037737052624e-06,
+      "loss": 0.5164,
+      "step": 17541
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4442609393095316e-06,
+      "loss": 0.4513,
+      "step": 17542
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4433182614118766e-06,
+      "loss": 0.5754,
+      "step": 17543
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.442375740031827e-06,
+      "loss": 0.5149,
+      "step": 17544
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4414333751889075e-06,
+      "loss": 0.5615,
+      "step": 17545
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4404911669026323e-06,
+      "loss": 0.6521,
+      "step": 17546
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4395491151925167e-06,
+      "loss": 0.6235,
+      "step": 17547
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4386072200780784e-06,
+      "loss": 0.5439,
+      "step": 17548
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4376654815788204e-06,
+      "loss": 0.5688,
+      "step": 17549
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4367238997142528e-06,
+      "loss": 0.5295,
+      "step": 17550
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4357824745038726e-06,
+      "loss": 0.5452,
+      "step": 17551
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4348412059671865e-06,
+      "loss": 0.5454,
+      "step": 17552
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4339000941236836e-06,
+      "loss": 0.575,
+      "step": 17553
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.432959138992862e-06,
+      "loss": 0.5212,
+      "step": 17554
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4320183405942057e-06,
+      "loss": 0.6489,
+      "step": 17555
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4310776989472074e-06,
+      "loss": 0.5032,
+      "step": 17556
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4301372140713464e-06,
+      "loss": 0.5193,
+      "step": 17557
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4291968859861005e-06,
+      "loss": 0.5232,
+      "step": 17558
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4282567147109503e-06,
+      "loss": 0.5874,
+      "step": 17559
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4273167002653653e-06,
+      "loss": 0.5193,
+      "step": 17560
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4263768426688204e-06,
+      "loss": 0.564,
+      "step": 17561
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4254371419407763e-06,
+      "loss": 0.4375,
+      "step": 17562
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4244975981007023e-06,
+      "loss": 0.6016,
+      "step": 17563
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4235582111680544e-06,
+      "loss": 0.5952,
+      "step": 17564
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4226189811622937e-06,
+      "loss": 0.5833,
+      "step": 17565
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4216799081028688e-06,
+      "loss": 0.6455,
+      "step": 17566
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.420740992009235e-06,
+      "loss": 0.5015,
+      "step": 17567
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4198022329008362e-06,
+      "loss": 0.5757,
+      "step": 17568
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.41886363079712e-06,
+      "loss": 0.5305,
+      "step": 17569
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4179251857175224e-06,
+      "loss": 0.5764,
+      "step": 17570
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4169868976814834e-06,
+      "loss": 0.4927,
+      "step": 17571
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.416048766708441e-06,
+      "loss": 0.6099,
+      "step": 17572
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4151107928178185e-06,
+      "loss": 0.5222,
+      "step": 17573
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4141729760290523e-06,
+      "loss": 0.6355,
+      "step": 17574
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4132353163615584e-06,
+      "loss": 0.5242,
+      "step": 17575
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.412297813834765e-06,
+      "loss": 0.5723,
+      "step": 17576
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.411360468468085e-06,
+      "loss": 0.5027,
+      "step": 17577
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.410423280280938e-06,
+      "loss": 0.5996,
+      "step": 17578
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4094862492927295e-06,
+      "loss": 0.5315,
+      "step": 17579
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.408549375522875e-06,
+      "loss": 0.5703,
+      "step": 17580
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4076126589907723e-06,
+      "loss": 0.5691,
+      "step": 17581
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.406676099715828e-06,
+      "loss": 0.6726,
+      "step": 17582
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4057396977174373e-06,
+      "loss": 0.5691,
+      "step": 17583
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.404803453015e-06,
+      "loss": 0.5806,
+      "step": 17584
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.403867365627901e-06,
+      "loss": 0.6338,
+      "step": 17585
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4029314355755374e-06,
+      "loss": 0.5266,
+      "step": 17586
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.40199566287729e-06,
+      "loss": 0.5159,
+      "step": 17587
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.401060047552538e-06,
+      "loss": 0.5828,
+      "step": 17588
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4001245896206658e-06,
+      "loss": 0.5012,
+      "step": 17589
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.399189289101045e-06,
+      "loss": 0.5122,
+      "step": 17590
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3982541460130526e-06,
+      "loss": 0.6494,
+      "step": 17591
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3973191603760517e-06,
+      "loss": 0.5693,
+      "step": 17592
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.396384332209415e-06,
+      "loss": 0.499,
+      "step": 17593
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3954496615324983e-06,
+      "loss": 0.5479,
+      "step": 17594
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3945151483646634e-06,
+      "loss": 0.615,
+      "step": 17595
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3935807927252707e-06,
+      "loss": 0.5474,
+      "step": 17596
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3926465946336675e-06,
+      "loss": 0.5815,
+      "step": 17597
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3917125541092066e-06,
+      "loss": 0.5476,
+      "step": 17598
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.390778671171231e-06,
+      "loss": 0.5447,
+      "step": 17599
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.389844945839086e-06,
+      "loss": 0.5898,
+      "step": 17600
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.388911378132114e-06,
+      "loss": 0.5469,
+      "step": 17601
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.387977968069648e-06,
+      "loss": 0.5603,
+      "step": 17602
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.38704471567102e-06,
+      "loss": 0.4934,
+      "step": 17603
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3861116209555647e-06,
+      "loss": 0.4724,
+      "step": 17604
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3851786839426015e-06,
+      "loss": 0.51,
+      "step": 17605
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.384245904651462e-06,
+      "loss": 0.5598,
+      "step": 17606
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3833132831014603e-06,
+      "loss": 0.5645,
+      "step": 17607
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.382380819311918e-06,
+      "loss": 0.5566,
+      "step": 17608
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3814485133021437e-06,
+      "loss": 0.5571,
+      "step": 17609
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.380516365091454e-06,
+      "loss": 0.5535,
+      "step": 17610
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3795843746991485e-06,
+      "loss": 0.5371,
+      "step": 17611
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3786525421445383e-06,
+      "loss": 0.646,
+      "step": 17612
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3777208674469176e-06,
+      "loss": 0.6138,
+      "step": 17613
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.37678935062559e-06,
+      "loss": 0.5774,
+      "step": 17614
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3758579916998438e-06,
+      "loss": 0.561,
+      "step": 17615
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.374926790688974e-06,
+      "loss": 0.5813,
+      "step": 17616
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3739957476122676e-06,
+      "loss": 0.5381,
+      "step": 17617
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3730648624890027e-06,
+      "loss": 0.6489,
+      "step": 17618
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3721341353384697e-06,
+      "loss": 0.5481,
+      "step": 17619
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.371203566179938e-06,
+      "loss": 0.5659,
+      "step": 17620
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3702731550326863e-06,
+      "loss": 0.5784,
+      "step": 17621
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3693429019159874e-06,
+      "loss": 0.4719,
+      "step": 17622
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.368412806849104e-06,
+      "loss": 0.5261,
+      "step": 17623
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.367482869851303e-06,
+      "loss": 0.5852,
+      "step": 17624
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3665530909418488e-06,
+      "loss": 0.5625,
+      "step": 17625
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.365623470139995e-06,
+      "loss": 0.6143,
+      "step": 17626
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3646940074650003e-06,
+      "loss": 0.5845,
+      "step": 17627
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.363764702936111e-06,
+      "loss": 0.5583,
+      "step": 17628
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3628355565725814e-06,
+      "loss": 0.5977,
+      "step": 17629
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3619065683936504e-06,
+      "loss": 0.4993,
+      "step": 17630
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3609777384185663e-06,
+      "loss": 0.6104,
+      "step": 17631
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.360049066666563e-06,
+      "loss": 0.5969,
+      "step": 17632
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.359120553156873e-06,
+      "loss": 0.5056,
+      "step": 17633
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.358192197908735e-06,
+      "loss": 0.5259,
+      "step": 17634
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3572640009413705e-06,
+      "loss": 0.531,
+      "step": 17635
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.356335962274012e-06,
+      "loss": 0.5784,
+      "step": 17636
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.355408081925874e-06,
+      "loss": 0.4644,
+      "step": 17637
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3544803599161827e-06,
+      "loss": 0.5923,
+      "step": 17638
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3535527962641456e-06,
+      "loss": 0.5359,
+      "step": 17639
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.352625390988983e-06,
+      "loss": 0.5112,
+      "step": 17640
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.351698144109896e-06,
+      "loss": 0.4932,
+      "step": 17641
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3507710556460982e-06,
+      "loss": 0.551,
+      "step": 17642
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3498441256167838e-06,
+      "loss": 0.5081,
+      "step": 17643
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.348917354041158e-06,
+      "loss": 0.4849,
+      "step": 17644
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3479907409384126e-06,
+      "loss": 0.5027,
+      "step": 17645
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.347064286327744e-06,
+      "loss": 0.5237,
+      "step": 17646
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3461379902283353e-06,
+      "loss": 0.564,
+      "step": 17647
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.345211852659377e-06,
+      "loss": 0.5667,
+      "step": 17648
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3442858736400533e-06,
+      "loss": 0.5669,
+      "step": 17649
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3433600531895386e-06,
+      "loss": 0.5833,
+      "step": 17650
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.342434391327014e-06,
+      "loss": 0.5825,
+      "step": 17651
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3415088880716463e-06,
+      "loss": 0.626,
+      "step": 17652
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3405835434426118e-06,
+      "loss": 0.646,
+      "step": 17653
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3396583574590705e-06,
+      "loss": 0.5269,
+      "step": 17654
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3387333301401905e-06,
+      "loss": 0.5334,
+      "step": 17655
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3378084615051255e-06,
+      "loss": 0.5889,
+      "step": 17656
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.336883751573038e-06,
+      "loss": 0.5781,
+      "step": 17657
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.335959200363076e-06,
+      "loss": 0.5671,
+      "step": 17658
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.335034807894394e-06,
+      "loss": 0.5566,
+      "step": 17659
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3341105741861336e-06,
+      "loss": 0.5962,
+      "step": 17660
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3331864992574427e-06,
+      "loss": 0.5879,
+      "step": 17661
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3322625831274594e-06,
+      "loss": 0.5127,
+      "step": 17662
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.331338825815316e-06,
+      "loss": 0.615,
+      "step": 17663
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3304152273401524e-06,
+      "loss": 0.4553,
+      "step": 17664
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3294917877210934e-06,
+      "loss": 0.6472,
+      "step": 17665
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.32856850697727e-06,
+      "loss": 0.5725,
+      "step": 17666
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3276453851278013e-06,
+      "loss": 0.583,
+      "step": 17667
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3267224221918126e-06,
+      "loss": 0.5089,
+      "step": 17668
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3257996181884148e-06,
+      "loss": 0.5288,
+      "step": 17669
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3248769731367282e-06,
+      "loss": 0.5642,
+      "step": 17670
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.323954487055856e-06,
+      "loss": 0.5745,
+      "step": 17671
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3230321599649085e-06,
+      "loss": 0.4932,
+      "step": 17672
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3221099918829926e-06,
+      "loss": 0.5232,
+      "step": 17673
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3211879828292026e-06,
+      "loss": 0.5305,
+      "step": 17674
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.320266132822642e-06,
+      "loss": 0.5842,
+      "step": 17675
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3193444418823976e-06,
+      "loss": 0.5488,
+      "step": 17676
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3184229100275667e-06,
+      "loss": 0.5889,
+      "step": 17677
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3175015372772303e-06,
+      "loss": 0.5981,
+      "step": 17678
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.316580323650478e-06,
+      "loss": 0.5095,
+      "step": 17679
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.315659269166385e-06,
+      "loss": 0.6038,
+      "step": 17680
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3147383738440343e-06,
+      "loss": 0.6519,
+      "step": 17681
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3138176377024945e-06,
+      "loss": 0.5681,
+      "step": 17682
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3128970607608413e-06,
+      "loss": 0.5229,
+      "step": 17683
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.311976643038136e-06,
+      "loss": 0.5667,
+      "step": 17684
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3110563845534504e-06,
+      "loss": 0.5776,
+      "step": 17685
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.310136285325838e-06,
+      "loss": 0.5449,
+      "step": 17686
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3092163453743633e-06,
+      "loss": 0.6199,
+      "step": 17687
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.308296564718072e-06,
+      "loss": 0.501,
+      "step": 17688
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.307376943376025e-06,
+      "loss": 0.592,
+      "step": 17689
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.30645748136726e-06,
+      "loss": 0.6011,
+      "step": 17690
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.305538178710831e-06,
+      "loss": 0.6294,
+      "step": 17691
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3046190354257726e-06,
+      "loss": 0.5312,
+      "step": 17692
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3037000515311226e-06,
+      "loss": 0.5396,
+      "step": 17693
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.302781227045918e-06,
+      "loss": 0.6309,
+      "step": 17694
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3018625619891875e-06,
+      "loss": 0.521,
+      "step": 17695
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3009440563799633e-06,
+      "loss": 0.5349,
+      "step": 17696
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3000257102372635e-06,
+      "loss": 0.5181,
+      "step": 17697
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.299107523580113e-06,
+      "loss": 0.6035,
+      "step": 17698
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.298189496427532e-06,
+      "loss": 0.5312,
+      "step": 17699
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.297271628798531e-06,
+      "loss": 0.48,
+      "step": 17700
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.296353920712122e-06,
+      "loss": 0.5527,
+      "step": 17701
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.295436372187316e-06,
+      "loss": 0.5784,
+      "step": 17702
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2945189832431147e-06,
+      "loss": 0.5298,
+      "step": 17703
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2936017538985222e-06,
+      "loss": 0.5479,
+      "step": 17704
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2926846841725313e-06,
+      "loss": 0.5317,
+      "step": 17705
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2917677740841447e-06,
+      "loss": 0.5444,
+      "step": 17706
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2908510236523486e-06,
+      "loss": 0.5554,
+      "step": 17707
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2899344328961294e-06,
+      "loss": 0.5762,
+      "step": 17708
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.289018001834478e-06,
+      "loss": 0.5769,
+      "step": 17709
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2881017304863683e-06,
+      "loss": 0.4851,
+      "step": 17710
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2871856188707876e-06,
+      "loss": 0.5864,
+      "step": 17711
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.286269667006702e-06,
+      "loss": 0.5944,
+      "step": 17712
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.285353874913091e-06,
+      "loss": 0.5483,
+      "step": 17713
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.284438242608915e-06,
+      "loss": 0.5464,
+      "step": 17714
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2835227701131467e-06,
+      "loss": 0.5957,
+      "step": 17715
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.282607457444742e-06,
+      "loss": 0.5105,
+      "step": 17716
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2816923046226634e-06,
+      "loss": 0.6199,
+      "step": 17717
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2807773116658626e-06,
+      "loss": 0.5356,
+      "step": 17718
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2798624785932954e-06,
+      "loss": 0.5291,
+      "step": 17719
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2789478054239044e-06,
+      "loss": 0.5137,
+      "step": 17720
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.278033292176641e-06,
+      "loss": 0.625,
+      "step": 17721
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2771189388704416e-06,
+      "loss": 0.5068,
+      "step": 17722
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.27620474552425e-06,
+      "loss": 0.6091,
+      "step": 17723
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2752907121569957e-06,
+      "loss": 0.5256,
+      "step": 17724
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2743768387876143e-06,
+      "loss": 0.5574,
+      "step": 17725
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.273463125435036e-06,
+      "loss": 0.5273,
+      "step": 17726
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.272549572118181e-06,
+      "loss": 0.6177,
+      "step": 17727
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.271636178855976e-06,
+      "loss": 0.623,
+      "step": 17728
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2707229456673364e-06,
+      "loss": 0.5791,
+      "step": 17729
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2698098725711802e-06,
+      "loss": 0.5066,
+      "step": 17730
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2688969595864162e-06,
+      "loss": 0.5156,
+      "step": 17731
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.267984206731957e-06,
+      "loss": 0.5745,
+      "step": 17732
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2670716140267036e-06,
+      "loss": 0.5383,
+      "step": 17733
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2661591814895622e-06,
+      "loss": 0.5393,
+      "step": 17734
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.265246909139428e-06,
+      "loss": 0.5356,
+      "step": 17735
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2643347969952003e-06,
+      "loss": 0.5845,
+      "step": 17736
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2634228450757688e-06,
+      "loss": 0.554,
+      "step": 17737
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2625110534000195e-06,
+      "loss": 0.5618,
+      "step": 17738
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.261599421986843e-06,
+      "loss": 0.5564,
+      "step": 17739
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2606879508551182e-06,
+      "loss": 0.6025,
+      "step": 17740
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2597766400237266e-06,
+      "loss": 0.5081,
+      "step": 17741
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.258865489511539e-06,
+      "loss": 0.6191,
+      "step": 17742
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2579544993374347e-06,
+      "loss": 0.5527,
+      "step": 17743
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.257043669520275e-06,
+      "loss": 0.5867,
+      "step": 17744
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2561330000789318e-06,
+      "loss": 0.5549,
+      "step": 17745
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2552224910322618e-06,
+      "loss": 0.6013,
+      "step": 17746
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2543121423991287e-06,
+      "loss": 0.6091,
+      "step": 17747
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2534019541983844e-06,
+      "loss": 0.5383,
+      "step": 17748
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.252491926448882e-06,
+      "loss": 0.5496,
+      "step": 17749
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2515820591694735e-06,
+      "loss": 0.5466,
+      "step": 17750
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.250672352379e-06,
+      "loss": 0.604,
+      "step": 17751
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2497628060963085e-06,
+      "loss": 0.5618,
+      "step": 17752
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2488534203402334e-06,
+      "loss": 0.4771,
+      "step": 17753
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2479441951296143e-06,
+      "loss": 0.5583,
+      "step": 17754
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2470351304832785e-06,
+      "loss": 0.5088,
+      "step": 17755
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.246126226420061e-06,
+      "loss": 0.5339,
+      "step": 17756
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2452174829587826e-06,
+      "loss": 0.5518,
+      "step": 17757
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2443089001182695e-06,
+      "loss": 0.5793,
+      "step": 17758
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2434004779173357e-06,
+      "loss": 0.604,
+      "step": 17759
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.242492216374803e-06,
+      "loss": 0.5486,
+      "step": 17760
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2415841155094775e-06,
+      "loss": 0.605,
+      "step": 17761
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2406761753401753e-06,
+      "loss": 0.5793,
+      "step": 17762
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.239768395885694e-06,
+      "loss": 0.6421,
+      "step": 17763
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.238860777164843e-06,
+      "loss": 0.5554,
+      "step": 17764
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2379533191964187e-06,
+      "loss": 0.5383,
+      "step": 17765
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.237046021999214e-06,
+      "loss": 0.5708,
+      "step": 17766
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2361388855920265e-06,
+      "loss": 0.5823,
+      "step": 17767
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.23523190999364e-06,
+      "loss": 0.5449,
+      "step": 17768
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2343250952228455e-06,
+      "loss": 0.561,
+      "step": 17769
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2334184412984194e-06,
+      "loss": 0.6201,
+      "step": 17770
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2325119482391466e-06,
+      "loss": 0.5864,
+      "step": 17771
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.231605616063799e-06,
+      "loss": 0.4497,
+      "step": 17772
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.230699444791151e-06,
+      "loss": 0.5073,
+      "step": 17773
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2297934344399696e-06,
+      "loss": 0.5945,
+      "step": 17774
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2288875850290203e-06,
+      "loss": 0.4963,
+      "step": 17775
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2279818965770705e-06,
+      "loss": 0.5684,
+      "step": 17776
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.227076369102873e-06,
+      "loss": 0.5146,
+      "step": 17777
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2261710026251857e-06,
+      "loss": 0.5315,
+      "step": 17778
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2252657971627654e-06,
+      "loss": 0.5078,
+      "step": 17779
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.224360752734357e-06,
+      "loss": 0.5457,
+      "step": 17780
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2234558693587027e-06,
+      "loss": 0.5522,
+      "step": 17781
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2225511470545523e-06,
+      "loss": 0.5574,
+      "step": 17782
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2216465858406378e-06,
+      "loss": 0.5449,
+      "step": 17783
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2207421857357013e-06,
+      "loss": 0.5017,
+      "step": 17784
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.21983794675847e-06,
+      "loss": 0.4976,
+      "step": 17785
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2189338689276773e-06,
+      "loss": 0.5762,
+      "step": 17786
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.218029952262044e-06,
+      "loss": 0.5779,
+      "step": 17787
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2171261967802983e-06,
+      "loss": 0.5522,
+      "step": 17788
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2162226025011522e-06,
+      "loss": 0.5112,
+      "step": 17789
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2153191694433296e-06,
+      "loss": 0.5334,
+      "step": 17790
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2144158976255346e-06,
+      "loss": 0.5173,
+      "step": 17791
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.213512787066483e-06,
+      "loss": 0.6304,
+      "step": 17792
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.212609837784875e-06,
+      "loss": 0.6382,
+      "step": 17793
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2117070497994174e-06,
+      "loss": 0.5303,
+      "step": 17794
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2108044231288084e-06,
+      "loss": 0.5508,
+      "step": 17795
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.209901957791739e-06,
+      "loss": 0.5291,
+      "step": 17796
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.208999653806907e-06,
+      "loss": 0.5867,
+      "step": 17797
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.208097511192998e-06,
+      "loss": 0.5823,
+      "step": 17798
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2071955299686974e-06,
+      "loss": 0.6365,
+      "step": 17799
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2062937101526927e-06,
+      "loss": 0.5361,
+      "step": 17800
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2053920517636563e-06,
+      "loss": 0.5369,
+      "step": 17801
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2044905548202668e-06,
+      "loss": 0.5857,
+      "step": 17802
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2035892193411988e-06,
+      "loss": 0.6062,
+      "step": 17803
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.202688045345116e-06,
+      "loss": 0.5662,
+      "step": 17804
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2017870328506885e-06,
+      "loss": 0.498,
+      "step": 17805
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2008861818765746e-06,
+      "loss": 0.5691,
+      "step": 17806
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1999854924414376e-06,
+      "loss": 0.584,
+      "step": 17807
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1990849645639278e-06,
+      "loss": 0.5432,
+      "step": 17808
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1981845982627024e-06,
+      "loss": 0.5703,
+      "step": 17809
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.197284393556408e-06,
+      "loss": 0.5474,
+      "step": 17810
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1963843504636883e-06,
+      "loss": 0.6165,
+      "step": 17811
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1954844690031885e-06,
+      "loss": 0.5991,
+      "step": 17812
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1945847491935434e-06,
+      "loss": 0.6204,
+      "step": 17813
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.193685191053394e-06,
+      "loss": 0.5103,
+      "step": 17814
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1927857946013675e-06,
+      "loss": 0.5398,
+      "step": 17815
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.191886559856097e-06,
+      "loss": 0.5527,
+      "step": 17816
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1909874868362015e-06,
+      "loss": 0.5383,
+      "step": 17817
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1900885755603095e-06,
+      "loss": 0.6248,
+      "step": 17818
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1891898260470356e-06,
+      "loss": 0.5276,
+      "step": 17819
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.188291238314999e-06,
+      "loss": 0.4785,
+      "step": 17820
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1873928123828057e-06,
+      "loss": 0.5586,
+      "step": 17821
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1864945482690726e-06,
+      "loss": 0.5105,
+      "step": 17822
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.185596445992396e-06,
+      "loss": 0.6106,
+      "step": 17823
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1846985055713856e-06,
+      "loss": 0.6135,
+      "step": 17824
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.183800727024633e-06,
+      "loss": 0.5496,
+      "step": 17825
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.182903110370738e-06,
+      "loss": 0.6355,
+      "step": 17826
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.182005655628293e-06,
+      "loss": 0.5225,
+      "step": 17827
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1811083628158823e-06,
+      "loss": 0.5835,
+      "step": 17828
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1802112319520965e-06,
+      "loss": 0.5493,
+      "step": 17829
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.179314263055512e-06,
+      "loss": 0.5481,
+      "step": 17830
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1784174561447126e-06,
+      "loss": 0.5591,
+      "step": 17831
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.177520811238267e-06,
+      "loss": 0.5427,
+      "step": 17832
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.176624328354755e-06,
+      "loss": 0.5466,
+      "step": 17833
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1757280075127364e-06,
+      "loss": 0.5759,
+      "step": 17834
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.174831848730784e-06,
+      "loss": 0.5444,
+      "step": 17835
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1739358520274524e-06,
+      "loss": 0.5535,
+      "step": 17836
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.173040017421306e-06,
+      "loss": 0.5898,
+      "step": 17837
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1721443449308954e-06,
+      "loss": 0.5989,
+      "step": 17838
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.171248834574775e-06,
+      "loss": 0.6758,
+      "step": 17839
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.170353486371494e-06,
+      "loss": 0.605,
+      "step": 17840
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.169458300339591e-06,
+      "loss": 0.5369,
+      "step": 17841
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1685632764976137e-06,
+      "loss": 0.5022,
+      "step": 17842
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.167668414864097e-06,
+      "loss": 0.5842,
+      "step": 17843
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1667737154575797e-06,
+      "loss": 0.5635,
+      "step": 17844
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1658791782965873e-06,
+      "loss": 0.6406,
+      "step": 17845
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1649848033996535e-06,
+      "loss": 0.5098,
+      "step": 17846
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1640905907852983e-06,
+      "loss": 0.4888,
+      "step": 17847
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1631965404720477e-06,
+      "loss": 0.6084,
+      "step": 17848
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.162302652478415e-06,
+      "loss": 0.5137,
+      "step": 17849
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1614089268229198e-06,
+      "loss": 0.5322,
+      "step": 17850
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1605153635240684e-06,
+      "loss": 0.4995,
+      "step": 17851
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1596219626003713e-06,
+      "loss": 0.509,
+      "step": 17852
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.158728724070335e-06,
+      "loss": 0.4636,
+      "step": 17853
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1578356479524552e-06,
+      "loss": 0.5015,
+      "step": 17854
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.156942734265237e-06,
+      "loss": 0.636,
+      "step": 17855
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1560499830271677e-06,
+      "loss": 0.5859,
+      "step": 17856
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.155157394256745e-06,
+      "loss": 0.5686,
+      "step": 17857
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1542649679724492e-06,
+      "loss": 0.4858,
+      "step": 17858
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.153372704192772e-06,
+      "loss": 0.5422,
+      "step": 17859
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.152480602936189e-06,
+      "loss": 0.6106,
+      "step": 17860
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1515886642211813e-06,
+      "loss": 0.5435,
+      "step": 17861
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1506968880662203e-06,
+      "loss": 0.584,
+      "step": 17862
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1498052744897803e-06,
+      "loss": 0.5085,
+      "step": 17863
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.148913823510325e-06,
+      "loss": 0.561,
+      "step": 17864
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1480225351463225e-06,
+      "loss": 0.6094,
+      "step": 17865
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1471314094162287e-06,
+      "loss": 0.6172,
+      "step": 17866
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.146240446338507e-06,
+      "loss": 0.5632,
+      "step": 17867
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.145349645931605e-06,
+      "loss": 0.5515,
+      "step": 17868
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1444590082139803e-06,
+      "loss": 0.4656,
+      "step": 17869
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.143568533204076e-06,
+      "loss": 0.5278,
+      "step": 17870
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1426782209203335e-06,
+      "loss": 0.5647,
+      "step": 17871
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1417880713812e-06,
+      "loss": 0.5659,
+      "step": 17872
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1408980846051063e-06,
+      "loss": 0.6055,
+      "step": 17873
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.140008260610492e-06,
+      "loss": 0.5027,
+      "step": 17874
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1391185994157816e-06,
+      "loss": 0.5642,
+      "step": 17875
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1382291010394062e-06,
+      "loss": 0.5186,
+      "step": 17876
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1373397654997905e-06,
+      "loss": 0.5129,
+      "step": 17877
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1364505928153502e-06,
+      "loss": 0.5869,
+      "step": 17878
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.135561583004505e-06,
+      "loss": 0.5757,
+      "step": 17879
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.134672736085672e-06,
+      "loss": 0.5938,
+      "step": 17880
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1337840520772545e-06,
+      "loss": 0.5417,
+      "step": 17881
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1328955309976664e-06,
+      "loss": 0.4773,
+      "step": 17882
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1320071728653037e-06,
+      "loss": 0.6252,
+      "step": 17883
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1311189776985743e-06,
+      "loss": 0.5781,
+      "step": 17884
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.13023094551587e-06,
+      "loss": 0.5378,
+      "step": 17885
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1293430763355827e-06,
+      "loss": 0.5994,
+      "step": 17886
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.128455370176108e-06,
+      "loss": 0.5554,
+      "step": 17887
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1275678270558263e-06,
+      "loss": 0.678,
+      "step": 17888
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1266804469931257e-06,
+      "loss": 0.605,
+      "step": 17889
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1257932300063822e-06,
+      "loss": 0.564,
+      "step": 17890
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1249061761139777e-06,
+      "loss": 0.5811,
+      "step": 17891
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1240192853342776e-06,
+      "loss": 0.5452,
+      "step": 17892
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.12313255768566e-06,
+      "loss": 0.5811,
+      "step": 17893
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1222459931864837e-06,
+      "loss": 0.5867,
+      "step": 17894
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1213595918551176e-06,
+      "loss": 0.594,
+      "step": 17895
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1204733537099153e-06,
+      "loss": 0.594,
+      "step": 17896
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.119587278769241e-06,
+      "loss": 0.5752,
+      "step": 17897
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.118701367051438e-06,
+      "loss": 0.4973,
+      "step": 17898
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1178156185748643e-06,
+      "loss": 0.543,
+      "step": 17899
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.116930033357859e-06,
+      "loss": 0.5701,
+      "step": 17900
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1160446114187717e-06,
+      "loss": 0.5261,
+      "step": 17901
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.115159352775935e-06,
+      "loss": 0.5828,
+      "step": 17902
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.114274257447687e-06,
+      "loss": 0.5286,
+      "step": 17903
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.113389325452364e-06,
+      "loss": 0.491,
+      "step": 17904
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.11250455680829e-06,
+      "loss": 0.6255,
+      "step": 17905
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1116199515337954e-06,
+      "loss": 0.5481,
+      "step": 17906
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.110735509647197e-06,
+      "loss": 0.6626,
+      "step": 17907
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1098512311668195e-06,
+      "loss": 0.5525,
+      "step": 17908
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.108967116110974e-06,
+      "loss": 0.5974,
+      "step": 17909
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.108083164497976e-06,
+      "loss": 0.5513,
+      "step": 17910
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.107199376346131e-06,
+      "loss": 0.5459,
+      "step": 17911
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1063157516737488e-06,
+      "loss": 0.5623,
+      "step": 17912
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1054322904991263e-06,
+      "loss": 0.6411,
+      "step": 17913
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1045489928405673e-06,
+      "loss": 0.5483,
+      "step": 17914
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.103665858716365e-06,
+      "loss": 0.5405,
+      "step": 17915
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1027828881448077e-06,
+      "loss": 0.5618,
+      "step": 17916
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1019000811441903e-06,
+      "loss": 0.5527,
+      "step": 17917
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.101017437732792e-06,
+      "loss": 0.5833,
+      "step": 17918
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1001349579289e-06,
+      "loss": 0.5693,
+      "step": 17919
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0992526417507876e-06,
+      "loss": 0.5251,
+      "step": 17920
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0983704892167347e-06,
+      "loss": 0.5093,
+      "step": 17921
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0974885003450063e-06,
+      "loss": 0.511,
+      "step": 17922
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.096606675153878e-06,
+      "loss": 0.5881,
+      "step": 17923
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.095725013661609e-06,
+      "loss": 0.5188,
+      "step": 17924
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0948435158864643e-06,
+      "loss": 0.5203,
+      "step": 17925
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0939621818466993e-06,
+      "loss": 0.5847,
+      "step": 17926
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0930810115605717e-06,
+      "loss": 0.5327,
+      "step": 17927
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0922000050463276e-06,
+      "loss": 0.5337,
+      "step": 17928
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0913191623222186e-06,
+      "loss": 0.571,
+      "step": 17929
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0904384834064906e-06,
+      "loss": 0.5508,
+      "step": 17930
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0895579683173806e-06,
+      "loss": 0.5735,
+      "step": 17931
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0886776170731303e-06,
+      "loss": 0.5845,
+      "step": 17932
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0877974296919688e-06,
+      "loss": 0.4895,
+      "step": 17933
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.086917406192134e-06,
+      "loss": 0.5374,
+      "step": 17934
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.086037546591845e-06,
+      "loss": 0.5137,
+      "step": 17935
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.085157850909334e-06,
+      "loss": 0.6316,
+      "step": 17936
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0842783191628145e-06,
+      "loss": 0.5747,
+      "step": 17937
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.083398951370511e-06,
+      "loss": 0.519,
+      "step": 17938
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.082519747550631e-06,
+      "loss": 0.5869,
+      "step": 17939
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0816407077213897e-06,
+      "loss": 0.5576,
+      "step": 17940
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.080761831900989e-06,
+      "loss": 0.5242,
+      "step": 17941
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0798831201076388e-06,
+      "loss": 0.5361,
+      "step": 17942
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0790045723595333e-06,
+      "loss": 0.5552,
+      "step": 17943
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.078126188674875e-06,
+      "loss": 0.5725,
+      "step": 17944
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.077247969071855e-06,
+      "loss": 0.5464,
+      "step": 17945
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.076369913568661e-06,
+      "loss": 0.5269,
+      "step": 17946
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0754920221834842e-06,
+      "loss": 0.5488,
+      "step": 17947
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0746142949345026e-06,
+      "loss": 0.5435,
+      "step": 17948
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0737367318399014e-06,
+      "loss": 0.5569,
+      "step": 17949
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0728593329178537e-06,
+      "loss": 0.6206,
+      "step": 17950
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0719820981865356e-06,
+      "loss": 0.4668,
+      "step": 17951
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.071105027664112e-06,
+      "loss": 0.5652,
+      "step": 17952
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0702281213687515e-06,
+      "loss": 0.5728,
+      "step": 17953
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0693513793186216e-06,
+      "loss": 0.5786,
+      "step": 17954
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0684748015318746e-06,
+      "loss": 0.5527,
+      "step": 17955
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0675983880266692e-06,
+      "loss": 0.5483,
+      "step": 17956
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.066722138821162e-06,
+      "loss": 0.55,
+      "step": 17957
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0658460539335e-06,
+      "loss": 0.5842,
+      "step": 17958
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.064970133381824e-06,
+      "loss": 0.5105,
+      "step": 17959
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0640943771842836e-06,
+      "loss": 0.5327,
+      "step": 17960
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0632187853590122e-06,
+      "loss": 0.5471,
+      "step": 17961
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0623433579241513e-06,
+      "loss": 0.5925,
+      "step": 17962
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.061468094897827e-06,
+      "loss": 0.6069,
+      "step": 17963
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.060592996298174e-06,
+      "loss": 0.5776,
+      "step": 17964
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.059718062143312e-06,
+      "loss": 0.5264,
+      "step": 17965
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.058843292451369e-06,
+      "loss": 0.5642,
+      "step": 17966
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.057968687240458e-06,
+      "loss": 0.5125,
+      "step": 17967
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.057094246528699e-06,
+      "loss": 0.4539,
+      "step": 17968
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.056219970334199e-06,
+      "loss": 0.6079,
+      "step": 17969
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.055345858675073e-06,
+      "loss": 0.5623,
+      "step": 17970
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0544719115694177e-06,
+      "loss": 0.5386,
+      "step": 17971
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0535981290353433e-06,
+      "loss": 0.5732,
+      "step": 17972
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0527245110909433e-06,
+      "loss": 0.5061,
+      "step": 17973
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0518510577543104e-06,
+      "loss": 0.5649,
+      "step": 17974
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.050977769043542e-06,
+      "loss": 0.5144,
+      "step": 17975
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.050104644976719e-06,
+      "loss": 0.5942,
+      "step": 17976
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.049231685571931e-06,
+      "loss": 0.6172,
+      "step": 17977
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0483588908472594e-06,
+      "loss": 0.5305,
+      "step": 17978
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.047486260820778e-06,
+      "loss": 0.584,
+      "step": 17979
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0466137955105637e-06,
+      "loss": 0.5811,
+      "step": 17980
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.04574149493469e-06,
+      "loss": 0.5342,
+      "step": 17981
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.044869359111219e-06,
+      "loss": 0.6382,
+      "step": 17982
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0439973880582196e-06,
+      "loss": 0.5762,
+      "step": 17983
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0431255817937477e-06,
+      "loss": 0.5698,
+      "step": 17984
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0422539403358665e-06,
+      "loss": 0.5051,
+      "step": 17985
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0413824637026236e-06,
+      "loss": 0.5708,
+      "step": 17986
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0405111519120746e-06,
+      "loss": 0.5581,
+      "step": 17987
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.039640004982264e-06,
+      "loss": 0.5898,
+      "step": 17988
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.038769022931234e-06,
+      "loss": 0.5178,
+      "step": 17989
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0378982057770292e-06,
+      "loss": 0.6184,
+      "step": 17990
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0370275535376805e-06,
+      "loss": 0.5798,
+      "step": 17991
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.036157066231227e-06,
+      "loss": 0.6216,
+      "step": 17992
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.035286743875694e-06,
+      "loss": 0.5112,
+      "step": 17993
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.034416586489113e-06,
+      "loss": 0.5857,
+      "step": 17994
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0335465940895004e-06,
+      "loss": 0.5342,
+      "step": 17995
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.032676766694883e-06,
+      "loss": 0.6074,
+      "step": 17996
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0318071043232724e-06,
+      "loss": 0.5471,
+      "step": 17997
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0309376069926847e-06,
+      "loss": 0.5991,
+      "step": 17998
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.030068274721124e-06,
+      "loss": 0.5845,
+      "step": 17999
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0291991075266036e-06,
+      "loss": 0.5459,
+      "step": 18000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.028330105427119e-06,
+      "loss": 0.6165,
+      "step": 18001
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.027461268440676e-06,
+      "loss": 0.5425,
+      "step": 18002
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.026592596585264e-06,
+      "loss": 0.521,
+      "step": 18003
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.025724089878882e-06,
+      "loss": 0.5837,
+      "step": 18004
+    },
+    {
+      "epoch": 4.0,
+      "eval_cIoU": 0.8040042519569397,
+      "eval_gIoU": 0.8009964227676392,
+      "eval_runtime": 120.3199,
+      "eval_samples_per_second": 31.674,
+      "eval_steps_per_second": 3.964,
+      "step": 18004
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0248557483395114e-06,
+      "loss": 0.4502,
+      "step": 18005
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0239875719851433e-06,
+      "loss": 0.4863,
+      "step": 18006
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0231195608337595e-06,
+      "loss": 0.5186,
+      "step": 18007
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.022251714903336e-06,
+      "loss": 0.4412,
+      "step": 18008
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0213840342118507e-06,
+      "loss": 0.4858,
+      "step": 18009
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0205165187772725e-06,
+      "loss": 0.4368,
+      "step": 18010
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0196491686175736e-06,
+      "loss": 0.4902,
+      "step": 18011
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.018781983750714e-06,
+      "loss": 0.4485,
+      "step": 18012
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0179149641946603e-06,
+      "loss": 0.4041,
+      "step": 18013
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0170481099673665e-06,
+      "loss": 0.429,
+      "step": 18014
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0161814210867913e-06,
+      "loss": 0.4524,
+      "step": 18015
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0153148975708813e-06,
+      "loss": 0.3914,
+      "step": 18016
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.01444853943759e-06,
+      "loss": 0.4133,
+      "step": 18017
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0135823467048587e-06,
+      "loss": 0.386,
+      "step": 18018
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.012716319390625e-06,
+      "loss": 0.5139,
+      "step": 18019
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0118504575128326e-06,
+      "loss": 0.5015,
+      "step": 18020
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0109847610894107e-06,
+      "loss": 0.416,
+      "step": 18021
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0101192301382956e-06,
+      "loss": 0.4026,
+      "step": 18022
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0092538646774073e-06,
+      "loss": 0.4922,
+      "step": 18023
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0083886647246766e-06,
+      "loss": 0.4419,
+      "step": 18024
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.007523630298018e-06,
+      "loss": 0.5488,
+      "step": 18025
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0066587614153545e-06,
+      "loss": 0.4832,
+      "step": 18026
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0057940580945943e-06,
+      "loss": 0.4702,
+      "step": 18027
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0049295203536523e-06,
+      "loss": 0.4678,
+      "step": 18028
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.00406514821043e-06,
+      "loss": 0.4059,
+      "step": 18029
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0032009416828323e-06,
+      "loss": 0.449,
+      "step": 18030
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.002336900788765e-06,
+      "loss": 0.4421,
+      "step": 18031
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0014730255461157e-06,
+      "loss": 0.4434,
+      "step": 18032
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0006093159727847e-06,
+      "loss": 0.488,
+      "step": 18033
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9997457720866554e-06,
+      "loss": 0.4409,
+      "step": 18034
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.998882393905619e-06,
+      "loss": 0.4807,
+      "step": 18035
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9980191814475538e-06,
+      "loss": 0.4854,
+      "step": 18036
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.997156134730345e-06,
+      "loss": 0.4492,
+      "step": 18037
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9962932537718603e-06,
+      "loss": 0.4785,
+      "step": 18038
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.995430538589981e-06,
+      "loss": 0.5261,
+      "step": 18039
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9945679892025673e-06,
+      "loss": 0.469,
+      "step": 18040
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.993705605627493e-06,
+      "loss": 0.395,
+      "step": 18041
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9928433878826127e-06,
+      "loss": 0.4502,
+      "step": 18042
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.991981335985792e-06,
+      "loss": 0.4526,
+      "step": 18043
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9911194499548793e-06,
+      "loss": 0.4795,
+      "step": 18044
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.990257729807733e-06,
+      "loss": 0.4668,
+      "step": 18045
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.989396175562196e-06,
+      "loss": 0.4053,
+      "step": 18046
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.988534787236117e-06,
+      "loss": 0.458,
+      "step": 18047
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9876735648473377e-06,
+      "loss": 0.4368,
+      "step": 18048
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9868125084136903e-06,
+      "loss": 0.4421,
+      "step": 18049
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.985951617953017e-06,
+      "loss": 0.4348,
+      "step": 18050
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9850908934831427e-06,
+      "loss": 0.4634,
+      "step": 18051
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9842303350219018e-06,
+      "loss": 0.4822,
+      "step": 18052
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.98336994258711e-06,
+      "loss": 0.4673,
+      "step": 18053
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9825097161965944e-06,
+      "loss": 0.4258,
+      "step": 18054
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9816496558681732e-06,
+      "loss": 0.4893,
+      "step": 18055
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.980789761619656e-06,
+      "loss": 0.4373,
+      "step": 18056
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9799300334688555e-06,
+      "loss": 0.4385,
+      "step": 18057
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9790704714335805e-06,
+      "loss": 0.4514,
+      "step": 18058
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.978211075531631e-06,
+      "loss": 0.4592,
+      "step": 18059
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9773518457808115e-06,
+      "loss": 0.4443,
+      "step": 18060
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9764927821989146e-06,
+      "loss": 0.4841,
+      "step": 18061
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.975633884803737e-06,
+      "loss": 0.4722,
+      "step": 18062
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9747751536130677e-06,
+      "loss": 0.4832,
+      "step": 18063
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.97391658864469e-06,
+      "loss": 0.4895,
+      "step": 18064
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9730581899163915e-06,
+      "loss": 0.4156,
+      "step": 18065
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.972199957445948e-06,
+      "loss": 0.3918,
+      "step": 18066
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9713418912511394e-06,
+      "loss": 0.4751,
+      "step": 18067
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.970483991349733e-06,
+      "loss": 0.4688,
+      "step": 18068
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.969626257759505e-06,
+      "loss": 0.4946,
+      "step": 18069
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.968768690498215e-06,
+      "loss": 0.4949,
+      "step": 18070
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9679112895836307e-06,
+      "loss": 0.428,
+      "step": 18071
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9670540550335036e-06,
+      "loss": 0.4124,
+      "step": 18072
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9661969868655984e-06,
+      "loss": 0.4927,
+      "step": 18073
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9653400850976588e-06,
+      "loss": 0.4512,
+      "step": 18074
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.96448334974744e-06,
+      "loss": 0.4653,
+      "step": 18075
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9636267808326815e-06,
+      "loss": 0.4988,
+      "step": 18076
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9627703783711304e-06,
+      "loss": 0.4878,
+      "step": 18077
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.961914142380521e-06,
+      "loss": 0.4082,
+      "step": 18078
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9610580728785876e-06,
+      "loss": 0.5239,
+      "step": 18079
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.960202169883063e-06,
+      "loss": 0.4075,
+      "step": 18080
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9593464334116773e-06,
+      "loss": 0.4072,
+      "step": 18081
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.958490863482151e-06,
+      "loss": 0.4048,
+      "step": 18082
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.957635460112206e-06,
+      "loss": 0.5305,
+      "step": 18083
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.956780223319563e-06,
+      "loss": 0.4146,
+      "step": 18084
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9559251531219315e-06,
+      "loss": 0.4446,
+      "step": 18085
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.955070249537028e-06,
+      "loss": 0.4358,
+      "step": 18086
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9542155125825514e-06,
+      "loss": 0.4465,
+      "step": 18087
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9533609422762147e-06,
+      "loss": 0.4697,
+      "step": 18088
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.952506538635709e-06,
+      "loss": 0.4446,
+      "step": 18089
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9516523016787393e-06,
+      "loss": 0.3955,
+      "step": 18090
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9507982314229935e-06,
+      "loss": 0.4136,
+      "step": 18091
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.949944327886164e-06,
+      "loss": 0.4385,
+      "step": 18092
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9490905910859382e-06,
+      "loss": 0.3701,
+      "step": 18093
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9482370210399937e-06,
+      "loss": 0.4724,
+      "step": 18094
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9473836177660176e-06,
+      "loss": 0.4155,
+      "step": 18095
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.946530381281678e-06,
+      "loss": 0.5103,
+      "step": 18096
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9456773116046556e-06,
+      "loss": 0.4783,
+      "step": 18097
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9448244087526138e-06,
+      "loss": 0.4961,
+      "step": 18098
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.943971672743221e-06,
+      "loss": 0.5073,
+      "step": 18099
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9431191035941377e-06,
+      "loss": 0.5129,
+      "step": 18100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9422667013230256e-06,
+      "loss": 0.4875,
+      "step": 18101
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9414144659475364e-06,
+      "loss": 0.4473,
+      "step": 18102
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.940562397485326e-06,
+      "loss": 0.4731,
+      "step": 18103
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.939710495954038e-06,
+      "loss": 0.3682,
+      "step": 18104
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9388587613713227e-06,
+      "loss": 0.4749,
+      "step": 18105
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.938007193754816e-06,
+      "loss": 0.4688,
+      "step": 18106
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9371557931221594e-06,
+      "loss": 0.4434,
+      "step": 18107
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.93630455949099e-06,
+      "loss": 0.4253,
+      "step": 18108
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9354534928789337e-06,
+      "loss": 0.3992,
+      "step": 18109
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.934602593303623e-06,
+      "loss": 0.3943,
+      "step": 18110
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9337518607826765e-06,
+      "loss": 0.4006,
+      "step": 18111
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.932901295333721e-06,
+      "loss": 0.4019,
+      "step": 18112
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.932050896974369e-06,
+      "loss": 0.5317,
+      "step": 18113
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9312006657222383e-06,
+      "loss": 0.4666,
+      "step": 18114
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.930350601594935e-06,
+      "loss": 0.4136,
+      "step": 18115
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.929500704610071e-06,
+      "loss": 0.415,
+      "step": 18116
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9286509747852445e-06,
+      "loss": 0.4714,
+      "step": 18117
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9278014121380605e-06,
+      "loss": 0.4829,
+      "step": 18118
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.926952016686111e-06,
+      "loss": 0.5215,
+      "step": 18119
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.926102788446994e-06,
+      "loss": 0.4092,
+      "step": 18120
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9252537274382934e-06,
+      "loss": 0.4211,
+      "step": 18121
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.924404833677601e-06,
+      "loss": 0.4597,
+      "step": 18122
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.923556107182498e-06,
+      "loss": 0.457,
+      "step": 18123
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9227075479705593e-06,
+      "loss": 0.4482,
+      "step": 18124
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.921859156059366e-06,
+      "loss": 0.4934,
+      "step": 18125
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.921010931466486e-06,
+      "loss": 0.4224,
+      "step": 18126
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9201628742094925e-06,
+      "loss": 0.3582,
+      "step": 18127
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9193149843059467e-06,
+      "loss": 0.4553,
+      "step": 18128
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9184672617734157e-06,
+      "loss": 0.4045,
+      "step": 18129
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.917619706629451e-06,
+      "loss": 0.4304,
+      "step": 18130
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.916772318891612e-06,
+      "loss": 0.48,
+      "step": 18131
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9159250985774524e-06,
+      "loss": 0.3787,
+      "step": 18132
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.915078045704515e-06,
+      "loss": 0.4517,
+      "step": 18133
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.914231160290346e-06,
+      "loss": 0.4229,
+      "step": 18134
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9133844423524905e-06,
+      "loss": 0.4565,
+      "step": 18135
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9125378919084803e-06,
+      "loss": 0.4148,
+      "step": 18136
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9116915089758548e-06,
+      "loss": 0.4373,
+      "step": 18137
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.910845293572141e-06,
+      "loss": 0.4041,
+      "step": 18138
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9099992457148654e-06,
+      "loss": 0.4707,
+      "step": 18139
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9091533654215564e-06,
+      "loss": 0.4714,
+      "step": 18140
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9083076527097287e-06,
+      "loss": 0.4578,
+      "step": 18141
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9074621075969035e-06,
+      "loss": 0.5042,
+      "step": 18142
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.906616730100591e-06,
+      "loss": 0.4302,
+      "step": 18143
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9057715202383043e-06,
+      "loss": 0.4814,
+      "step": 18144
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9049264780275456e-06,
+      "loss": 0.5288,
+      "step": 18145
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9040816034858233e-06,
+      "loss": 0.491,
+      "step": 18146
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9032368966306303e-06,
+      "loss": 0.4255,
+      "step": 18147
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9023923574794688e-06,
+      "loss": 0.4548,
+      "step": 18148
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9015479860498264e-06,
+      "loss": 0.4143,
+      "step": 18149
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9007037823591966e-06,
+      "loss": 0.4829,
+      "step": 18150
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.89985974642506e-06,
+      "loss": 0.4651,
+      "step": 18151
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8990158782649038e-06,
+      "loss": 0.4736,
+      "step": 18152
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8981721778962037e-06,
+      "loss": 0.4131,
+      "step": 18153
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8973286453364337e-06,
+      "loss": 0.4651,
+      "step": 18154
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8964852806030688e-06,
+      "loss": 0.4285,
+      "step": 18155
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8956420837135725e-06,
+      "loss": 0.4233,
+      "step": 18156
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8947990546854123e-06,
+      "loss": 0.4397,
+      "step": 18157
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8939561935360529e-06,
+      "loss": 0.4626,
+      "step": 18158
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8931135002829448e-06,
+      "loss": 0.4678,
+      "step": 18159
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8922709749435465e-06,
+      "loss": 0.4485,
+      "step": 18160
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8914286175353114e-06,
+      "loss": 0.4309,
+      "step": 18161
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8905864280756814e-06,
+      "loss": 0.481,
+      "step": 18162
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8897444065821046e-06,
+      "loss": 0.4458,
+      "step": 18163
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.888902553072016e-06,
+      "loss": 0.4629,
+      "step": 18164
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8880608675628597e-06,
+      "loss": 0.4949,
+      "step": 18165
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8872193500720647e-06,
+      "loss": 0.4093,
+      "step": 18166
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8863780006170596e-06,
+      "loss": 0.3865,
+      "step": 18167
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8855368192152756e-06,
+      "loss": 0.4106,
+      "step": 18168
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8846958058841292e-06,
+      "loss": 0.4524,
+      "step": 18169
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8838549606410461e-06,
+      "loss": 0.4744,
+      "step": 18170
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.883014283503437e-06,
+      "loss": 0.4609,
+      "step": 18171
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8821737744887203e-06,
+      "loss": 0.4363,
+      "step": 18172
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8813334336142985e-06,
+      "loss": 0.4792,
+      "step": 18173
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8804932608975834e-06,
+      "loss": 0.4336,
+      "step": 18174
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8796532563559722e-06,
+      "loss": 0.4404,
+      "step": 18175
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8788134200068664e-06,
+      "loss": 0.4302,
+      "step": 18176
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8779737518676589e-06,
+      "loss": 0.4395,
+      "step": 18177
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8771342519557446e-06,
+      "loss": 0.4143,
+      "step": 18178
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8762949202885061e-06,
+      "loss": 0.4326,
+      "step": 18179
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8754557568833354e-06,
+      "loss": 0.3815,
+      "step": 18180
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8746167617576061e-06,
+      "loss": 0.3916,
+      "step": 18181
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8737779349287021e-06,
+      "loss": 0.4871,
+      "step": 18182
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.872939276413992e-06,
+      "loss": 0.5137,
+      "step": 18183
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8721007862308505e-06,
+      "loss": 0.4578,
+      "step": 18184
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8712624643966448e-06,
+      "loss": 0.4553,
+      "step": 18185
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8704243109287356e-06,
+      "loss": 0.4375,
+      "step": 18186
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.869586325844488e-06,
+      "loss": 0.4373,
+      "step": 18187
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8687485091612534e-06,
+      "loss": 0.479,
+      "step": 18188
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8679108608963903e-06,
+      "loss": 0.4392,
+      "step": 18189
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8670733810672425e-06,
+      "loss": 0.4556,
+      "step": 18190
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8662360696911618e-06,
+      "loss": 0.5151,
+      "step": 18191
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8653989267854866e-06,
+      "loss": 0.4827,
+      "step": 18192
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8645619523675606e-06,
+      "loss": 0.3994,
+      "step": 18193
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8637251464547146e-06,
+      "loss": 0.4182,
+      "step": 18194
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8628885090642857e-06,
+      "loss": 0.4854,
+      "step": 18195
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8620520402136012e-06,
+      "loss": 0.4719,
+      "step": 18196
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8612157399199838e-06,
+      "loss": 0.418,
+      "step": 18197
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8603796082007597e-06,
+      "loss": 0.4541,
+      "step": 18198
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8595436450732419e-06,
+      "loss": 0.4841,
+      "step": 18199
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8587078505547507e-06,
+      "loss": 0.4614,
+      "step": 18200
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8578722246625935e-06,
+      "loss": 0.418,
+      "step": 18201
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.857036767414081e-06,
+      "loss": 0.4543,
+      "step": 18202
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8562014788265148e-06,
+      "loss": 0.4373,
+      "step": 18203
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8553663589171999e-06,
+      "loss": 0.4565,
+      "step": 18204
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8545314077034281e-06,
+      "loss": 0.4861,
+      "step": 18205
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8536966252025002e-06,
+      "loss": 0.46,
+      "step": 18206
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8528620114316986e-06,
+      "loss": 0.4299,
+      "step": 18207
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8520275664083155e-06,
+      "loss": 0.416,
+      "step": 18208
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8511932901496355e-06,
+      "loss": 0.5,
+      "step": 18209
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8503591826729339e-06,
+      "loss": 0.4156,
+      "step": 18210
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8495252439954924e-06,
+      "loss": 0.4705,
+      "step": 18211
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8486914741345774e-06,
+      "loss": 0.4556,
+      "step": 18212
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.847857873107466e-06,
+      "loss": 0.4678,
+      "step": 18213
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8470244409314165e-06,
+      "loss": 0.4966,
+      "step": 18214
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8461911776236973e-06,
+      "loss": 0.4282,
+      "step": 18215
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8453580832015627e-06,
+      "loss": 0.425,
+      "step": 18216
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8445251576822742e-06,
+      "loss": 0.3953,
+      "step": 18217
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8436924010830758e-06,
+      "loss": 0.4183,
+      "step": 18218
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.842859813421224e-06,
+      "loss": 0.4692,
+      "step": 18219
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8420273947139566e-06,
+      "loss": 0.4341,
+      "step": 18220
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8411951449785215e-06,
+      "loss": 0.4802,
+      "step": 18221
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8403630642321502e-06,
+      "loss": 0.4958,
+      "step": 18222
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8395311524920845e-06,
+      "loss": 0.4167,
+      "step": 18223
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8386994097755473e-06,
+      "loss": 0.4294,
+      "step": 18224
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8378678360997737e-06,
+      "loss": 0.4548,
+      "step": 18225
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8370364314819832e-06,
+      "loss": 0.4492,
+      "step": 18226
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8362051959393945e-06,
+      "loss": 0.4407,
+      "step": 18227
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8353741294892302e-06,
+      "loss": 0.4738,
+      "step": 18228
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8345432321486988e-06,
+      "loss": 0.4197,
+      "step": 18229
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8337125039350135e-06,
+      "loss": 0.4905,
+      "step": 18230
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8328819448653778e-06,
+      "loss": 0.4719,
+      "step": 18231
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8320515549569984e-06,
+      "loss": 0.4587,
+      "step": 18232
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.831221334227069e-06,
+      "loss": 0.4163,
+      "step": 18233
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8303912826927906e-06,
+      "loss": 0.4116,
+      "step": 18234
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8295614003713557e-06,
+      "loss": 0.4714,
+      "step": 18235
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.82873168727995e-06,
+      "loss": 0.4568,
+      "step": 18236
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8279021434357592e-06,
+      "loss": 0.4084,
+      "step": 18237
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8270727688559697e-06,
+      "loss": 0.4697,
+      "step": 18238
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8262435635577546e-06,
+      "loss": 0.4175,
+      "step": 18239
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8254145275582936e-06,
+      "loss": 0.4678,
+      "step": 18240
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8245856608747548e-06,
+      "loss": 0.4673,
+      "step": 18241
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8237569635243046e-06,
+      "loss": 0.4561,
+      "step": 18242
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8229284355241117e-06,
+      "loss": 0.4707,
+      "step": 18243
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.822100076891332e-06,
+      "loss": 0.48,
+      "step": 18244
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.821271887643129e-06,
+      "loss": 0.4207,
+      "step": 18245
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8204438677966508e-06,
+      "loss": 0.429,
+      "step": 18246
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8196160173690513e-06,
+      "loss": 0.4441,
+      "step": 18247
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.818788336377475e-06,
+      "loss": 0.4141,
+      "step": 18248
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8179608248390679e-06,
+      "loss": 0.4219,
+      "step": 18249
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8171334827709663e-06,
+      "loss": 0.4624,
+      "step": 18250
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.816306310190311e-06,
+      "loss": 0.4358,
+      "step": 18251
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8154793071142296e-06,
+      "loss": 0.4539,
+      "step": 18252
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.814652473559857e-06,
+      "loss": 0.4788,
+      "step": 18253
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.813825809544313e-06,
+      "loss": 0.4592,
+      "step": 18254
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8129993150847257e-06,
+      "loss": 0.4641,
+      "step": 18255
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.812172990198211e-06,
+      "loss": 0.4856,
+      "step": 18256
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8113468349018815e-06,
+      "loss": 0.4194,
+      "step": 18257
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8105208492128511e-06,
+      "loss": 0.4373,
+      "step": 18258
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8096950331482321e-06,
+      "loss": 0.4619,
+      "step": 18259
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8088693867251227e-06,
+      "loss": 0.4541,
+      "step": 18260
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8080439099606273e-06,
+      "loss": 0.4287,
+      "step": 18261
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8072186028718453e-06,
+      "loss": 0.4941,
+      "step": 18262
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8063934654758664e-06,
+      "loss": 0.4607,
+      "step": 18263
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.805568497789786e-06,
+      "loss": 0.4417,
+      "step": 18264
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8047436998306866e-06,
+      "loss": 0.4717,
+      "step": 18265
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8039190716156562e-06,
+      "loss": 0.4634,
+      "step": 18266
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.803094613161771e-06,
+      "loss": 0.488,
+      "step": 18267
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8022703244861118e-06,
+      "loss": 0.4683,
+      "step": 18268
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.801446205605747e-06,
+      "loss": 0.4736,
+      "step": 18269
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8006222565377507e-06,
+      "loss": 0.5022,
+      "step": 18270
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.799798477299186e-06,
+      "loss": 0.4658,
+      "step": 18271
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7989748679071139e-06,
+      "loss": 0.4702,
+      "step": 18272
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7981514283785983e-06,
+      "loss": 0.4404,
+      "step": 18273
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.797328158730689e-06,
+      "loss": 0.4248,
+      "step": 18274
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7965050589804434e-06,
+      "loss": 0.4056,
+      "step": 18275
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7956821291449046e-06,
+      "loss": 0.4319,
+      "step": 18276
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7948593692411232e-06,
+      "loss": 0.426,
+      "step": 18277
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7940367792861347e-06,
+      "loss": 0.4089,
+      "step": 18278
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.793214359296983e-06,
+      "loss": 0.4075,
+      "step": 18279
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.792392109290697e-06,
+      "loss": 0.3899,
+      "step": 18280
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7915700292843118e-06,
+      "loss": 0.4827,
+      "step": 18281
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7907481192948505e-06,
+      "loss": 0.437,
+      "step": 18282
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7899263793393429e-06,
+      "loss": 0.5339,
+      "step": 18283
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7891048094348018e-06,
+      "loss": 0.4666,
+      "step": 18284
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.788283409598248e-06,
+      "loss": 0.4429,
+      "step": 18285
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7874621798466973e-06,
+      "loss": 0.4878,
+      "step": 18286
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7866411201971546e-06,
+      "loss": 0.4607,
+      "step": 18287
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7858202306666296e-06,
+      "loss": 0.4209,
+      "step": 18288
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7849995112721209e-06,
+      "loss": 0.4495,
+      "step": 18289
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7841789620306337e-06,
+      "loss": 0.4785,
+      "step": 18290
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.783358582959157e-06,
+      "loss": 0.4119,
+      "step": 18291
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7825383740746881e-06,
+      "loss": 0.4541,
+      "step": 18292
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7817183353942114e-06,
+      "loss": 0.4685,
+      "step": 18293
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7808984669347162e-06,
+      "loss": 0.4644,
+      "step": 18294
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7800787687131803e-06,
+      "loss": 0.511,
+      "step": 18295
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7792592407465848e-06,
+      "loss": 0.5498,
+      "step": 18296
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7784398830519002e-06,
+      "loss": 0.4482,
+      "step": 18297
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7776206956461016e-06,
+      "loss": 0.4829,
+      "step": 18298
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.776801678546153e-06,
+      "loss": 0.5054,
+      "step": 18299
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7759828317690209e-06,
+      "loss": 0.4729,
+      "step": 18300
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7751641553316656e-06,
+      "loss": 0.4873,
+      "step": 18301
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7743456492510402e-06,
+      "loss": 0.4143,
+      "step": 18302
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7735273135441034e-06,
+      "loss": 0.4048,
+      "step": 18303
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.772709148227799e-06,
+      "loss": 0.4253,
+      "step": 18304
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7718911533190798e-06,
+      "loss": 0.4749,
+      "step": 18305
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7710733288348826e-06,
+      "loss": 0.4858,
+      "step": 18306
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7702556747921517e-06,
+      "loss": 0.4543,
+      "step": 18307
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7694381912078173e-06,
+      "loss": 0.4272,
+      "step": 18308
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7686208780988168e-06,
+      "loss": 0.4387,
+      "step": 18309
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7678037354820733e-06,
+      "loss": 0.446,
+      "step": 18310
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7669867633745164e-06,
+      "loss": 0.3826,
+      "step": 18311
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7661699617930671e-06,
+      "loss": 0.4541,
+      "step": 18312
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7653533307546411e-06,
+      "loss": 0.469,
+      "step": 18313
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7645368702761534e-06,
+      "loss": 0.5044,
+      "step": 18314
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7637205803745182e-06,
+      "loss": 0.4468,
+      "step": 18315
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7629044610666412e-06,
+      "loss": 0.4756,
+      "step": 18316
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.762088512369422e-06,
+      "loss": 0.4568,
+      "step": 18317
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.761272734299767e-06,
+      "loss": 0.4238,
+      "step": 18318
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7604571268745685e-06,
+      "loss": 0.3538,
+      "step": 18319
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7596416901107238e-06,
+      "loss": 0.4939,
+      "step": 18320
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7588264240251174e-06,
+      "loss": 0.4303,
+      "step": 18321
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7580113286346413e-06,
+      "loss": 0.3865,
+      "step": 18322
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7571964039561728e-06,
+      "loss": 0.4771,
+      "step": 18323
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7563816500065956e-06,
+      "loss": 0.4194,
+      "step": 18324
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7555670668027813e-06,
+      "loss": 0.4133,
+      "step": 18325
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7547526543616067e-06,
+      "loss": 0.4458,
+      "step": 18326
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7539384126999337e-06,
+      "loss": 0.4092,
+      "step": 18327
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7531243418346344e-06,
+      "loss": 0.3987,
+      "step": 18328
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7523104417825642e-06,
+      "loss": 0.4702,
+      "step": 18329
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7514967125605863e-06,
+      "loss": 0.4666,
+      "step": 18330
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7506831541855518e-06,
+      "loss": 0.4375,
+      "step": 18331
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7498697666743102e-06,
+      "loss": 0.4653,
+      "step": 18332
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7490565500437129e-06,
+      "loss": 0.449,
+      "step": 18333
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7482435043105984e-06,
+      "loss": 0.5181,
+      "step": 18334
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7474306294918098e-06,
+      "loss": 0.4041,
+      "step": 18335
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7466179256041872e-06,
+      "loss": 0.4866,
+      "step": 18336
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7458053926645579e-06,
+      "loss": 0.4077,
+      "step": 18337
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7449930306897533e-06,
+      "loss": 0.4539,
+      "step": 18338
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.744180839696602e-06,
+      "loss": 0.4468,
+      "step": 18339
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7433688197019226e-06,
+      "loss": 0.377,
+      "step": 18340
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7425569707225386e-06,
+      "loss": 0.4578,
+      "step": 18341
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7417452927752608e-06,
+      "loss": 0.4255,
+      "step": 18342
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7409337858769048e-06,
+      "loss": 0.4939,
+      "step": 18343
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7401224500442748e-06,
+      "loss": 0.4271,
+      "step": 18344
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7393112852941807e-06,
+      "loss": 0.4254,
+      "step": 18345
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7385002916434203e-06,
+      "loss": 0.4707,
+      "step": 18346
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7376894691087899e-06,
+      "loss": 0.4082,
+      "step": 18347
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7368788177070883e-06,
+      "loss": 0.4272,
+      "step": 18348
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7360683374551002e-06,
+      "loss": 0.4812,
+      "step": 18349
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7352580283696197e-06,
+      "loss": 0.4668,
+      "step": 18350
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7344478904674234e-06,
+      "loss": 0.4866,
+      "step": 18351
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7336379237652966e-06,
+      "loss": 0.4126,
+      "step": 18352
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7328281282800119e-06,
+      "loss": 0.4692,
+      "step": 18353
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7320185040283454e-06,
+      "loss": 0.4119,
+      "step": 18354
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7312090510270629e-06,
+      "loss": 0.417,
+      "step": 18355
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7303997692929342e-06,
+      "loss": 0.4702,
+      "step": 18356
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.729590658842717e-06,
+      "loss": 0.4644,
+      "step": 18357
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7287817196931755e-06,
+      "loss": 0.4648,
+      "step": 18358
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7279729518610588e-06,
+      "loss": 0.5032,
+      "step": 18359
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7271643553631234e-06,
+      "loss": 0.4536,
+      "step": 18360
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7263559302161138e-06,
+      "loss": 0.4636,
+      "step": 18361
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7255476764367763e-06,
+      "loss": 0.4714,
+      "step": 18362
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7247395940418532e-06,
+      "loss": 0.4924,
+      "step": 18363
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.723931683048078e-06,
+      "loss": 0.509,
+      "step": 18364
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.723123943472189e-06,
+      "loss": 0.4871,
+      "step": 18365
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7223163753309125e-06,
+      "loss": 0.4675,
+      "step": 18366
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.72150897864098e-06,
+      "loss": 0.4521,
+      "step": 18367
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7207017534191083e-06,
+      "loss": 0.4573,
+      "step": 18368
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.719894699682023e-06,
+      "loss": 0.4232,
+      "step": 18369
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7190878174464354e-06,
+      "loss": 0.4487,
+      "step": 18370
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7182811067290629e-06,
+      "loss": 0.5037,
+      "step": 18371
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7174745675466099e-06,
+      "loss": 0.3911,
+      "step": 18372
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7166681999157853e-06,
+      "loss": 0.4104,
+      "step": 18373
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7158620038532903e-06,
+      "loss": 0.5437,
+      "step": 18374
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7150559793758192e-06,
+      "loss": 0.4485,
+      "step": 18375
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7142501265000732e-06,
+      "loss": 0.436,
+      "step": 18376
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7134444452427367e-06,
+      "loss": 0.4224,
+      "step": 18377
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7126389356205042e-06,
+      "loss": 0.5271,
+      "step": 18378
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7118335976500534e-06,
+      "loss": 0.4351,
+      "step": 18379
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7110284313480718e-06,
+      "loss": 0.4221,
+      "step": 18380
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7102234367312288e-06,
+      "loss": 0.4081,
+      "step": 18381
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7094186138162039e-06,
+      "loss": 0.438,
+      "step": 18382
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.708613962619663e-06,
+      "loss": 0.458,
+      "step": 18383
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.707809483158276e-06,
+      "loss": 0.4468,
+      "step": 18384
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7070051754487017e-06,
+      "loss": 0.5183,
+      "step": 18385
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7062010395076035e-06,
+      "loss": 0.4038,
+      "step": 18386
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7053970753516325e-06,
+      "loss": 0.479,
+      "step": 18387
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7045932829974443e-06,
+      "loss": 0.4963,
+      "step": 18388
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7037896624616868e-06,
+      "loss": 0.4294,
+      "step": 18389
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.702986213761003e-06,
+      "loss": 0.4595,
+      "step": 18390
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.702182936912039e-06,
+      "loss": 0.449,
+      "step": 18391
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7013798319314257e-06,
+      "loss": 0.4663,
+      "step": 18392
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.700576898835804e-06,
+      "loss": 0.4346,
+      "step": 18393
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6997741376417997e-06,
+      "loss": 0.5083,
+      "step": 18394
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6989715483660452e-06,
+      "loss": 0.3938,
+      "step": 18395
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.698169131025157e-06,
+      "loss": 0.3544,
+      "step": 18396
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6973668856357627e-06,
+      "loss": 0.4493,
+      "step": 18397
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6965648122144717e-06,
+      "loss": 0.4885,
+      "step": 18398
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.695762910777904e-06,
+      "loss": 0.468,
+      "step": 18399
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6949611813426625e-06,
+      "loss": 0.4128,
+      "step": 18400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.694159623925359e-06,
+      "loss": 0.4719,
+      "step": 18401
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6933582385425884e-06,
+      "loss": 0.4875,
+      "step": 18402
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.692557025210958e-06,
+      "loss": 0.3828,
+      "step": 18403
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6917559839470577e-06,
+      "loss": 0.4692,
+      "step": 18404
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6909551147674775e-06,
+      "loss": 0.4666,
+      "step": 18405
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6901544176888106e-06,
+      "loss": 0.418,
+      "step": 18406
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6893538927276353e-06,
+      "loss": 0.4475,
+      "step": 18407
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6885535399005393e-06,
+      "loss": 0.4016,
+      "step": 18408
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6877533592240936e-06,
+      "loss": 0.5149,
+      "step": 18409
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6869533507148762e-06,
+      "loss": 0.4141,
+      "step": 18410
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6861535143894537e-06,
+      "loss": 0.4521,
+      "step": 18411
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.685353850264395e-06,
+      "loss": 0.4604,
+      "step": 18412
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6845543583562663e-06,
+      "loss": 0.449,
+      "step": 18413
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6837550386816193e-06,
+      "loss": 0.4424,
+      "step": 18414
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6829558912570154e-06,
+      "loss": 0.4817,
+      "step": 18415
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6821569160990082e-06,
+      "loss": 0.4525,
+      "step": 18416
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6813581132241407e-06,
+      "loss": 0.4348,
+      "step": 18417
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6805594826489647e-06,
+      "loss": 0.4443,
+      "step": 18418
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6797610243900174e-06,
+      "loss": 0.3596,
+      "step": 18419
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6789627384638364e-06,
+      "loss": 0.4998,
+      "step": 18420
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6781646248869598e-06,
+      "loss": 0.4492,
+      "step": 18421
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6773666836759138e-06,
+      "loss": 0.4834,
+      "step": 18422
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6765689148472319e-06,
+      "loss": 0.4731,
+      "step": 18423
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6757713184174306e-06,
+      "loss": 0.4873,
+      "step": 18424
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.674973894403037e-06,
+      "loss": 0.4102,
+      "step": 18425
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6741766428205619e-06,
+      "loss": 0.4509,
+      "step": 18426
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6733795636865236e-06,
+      "loss": 0.4702,
+      "step": 18427
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6725826570174264e-06,
+      "loss": 0.4541,
+      "step": 18428
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6717859228297816e-06,
+      "loss": 0.4495,
+      "step": 18429
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.670989361140085e-06,
+      "loss": 0.4214,
+      "step": 18430
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6701929719648435e-06,
+      "loss": 0.4382,
+      "step": 18431
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6693967553205448e-06,
+      "loss": 0.4465,
+      "step": 18432
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6686007112236857e-06,
+      "loss": 0.5093,
+      "step": 18433
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.667804839690752e-06,
+      "loss": 0.4116,
+      "step": 18434
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.667009140738226e-06,
+      "loss": 0.4568,
+      "step": 18435
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6662136143825913e-06,
+      "loss": 0.4548,
+      "step": 18436
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6654182606403268e-06,
+      "loss": 0.4592,
+      "step": 18437
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6646230795279028e-06,
+      "loss": 0.4202,
+      "step": 18438
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6638280710617903e-06,
+      "loss": 0.436,
+      "step": 18439
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6630332352584587e-06,
+      "loss": 0.4697,
+      "step": 18440
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.662238572134367e-06,
+      "loss": 0.3894,
+      "step": 18441
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6614440817059785e-06,
+      "loss": 0.4161,
+      "step": 18442
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6606497639897457e-06,
+      "loss": 0.4333,
+      "step": 18443
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6598556190021242e-06,
+      "loss": 0.4227,
+      "step": 18444
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6590616467595577e-06,
+      "loss": 0.5103,
+      "step": 18445
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6582678472784975e-06,
+      "loss": 0.4277,
+      "step": 18446
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6574742205753803e-06,
+      "loss": 0.4719,
+      "step": 18447
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6566807666666474e-06,
+      "loss": 0.4946,
+      "step": 18448
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6558874855687313e-06,
+      "loss": 0.4302,
+      "step": 18449
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.655094377298062e-06,
+      "loss": 0.4756,
+      "step": 18450
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.654301441871069e-06,
+      "loss": 0.4104,
+      "step": 18451
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6535086793041733e-06,
+      "loss": 0.4458,
+      "step": 18452
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6527160896137996e-06,
+      "loss": 0.4492,
+      "step": 18453
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.651923672816358e-06,
+      "loss": 0.4167,
+      "step": 18454
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6511314289282664e-06,
+      "loss": 0.4343,
+      "step": 18455
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.650339357965932e-06,
+      "loss": 0.47,
+      "step": 18456
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6495474599457618e-06,
+      "loss": 0.4368,
+      "step": 18457
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6487557348841554e-06,
+      "loss": 0.418,
+      "step": 18458
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6479641827975167e-06,
+      "loss": 0.4531,
+      "step": 18459
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.647172803702234e-06,
+      "loss": 0.437,
+      "step": 18460
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6463815976147035e-06,
+      "loss": 0.5049,
+      "step": 18461
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6455905645513105e-06,
+      "loss": 0.4019,
+      "step": 18462
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6447997045284413e-06,
+      "loss": 0.4546,
+      "step": 18463
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6440090175624733e-06,
+      "loss": 0.4468,
+      "step": 18464
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6432185036697867e-06,
+      "loss": 0.4646,
+      "step": 18465
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.642428162866756e-06,
+      "loss": 0.4956,
+      "step": 18466
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.641637995169747e-06,
+      "loss": 0.4343,
+      "step": 18467
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6408480005951311e-06,
+      "loss": 0.4565,
+      "step": 18468
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.640058179159265e-06,
+      "loss": 0.4768,
+      "step": 18469
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6392685308785139e-06,
+      "loss": 0.4712,
+      "step": 18470
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6384790557692277e-06,
+      "loss": 0.4316,
+      "step": 18471
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.637689753847763e-06,
+      "loss": 0.3967,
+      "step": 18472
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6369006251304653e-06,
+      "loss": 0.4924,
+      "step": 18473
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.636111669633682e-06,
+      "loss": 0.4636,
+      "step": 18474
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6353228873737515e-06,
+      "loss": 0.4421,
+      "step": 18475
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6345342783670147e-06,
+      "loss": 0.4375,
+      "step": 18476
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6337458426298003e-06,
+      "loss": 0.4771,
+      "step": 18477
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6329575801784458e-06,
+      "loss": 0.4775,
+      "step": 18478
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.632169491029274e-06,
+      "loss": 0.4631,
+      "step": 18479
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6313815751986063e-06,
+      "loss": 0.5137,
+      "step": 18480
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6305938327027671e-06,
+      "loss": 0.415,
+      "step": 18481
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.629806263558068e-06,
+      "loss": 0.4463,
+      "step": 18482
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.629018867780826e-06,
+      "loss": 0.5037,
+      "step": 18483
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6282316453873448e-06,
+      "loss": 0.5093,
+      "step": 18484
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6274445963939356e-06,
+      "loss": 0.4246,
+      "step": 18485
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6266577208168944e-06,
+      "loss": 0.4384,
+      "step": 18486
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6258710186725235e-06,
+      "loss": 0.4434,
+      "step": 18487
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6250844899771146e-06,
+      "loss": 0.4805,
+      "step": 18488
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6242981347469599e-06,
+      "loss": 0.432,
+      "step": 18489
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.62351195299835e-06,
+      "loss": 0.4387,
+      "step": 18490
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6227259447475619e-06,
+      "loss": 0.4404,
+      "step": 18491
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.62194011001088e-06,
+      "loss": 0.5173,
+      "step": 18492
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6211544488045839e-06,
+      "loss": 0.425,
+      "step": 18493
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6203689611449415e-06,
+      "loss": 0.4536,
+      "step": 18494
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6195836470482218e-06,
+      "loss": 0.4539,
+      "step": 18495
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6187985065306955e-06,
+      "loss": 0.502,
+      "step": 18496
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6180135396086183e-06,
+      "loss": 0.4338,
+      "step": 18497
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.617228746298256e-06,
+      "loss": 0.4451,
+      "step": 18498
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.616444126615857e-06,
+      "loss": 0.5393,
+      "step": 18499
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.615659680577678e-06,
+      "loss": 0.401,
+      "step": 18500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6148754081999618e-06,
+      "loss": 0.4497,
+      "step": 18501
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6140913094989586e-06,
+      "loss": 0.5317,
+      "step": 18502
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.613307384490903e-06,
+      "loss": 0.4451,
+      "step": 18503
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6125236331920369e-06,
+      "loss": 0.4548,
+      "step": 18504
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6117400556185891e-06,
+      "loss": 0.4087,
+      "step": 18505
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6109566517867959e-06,
+      "loss": 0.4336,
+      "step": 18506
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.610173421712875e-06,
+      "loss": 0.457,
+      "step": 18507
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6093903654130572e-06,
+      "loss": 0.4126,
+      "step": 18508
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.608607482903558e-06,
+      "loss": 0.427,
+      "step": 18509
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6078247742005903e-06,
+      "loss": 0.4414,
+      "step": 18510
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.607042239320371e-06,
+      "loss": 0.4023,
+      "step": 18511
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6062598782791028e-06,
+      "loss": 0.4554,
+      "step": 18512
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6054776910929936e-06,
+      "loss": 0.4766,
+      "step": 18513
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6046956777782475e-06,
+      "loss": 0.3892,
+      "step": 18514
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6039138383510556e-06,
+      "loss": 0.3799,
+      "step": 18515
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6031321728276138e-06,
+      "loss": 0.4746,
+      "step": 18516
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6023506812241163e-06,
+      "loss": 0.4502,
+      "step": 18517
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6015693635567441e-06,
+      "loss": 0.4333,
+      "step": 18518
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6007882198416858e-06,
+      "loss": 0.405,
+      "step": 18519
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6000072500951148e-06,
+      "loss": 0.4922,
+      "step": 18520
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5992264543332125e-06,
+      "loss": 0.4705,
+      "step": 18521
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5984458325721464e-06,
+      "loss": 0.4253,
+      "step": 18522
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5976653848280899e-06,
+      "loss": 0.4668,
+      "step": 18523
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5968851111172046e-06,
+      "loss": 0.4233,
+      "step": 18524
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.596105011455651e-06,
+      "loss": 0.407,
+      "step": 18525
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5953250858595903e-06,
+      "loss": 0.512,
+      "step": 18526
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5945453343451721e-06,
+      "loss": 0.4495,
+      "step": 18527
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5937657569285537e-06,
+      "loss": 0.5178,
+      "step": 18528
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.592986353625875e-06,
+      "loss": 0.5164,
+      "step": 18529
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5922071244532854e-06,
+      "loss": 0.5166,
+      "step": 18530
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.591428069426919e-06,
+      "loss": 0.4551,
+      "step": 18531
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5906491885629171e-06,
+      "loss": 0.4817,
+      "step": 18532
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.589870481877409e-06,
+      "loss": 0.3872,
+      "step": 18533
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5890919493865254e-06,
+      "loss": 0.4817,
+      "step": 18534
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5883135911063896e-06,
+      "loss": 0.4741,
+      "step": 18535
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5875354070531257e-06,
+      "loss": 0.4802,
+      "step": 18536
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5867573972428496e-06,
+      "loss": 0.4536,
+      "step": 18537
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5859795616916784e-06,
+      "loss": 0.4543,
+      "step": 18538
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5852019004157204e-06,
+      "loss": 0.4775,
+      "step": 18539
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5844244134310849e-06,
+      "loss": 0.3877,
+      "step": 18540
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5836471007538734e-06,
+      "loss": 0.4033,
+      "step": 18541
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5828699624001864e-06,
+      "loss": 0.4639,
+      "step": 18542
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5820929983861245e-06,
+      "loss": 0.426,
+      "step": 18543
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5813162087277745e-06,
+      "loss": 0.5093,
+      "step": 18544
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5805395934412304e-06,
+      "loss": 0.447,
+      "step": 18545
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5797631525425738e-06,
+      "loss": 0.489,
+      "step": 18546
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5789868860478918e-06,
+      "loss": 0.4387,
+      "step": 18547
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.578210793973256e-06,
+      "loss": 0.4443,
+      "step": 18548
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5774348763347491e-06,
+      "loss": 0.3918,
+      "step": 18549
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5766591331484349e-06,
+      "loss": 0.4133,
+      "step": 18550
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.575883564430386e-06,
+      "loss": 0.5554,
+      "step": 18551
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.575108170196663e-06,
+      "loss": 0.4675,
+      "step": 18552
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5743329504633298e-06,
+      "loss": 0.4216,
+      "step": 18553
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5735579052464412e-06,
+      "loss": 0.4783,
+      "step": 18554
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5727830345620475e-06,
+      "loss": 0.4375,
+      "step": 18555
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5720083384262042e-06,
+      "loss": 0.5012,
+      "step": 18556
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.57123381685495e-06,
+      "loss": 0.489,
+      "step": 18557
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.570459469864335e-06,
+      "loss": 0.3967,
+      "step": 18558
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5696852974703901e-06,
+      "loss": 0.4604,
+      "step": 18559
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5689112996891576e-06,
+      "loss": 0.4138,
+      "step": 18560
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.568137476536663e-06,
+      "loss": 0.4426,
+      "step": 18561
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5673638280289383e-06,
+      "loss": 0.4431,
+      "step": 18562
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5665903541820038e-06,
+      "loss": 0.4573,
+      "step": 18563
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5658170550118845e-06,
+      "loss": 0.4548,
+      "step": 18564
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5650439305345932e-06,
+      "loss": 0.4109,
+      "step": 18565
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5642709807661449e-06,
+      "loss": 0.4556,
+      "step": 18566
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5634982057225512e-06,
+      "loss": 0.4336,
+      "step": 18567
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5627256054198136e-06,
+      "loss": 0.4922,
+      "step": 18568
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5619531798739408e-06,
+      "loss": 0.4399,
+      "step": 18569
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.561180929100926e-06,
+      "loss": 0.4355,
+      "step": 18570
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5604088531167683e-06,
+      "loss": 0.4636,
+      "step": 18571
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5596369519374554e-06,
+      "loss": 0.4326,
+      "step": 18572
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5588652255789804e-06,
+      "loss": 0.4175,
+      "step": 18573
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5580936740573226e-06,
+      "loss": 0.3862,
+      "step": 18574
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.557322297388466e-06,
+      "loss": 0.4543,
+      "step": 18575
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5565510955883855e-06,
+      "loss": 0.4692,
+      "step": 18576
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5557800686730584e-06,
+      "loss": 0.4226,
+      "step": 18577
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5550092166584486e-06,
+      "loss": 0.5154,
+      "step": 18578
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5542385395605285e-06,
+      "loss": 0.4521,
+      "step": 18579
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5534680373952559e-06,
+      "loss": 0.4568,
+      "step": 18580
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5526977101785945e-06,
+      "loss": 0.458,
+      "step": 18581
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5519275579264936e-06,
+      "loss": 0.4453,
+      "step": 18582
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5511575806549118e-06,
+      "loss": 0.4232,
+      "step": 18583
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5503877783797938e-06,
+      "loss": 0.4299,
+      "step": 18584
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5496181511170816e-06,
+      "loss": 0.4517,
+      "step": 18585
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.54884869888272e-06,
+      "loss": 0.5276,
+      "step": 18586
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5480794216926443e-06,
+      "loss": 0.4644,
+      "step": 18587
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5473103195627903e-06,
+      "loss": 0.4285,
+      "step": 18588
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5465413925090833e-06,
+      "loss": 0.3696,
+      "step": 18589
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5457726405474539e-06,
+      "loss": 0.4214,
+      "step": 18590
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5450040636938257e-06,
+      "loss": 0.4575,
+      "step": 18591
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5442356619641141e-06,
+      "loss": 0.4771,
+      "step": 18592
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5434674353742364e-06,
+      "loss": 0.4387,
+      "step": 18593
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.542699383940106e-06,
+      "loss": 0.468,
+      "step": 18594
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5419315076776275e-06,
+      "loss": 0.4512,
+      "step": 18595
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5411638066027113e-06,
+      "loss": 0.437,
+      "step": 18596
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5403962807312535e-06,
+      "loss": 0.469,
+      "step": 18597
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5396289300791501e-06,
+      "loss": 0.4927,
+      "step": 18598
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5388617546623007e-06,
+      "loss": 0.4512,
+      "step": 18599
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5380947544965895e-06,
+      "loss": 0.4995,
+      "step": 18600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5373279295979081e-06,
+      "loss": 0.4333,
+      "step": 18601
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.536561279982135e-06,
+      "loss": 0.467,
+      "step": 18602
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5357948056651528e-06,
+      "loss": 0.4983,
+      "step": 18603
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5350285066628346e-06,
+      "loss": 0.3521,
+      "step": 18604
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5342623829910542e-06,
+      "loss": 0.5359,
+      "step": 18605
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5334964346656777e-06,
+      "loss": 0.4631,
+      "step": 18606
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5327306617025729e-06,
+      "loss": 0.4348,
+      "step": 18607
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5319650641175966e-06,
+      "loss": 0.4885,
+      "step": 18608
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5311996419266118e-06,
+      "loss": 0.4727,
+      "step": 18609
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.530434395145466e-06,
+      "loss": 0.405,
+      "step": 18610
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5296693237900161e-06,
+      "loss": 0.5017,
+      "step": 18611
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5289044278761046e-06,
+      "loss": 0.4287,
+      "step": 18612
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5281397074195724e-06,
+      "loss": 0.4756,
+      "step": 18613
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.527375162436261e-06,
+      "loss": 0.4966,
+      "step": 18614
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.526610792942008e-06,
+      "loss": 0.4924,
+      "step": 18615
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5258465989526428e-06,
+      "loss": 0.4812,
+      "step": 18616
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5250825804839965e-06,
+      "loss": 0.376,
+      "step": 18617
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5243187375518897e-06,
+      "loss": 0.4431,
+      "step": 18618
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5235550701721446e-06,
+      "loss": 0.4165,
+      "step": 18619
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.522791578360583e-06,
+      "loss": 0.4797,
+      "step": 18620
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.522028262133013e-06,
+      "loss": 0.4302,
+      "step": 18621
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5212651215052498e-06,
+      "loss": 0.4602,
+      "step": 18622
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5205021564930944e-06,
+      "loss": 0.4175,
+      "step": 18623
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5197393671123562e-06,
+      "loss": 0.4727,
+      "step": 18624
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5189767533788268e-06,
+      "loss": 0.428,
+      "step": 18625
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5182143153083095e-06,
+      "loss": 0.4585,
+      "step": 18626
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5174520529165926e-06,
+      "loss": 0.4434,
+      "step": 18627
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5166899662194623e-06,
+      "loss": 0.4326,
+      "step": 18628
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.515928055232707e-06,
+      "loss": 0.3999,
+      "step": 18629
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5151663199721056e-06,
+      "loss": 0.4163,
+      "step": 18630
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5144047604534372e-06,
+      "loss": 0.4624,
+      "step": 18631
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.513643376692473e-06,
+      "loss": 0.4583,
+      "step": 18632
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5128821687049866e-06,
+      "loss": 0.459,
+      "step": 18633
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5121211365067401e-06,
+      "loss": 0.4233,
+      "step": 18634
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.511360280113502e-06,
+      "loss": 0.4829,
+      "step": 18635
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5105995995410261e-06,
+      "loss": 0.4637,
+      "step": 18636
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5098390948050723e-06,
+      "loss": 0.5361,
+      "step": 18637
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.509078765921389e-06,
+      "loss": 0.4482,
+      "step": 18638
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.508318612905728e-06,
+      "loss": 0.4426,
+      "step": 18639
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5075586357738291e-06,
+      "loss": 0.4475,
+      "step": 18640
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5067988345414397e-06,
+      "loss": 0.4224,
+      "step": 18641
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5060392092242904e-06,
+      "loss": 0.4431,
+      "step": 18642
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5052797598381185e-06,
+      "loss": 0.4414,
+      "step": 18643
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5045204863986562e-06,
+      "loss": 0.4705,
+      "step": 18644
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5037613889216252e-06,
+      "loss": 0.4353,
+      "step": 18645
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5030024674227529e-06,
+      "loss": 0.4531,
+      "step": 18646
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5022437219177544e-06,
+      "loss": 0.4548,
+      "step": 18647
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5014851524223483e-06,
+      "loss": 0.4509,
+      "step": 18648
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5007267589522435e-06,
+      "loss": 0.4709,
+      "step": 18649
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4999685415231525e-06,
+      "loss": 0.4675,
+      "step": 18650
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.499210500150774e-06,
+      "loss": 0.4382,
+      "step": 18651
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4984526348508144e-06,
+      "loss": 0.4331,
+      "step": 18652
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4976949456389678e-06,
+      "loss": 0.4441,
+      "step": 18653
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4969374325309294e-06,
+      "loss": 0.4419,
+      "step": 18654
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4961800955423866e-06,
+      "loss": 0.4214,
+      "step": 18655
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4954229346890303e-06,
+      "loss": 0.3907,
+      "step": 18656
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4946659499865402e-06,
+      "loss": 0.4229,
+      "step": 18657
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4939091414505936e-06,
+      "loss": 0.4978,
+      "step": 18658
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4931525090968702e-06,
+      "loss": 0.4155,
+      "step": 18659
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4923960529410375e-06,
+      "loss": 0.4309,
+      "step": 18660
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4916397729987676e-06,
+      "loss": 0.4451,
+      "step": 18661
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4908836692857198e-06,
+      "loss": 0.4568,
+      "step": 18662
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4901277418175607e-06,
+      "loss": 0.4326,
+      "step": 18663
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4893719906099424e-06,
+      "loss": 0.4741,
+      "step": 18664
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.488616415678522e-06,
+      "loss": 0.4624,
+      "step": 18665
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4878610170389473e-06,
+      "loss": 0.4902,
+      "step": 18666
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4871057947068646e-06,
+      "loss": 0.4348,
+      "step": 18667
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4863507486979189e-06,
+      "loss": 0.4456,
+      "step": 18668
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4855958790277448e-06,
+      "loss": 0.5212,
+      "step": 18669
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4848411857119794e-06,
+      "loss": 0.4854,
+      "step": 18670
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.484086668766258e-06,
+      "loss": 0.4246,
+      "step": 18671
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.483332328206205e-06,
+      "loss": 0.4878,
+      "step": 18672
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4825781640474423e-06,
+      "loss": 0.46,
+      "step": 18673
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4818241763055963e-06,
+      "loss": 0.4314,
+      "step": 18674
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4810703649962776e-06,
+      "loss": 0.4207,
+      "step": 18675
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.480316730135105e-06,
+      "loss": 0.428,
+      "step": 18676
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.479563271737684e-06,
+      "loss": 0.3943,
+      "step": 18677
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.478809989819624e-06,
+      "loss": 0.5215,
+      "step": 18678
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4780568843965238e-06,
+      "loss": 0.4133,
+      "step": 18679
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4773039554839862e-06,
+      "loss": 0.4221,
+      "step": 18680
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4765512030976026e-06,
+      "loss": 0.4448,
+      "step": 18681
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4757986272529667e-06,
+      "loss": 0.4153,
+      "step": 18682
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.475046227965664e-06,
+      "loss": 0.4856,
+      "step": 18683
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4742940052512822e-06,
+      "loss": 0.4768,
+      "step": 18684
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4735419591253963e-06,
+      "loss": 0.4419,
+      "step": 18685
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4727900896035896e-06,
+      "loss": 0.4434,
+      "step": 18686
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4720383967014307e-06,
+      "loss": 0.4333,
+      "step": 18687
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4712868804344881e-06,
+      "loss": 0.4255,
+      "step": 18688
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.470535540818332e-06,
+      "loss": 0.4189,
+      "step": 18689
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4697843778685195e-06,
+      "loss": 0.4282,
+      "step": 18690
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4690333916006117e-06,
+      "loss": 0.4678,
+      "step": 18691
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4682825820301649e-06,
+      "loss": 0.4521,
+      "step": 18692
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4675319491727268e-06,
+      "loss": 0.4229,
+      "step": 18693
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4667814930438483e-06,
+      "loss": 0.4492,
+      "step": 18694
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4660312136590705e-06,
+      "loss": 0.468,
+      "step": 18695
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4652811110339338e-06,
+      "loss": 0.4365,
+      "step": 18696
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.464531185183977e-06,
+      "loss": 0.4697,
+      "step": 18697
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4637814361247305e-06,
+      "loss": 0.4468,
+      "step": 18698
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4630318638717255e-06,
+      "loss": 0.4998,
+      "step": 18699
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4622824684404845e-06,
+      "loss": 0.4844,
+      "step": 18700
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.461533249846533e-06,
+      "loss": 0.4807,
+      "step": 18701
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.460784208105388e-06,
+      "loss": 0.4683,
+      "step": 18702
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.460035343232561e-06,
+      "loss": 0.4216,
+      "step": 18703
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.459286655243567e-06,
+      "loss": 0.4246,
+      "step": 18704
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4585381441539092e-06,
+      "loss": 0.4045,
+      "step": 18705
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4577898099790966e-06,
+      "loss": 0.4321,
+      "step": 18706
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.457041652734622e-06,
+      "loss": 0.4795,
+      "step": 18707
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4562936724359889e-06,
+      "loss": 0.4656,
+      "step": 18708
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4555458690986835e-06,
+      "loss": 0.4224,
+      "step": 18709
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.454798242738199e-06,
+      "loss": 0.4358,
+      "step": 18710
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4540507933700175e-06,
+      "loss": 0.4492,
+      "step": 18711
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4533035210096247e-06,
+      "loss": 0.46,
+      "step": 18712
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4525564256724934e-06,
+      "loss": 0.459,
+      "step": 18713
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4518095073741013e-06,
+      "loss": 0.3984,
+      "step": 18714
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.451062766129917e-06,
+      "loss": 0.4758,
+      "step": 18715
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4503162019554097e-06,
+      "loss": 0.4354,
+      "step": 18716
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4495698148660387e-06,
+      "loss": 0.4644,
+      "step": 18717
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4488236048772685e-06,
+      "loss": 0.5298,
+      "step": 18718
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4480775720045504e-06,
+      "loss": 0.416,
+      "step": 18719
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4473317162633383e-06,
+      "loss": 0.4321,
+      "step": 18720
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4465860376690833e-06,
+      "loss": 0.4392,
+      "step": 18721
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4458405362372263e-06,
+      "loss": 0.4358,
+      "step": 18722
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4450952119832117e-06,
+      "loss": 0.4214,
+      "step": 18723
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4443500649224729e-06,
+      "loss": 0.4792,
+      "step": 18724
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4436050950704494e-06,
+      "loss": 0.429,
+      "step": 18725
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4428603024425647e-06,
+      "loss": 0.4521,
+      "step": 18726
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4421156870542518e-06,
+      "loss": 0.4236,
+      "step": 18727
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4413712489209287e-06,
+      "loss": 0.4216,
+      "step": 18728
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4406269880580193e-06,
+      "loss": 0.4929,
+      "step": 18729
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4398829044809326e-06,
+      "loss": 0.4324,
+      "step": 18730
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4391389982050873e-06,
+      "loss": 0.3899,
+      "step": 18731
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4383952692458881e-06,
+      "loss": 0.428,
+      "step": 18732
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4376517176187365e-06,
+      "loss": 0.4434,
+      "step": 18733
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4369083433390396e-06,
+      "loss": 0.4211,
+      "step": 18734
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4361651464221882e-06,
+      "loss": 0.5398,
+      "step": 18735
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4354221268835823e-06,
+      "loss": 0.4719,
+      "step": 18736
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.434679284738606e-06,
+      "loss": 0.4797,
+      "step": 18737
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4339366200026495e-06,
+      "loss": 0.4183,
+      "step": 18738
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4331941326910914e-06,
+      "loss": 0.5454,
+      "step": 18739
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.432451822819314e-06,
+      "loss": 0.4939,
+      "step": 18740
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4317096904026894e-06,
+      "loss": 0.4902,
+      "step": 18741
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4309677354565932e-06,
+      "loss": 0.4636,
+      "step": 18742
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4302259579963872e-06,
+      "loss": 0.4363,
+      "step": 18743
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4294843580374385e-06,
+      "loss": 0.4568,
+      "step": 18744
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.428742935595111e-06,
+      "loss": 0.4751,
+      "step": 18745
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.428001690684755e-06,
+      "loss": 0.4246,
+      "step": 18746
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.427260623321729e-06,
+      "loss": 0.5178,
+      "step": 18747
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4265197335213764e-06,
+      "loss": 0.4155,
+      "step": 18748
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4257790212990507e-06,
+      "loss": 0.499,
+      "step": 18749
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4250384866700863e-06,
+      "loss": 0.4187,
+      "step": 18750
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4242981296498258e-06,
+      "loss": 0.3933,
+      "step": 18751
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4235579502536022e-06,
+      "loss": 0.449,
+      "step": 18752
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4228179484967487e-06,
+      "loss": 0.4866,
+      "step": 18753
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4220781243945881e-06,
+      "loss": 0.4482,
+      "step": 18754
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4213384779624494e-06,
+      "loss": 0.4417,
+      "step": 18755
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4205990092156486e-06,
+      "loss": 0.4875,
+      "step": 18756
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.419859718169504e-06,
+      "loss": 0.5054,
+      "step": 18757
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4191206048393259e-06,
+      "loss": 0.4133,
+      "step": 18758
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4183816692404273e-06,
+      "loss": 0.4312,
+      "step": 18759
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4176429113881074e-06,
+      "loss": 0.425,
+      "step": 18760
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4169043312976737e-06,
+      "loss": 0.3958,
+      "step": 18761
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.416165928984421e-06,
+      "loss": 0.4993,
+      "step": 18762
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4154277044636423e-06,
+      "loss": 0.4717,
+      "step": 18763
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.414689657750632e-06,
+      "loss": 0.4668,
+      "step": 18764
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4139517888606714e-06,
+      "loss": 0.4836,
+      "step": 18765
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.413214097809048e-06,
+      "loss": 0.4558,
+      "step": 18766
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4124765846110388e-06,
+      "loss": 0.458,
+      "step": 18767
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4117392492819205e-06,
+      "loss": 0.4402,
+      "step": 18768
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4110020918369672e-06,
+      "loss": 0.4011,
+      "step": 18769
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4102651122914434e-06,
+      "loss": 0.4797,
+      "step": 18770
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4095283106606173e-06,
+      "loss": 0.5005,
+      "step": 18771
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4087916869597462e-06,
+      "loss": 0.4565,
+      "step": 18772
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4080552412040893e-06,
+      "loss": 0.4615,
+      "step": 18773
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4073189734089032e-06,
+      "loss": 0.4106,
+      "step": 18774
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4065828835894323e-06,
+      "loss": 0.4016,
+      "step": 18775
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.405846971760928e-06,
+      "loss": 0.48,
+      "step": 18776
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.40511123793863e-06,
+      "loss": 0.4448,
+      "step": 18777
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4043756821377751e-06,
+      "loss": 0.4783,
+      "step": 18778
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4036403043736034e-06,
+      "loss": 0.4182,
+      "step": 18779
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4029051046613418e-06,
+      "loss": 0.448,
+      "step": 18780
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.402170083016222e-06,
+      "loss": 0.5078,
+      "step": 18781
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4014352394534647e-06,
+      "loss": 0.3777,
+      "step": 18782
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.400700573988294e-06,
+      "loss": 0.3989,
+      "step": 18783
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3999660866359222e-06,
+      "loss": 0.4539,
+      "step": 18784
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3992317774115672e-06,
+      "loss": 0.4956,
+      "step": 18785
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3984976463304334e-06,
+      "loss": 0.4395,
+      "step": 18786
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.397763693407731e-06,
+      "loss": 0.5002,
+      "step": 18787
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.397029918658659e-06,
+      "loss": 0.4373,
+      "step": 18788
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.396296322098417e-06,
+      "loss": 0.4395,
+      "step": 18789
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.395562903742199e-06,
+      "loss": 0.4275,
+      "step": 18790
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3948296636051973e-06,
+      "loss": 0.5161,
+      "step": 18791
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3940966017025981e-06,
+      "loss": 0.48,
+      "step": 18792
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.393363718049583e-06,
+      "loss": 0.428,
+      "step": 18793
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.392631012661334e-06,
+      "loss": 0.438,
+      "step": 18794
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.391898485553028e-06,
+      "loss": 0.4368,
+      "step": 18795
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3911661367398355e-06,
+      "loss": 0.4014,
+      "step": 18796
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3904339662369249e-06,
+      "loss": 0.4597,
+      "step": 18797
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3897019740594653e-06,
+      "loss": 0.4631,
+      "step": 18798
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3889701602226135e-06,
+      "loss": 0.4402,
+      "step": 18799
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3882385247415298e-06,
+      "loss": 0.3701,
+      "step": 18800
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3875070676313663e-06,
+      "loss": 0.4331,
+      "step": 18801
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3867757889072763e-06,
+      "loss": 0.4683,
+      "step": 18802
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3860446885844026e-06,
+      "loss": 0.4939,
+      "step": 18803
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3853137666778915e-06,
+      "loss": 0.4194,
+      "step": 18804
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.384583023202881e-06,
+      "loss": 0.4888,
+      "step": 18805
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.383852458174504e-06,
+      "loss": 0.4263,
+      "step": 18806
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3831220716078963e-06,
+      "loss": 0.4517,
+      "step": 18807
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.382391863518182e-06,
+      "loss": 0.4983,
+      "step": 18808
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3816618339204902e-06,
+      "loss": 0.3879,
+      "step": 18809
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3809319828299372e-06,
+      "loss": 0.4336,
+      "step": 18810
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3802023102616447e-06,
+      "loss": 0.3779,
+      "step": 18811
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3794728162307203e-06,
+      "loss": 0.4291,
+      "step": 18812
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3787435007522786e-06,
+      "loss": 0.4651,
+      "step": 18813
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3780143638414223e-06,
+      "loss": 0.4583,
+      "step": 18814
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3772854055132567e-06,
+      "loss": 0.4934,
+      "step": 18815
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3765566257828765e-06,
+      "loss": 0.5066,
+      "step": 18816
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.375828024665381e-06,
+      "loss": 0.4126,
+      "step": 18817
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3750996021758568e-06,
+      "loss": 0.4199,
+      "step": 18818
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.374371358329395e-06,
+      "loss": 0.4458,
+      "step": 18819
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.373643293141076e-06,
+      "loss": 0.4568,
+      "step": 18820
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.372915406625982e-06,
+      "loss": 0.4753,
+      "step": 18821
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.372187698799191e-06,
+      "loss": 0.4229,
+      "step": 18822
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3714601696757713e-06,
+      "loss": 0.4419,
+      "step": 18823
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3707328192707958e-06,
+      "loss": 0.4463,
+      "step": 18824
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3700056475993273e-06,
+      "loss": 0.457,
+      "step": 18825
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3692786546764291e-06,
+      "loss": 0.4143,
+      "step": 18826
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.368551840517156e-06,
+      "loss": 0.4412,
+      "step": 18827
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.367825205136567e-06,
+      "loss": 0.4641,
+      "step": 18828
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3670987485497067e-06,
+      "loss": 0.4087,
+      "step": 18829
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3663724707716276e-06,
+      "loss": 0.4343,
+      "step": 18830
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3656463718173674e-06,
+      "loss": 0.4487,
+      "step": 18831
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3649204517019699e-06,
+      "loss": 0.5015,
+      "step": 18832
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3641947104404663e-06,
+      "loss": 0.4351,
+      "step": 18833
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3634691480478934e-06,
+      "loss": 0.4546,
+      "step": 18834
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3627437645392771e-06,
+      "loss": 0.4368,
+      "step": 18835
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3620185599296398e-06,
+      "loss": 0.4463,
+      "step": 18836
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.361293534234006e-06,
+      "loss": 0.4507,
+      "step": 18837
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3605686874673885e-06,
+      "loss": 0.4578,
+      "step": 18838
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.359844019644806e-06,
+      "loss": 0.4688,
+      "step": 18839
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.359119530781262e-06,
+      "loss": 0.4226,
+      "step": 18840
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3583952208917694e-06,
+      "loss": 0.4255,
+      "step": 18841
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3576710899913237e-06,
+      "loss": 0.4072,
+      "step": 18842
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3569471380949284e-06,
+      "loss": 0.4719,
+      "step": 18843
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.356223365217575e-06,
+      "loss": 0.4136,
+      "step": 18844
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3554997713742557e-06,
+      "loss": 0.5066,
+      "step": 18845
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.354776356579961e-06,
+      "loss": 0.4976,
+      "step": 18846
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3540531208496698e-06,
+      "loss": 0.4214,
+      "step": 18847
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.353330064198366e-06,
+      "loss": 0.4309,
+      "step": 18848
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3526071866410229e-06,
+      "loss": 0.4148,
+      "step": 18849
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3518844881926153e-06,
+      "loss": 0.4939,
+      "step": 18850
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3511619688681099e-06,
+      "loss": 0.4194,
+      "step": 18851
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.350439628682475e-06,
+      "loss": 0.5046,
+      "step": 18852
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3497174676506675e-06,
+      "loss": 0.4248,
+      "step": 18853
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3489954857876497e-06,
+      "loss": 0.4392,
+      "step": 18854
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.348273683108372e-06,
+      "loss": 0.479,
+      "step": 18855
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3475520596277891e-06,
+      "loss": 0.4377,
+      "step": 18856
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3468306153608423e-06,
+      "loss": 0.4419,
+      "step": 18857
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3461093503224798e-06,
+      "loss": 0.4595,
+      "step": 18858
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.345388264527635e-06,
+      "loss": 0.4354,
+      "step": 18859
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3446673579912495e-06,
+      "loss": 0.4375,
+      "step": 18860
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.34394663072825e-06,
+      "loss": 0.4629,
+      "step": 18861
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3432260827535682e-06,
+      "loss": 0.3997,
+      "step": 18862
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3425057140821252e-06,
+      "loss": 0.4382,
+      "step": 18863
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3417855247288458e-06,
+      "loss": 0.4226,
+      "step": 18864
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3410655147086439e-06,
+      "loss": 0.4553,
+      "step": 18865
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3403456840364305e-06,
+      "loss": 0.425,
+      "step": 18866
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3396260327271204e-06,
+      "loss": 0.4797,
+      "step": 18867
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3389065607956141e-06,
+      "loss": 0.4176,
+      "step": 18868
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3381872682568186e-06,
+      "loss": 0.4561,
+      "step": 18869
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3374681551256286e-06,
+      "loss": 0.4778,
+      "step": 18870
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3367492214169386e-06,
+      "loss": 0.4258,
+      "step": 18871
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3360304671456449e-06,
+      "loss": 0.4435,
+      "step": 18872
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3353118923266262e-06,
+      "loss": 0.4504,
+      "step": 18873
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.334593496974772e-06,
+      "loss": 0.4419,
+      "step": 18874
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3338752811049627e-06,
+      "loss": 0.4849,
+      "step": 18875
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3331572447320695e-06,
+      "loss": 0.4585,
+      "step": 18876
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3324393878709697e-06,
+      "loss": 0.4731,
+      "step": 18877
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3317217105365277e-06,
+      "loss": 0.3789,
+      "step": 18878
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3310042127436118e-06,
+      "loss": 0.4304,
+      "step": 18879
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3302868945070812e-06,
+      "loss": 0.4875,
+      "step": 18880
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.329569755841792e-06,
+      "loss": 0.4578,
+      "step": 18881
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.328852796762602e-06,
+      "loss": 0.4675,
+      "step": 18882
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3281360172843572e-06,
+      "loss": 0.5159,
+      "step": 18883
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.327419417421907e-06,
+      "loss": 0.4729,
+      "step": 18884
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3267029971900902e-06,
+      "loss": 0.4817,
+      "step": 18885
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.32598675660375e-06,
+      "loss": 0.4258,
+      "step": 18886
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3252706956777173e-06,
+      "loss": 0.4351,
+      "step": 18887
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3245548144268284e-06,
+      "loss": 0.4624,
+      "step": 18888
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3238391128659046e-06,
+      "loss": 0.4858,
+      "step": 18889
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3231235910097773e-06,
+      "loss": 0.4033,
+      "step": 18890
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3224082488732593e-06,
+      "loss": 0.4851,
+      "step": 18891
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.321693086471174e-06,
+      "loss": 0.4302,
+      "step": 18892
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3209781038183279e-06,
+      "loss": 0.4338,
+      "step": 18893
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3202633009295352e-06,
+      "loss": 0.3794,
+      "step": 18894
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3195486778195977e-06,
+      "loss": 0.4675,
+      "step": 18895
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.31883423450332e-06,
+      "loss": 0.47,
+      "step": 18896
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3181199709954962e-06,
+      "loss": 0.4397,
+      "step": 18897
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.317405887310923e-06,
+      "loss": 0.4412,
+      "step": 18898
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.316691983464392e-06,
+      "loss": 0.4138,
+      "step": 18899
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.315978259470687e-06,
+      "loss": 0.4714,
+      "step": 18900
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3152647153445941e-06,
+      "loss": 0.4373,
+      "step": 18901
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.314551351100889e-06,
+      "loss": 0.4795,
+      "step": 18902
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.313838166754351e-06,
+      "loss": 0.4778,
+      "step": 18903
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3131251623197484e-06,
+      "loss": 0.4243,
+      "step": 18904
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3124123378118526e-06,
+      "loss": 0.4502,
+      "step": 18905
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3116996932454241e-06,
+      "loss": 0.4688,
+      "step": 18906
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3109872286352276e-06,
+      "loss": 0.4641,
+      "step": 18907
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3102749439960161e-06,
+      "loss": 0.4307,
+      "step": 18908
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3095628393425474e-06,
+      "loss": 0.4841,
+      "step": 18909
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3088509146895677e-06,
+      "loss": 0.3838,
+      "step": 18910
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3081391700518209e-06,
+      "loss": 0.4409,
+      "step": 18911
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.307427605444055e-06,
+      "loss": 0.4294,
+      "step": 18912
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3067162208810014e-06,
+      "loss": 0.4788,
+      "step": 18913
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3060050163774007e-06,
+      "loss": 0.4495,
+      "step": 18914
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.305293991947979e-06,
+      "loss": 0.4048,
+      "step": 18915
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3045831476074665e-06,
+      "loss": 0.4299,
+      "step": 18916
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3038724833705851e-06,
+      "loss": 0.4482,
+      "step": 18917
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.303161999252055e-06,
+      "loss": 0.4539,
+      "step": 18918
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3024516952665912e-06,
+      "loss": 0.4028,
+      "step": 18919
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3017415714289073e-06,
+      "loss": 0.4377,
+      "step": 18920
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3010316277537094e-06,
+      "loss": 0.4012,
+      "step": 18921
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3003218642557035e-06,
+      "loss": 0.3484,
+      "step": 18922
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2996122809495926e-06,
+      "loss": 0.4417,
+      "step": 18923
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2989028778500701e-06,
+      "loss": 0.4546,
+      "step": 18924
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2981936549718343e-06,
+      "loss": 0.4829,
+      "step": 18925
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2974846123295692e-06,
+      "loss": 0.4424,
+      "step": 18926
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2967757499379662e-06,
+      "loss": 0.395,
+      "step": 18927
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2960670678117026e-06,
+      "loss": 0.4121,
+      "step": 18928
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2953585659654622e-06,
+      "loss": 0.4573,
+      "step": 18929
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2946502444139154e-06,
+      "loss": 0.5166,
+      "step": 18930
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.293942103171737e-06,
+      "loss": 0.4524,
+      "step": 18931
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.29323414225359e-06,
+      "loss": 0.457,
+      "step": 18932
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2925263616741424e-06,
+      "loss": 0.4778,
+      "step": 18933
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2918187614480493e-06,
+      "loss": 0.4407,
+      "step": 18934
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2911113415899734e-06,
+      "loss": 0.4998,
+      "step": 18935
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2904041021145597e-06,
+      "loss": 0.4458,
+      "step": 18936
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2896970430364642e-06,
+      "loss": 0.4158,
+      "step": 18937
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.288990164370325e-06,
+      "loss": 0.4473,
+      "step": 18938
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2882834661307886e-06,
+      "loss": 0.4578,
+      "step": 18939
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2875769483324907e-06,
+      "loss": 0.4293,
+      "step": 18940
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.286870610990063e-06,
+      "loss": 0.4436,
+      "step": 18941
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2861644541181396e-06,
+      "loss": 0.437,
+      "step": 18942
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2854584777313406e-06,
+      "loss": 0.481,
+      "step": 18943
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2847526818442957e-06,
+      "loss": 0.462,
+      "step": 18944
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2840470664716188e-06,
+      "loss": 0.4697,
+      "step": 18945
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.283341631627928e-06,
+      "loss": 0.4382,
+      "step": 18946
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2826363773278305e-06,
+      "loss": 0.4207,
+      "step": 18947
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.281931303585936e-06,
+      "loss": 0.4512,
+      "step": 18948
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2812264104168514e-06,
+      "loss": 0.4175,
+      "step": 18949
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.280521697835172e-06,
+      "loss": 0.4531,
+      "step": 18950
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.279817165855497e-06,
+      "loss": 0.4426,
+      "step": 18951
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2791128144924193e-06,
+      "loss": 0.4751,
+      "step": 18952
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2784086437605249e-06,
+      "loss": 0.4543,
+      "step": 18953
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.277704653674403e-06,
+      "loss": 0.4973,
+      "step": 18954
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2770008442486327e-06,
+      "loss": 0.4756,
+      "step": 18955
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2762972154977893e-06,
+      "loss": 0.4478,
+      "step": 18956
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2755937674364516e-06,
+      "loss": 0.4675,
+      "step": 18957
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2748905000791844e-06,
+      "loss": 0.4543,
+      "step": 18958
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2741874134405596e-06,
+      "loss": 0.4773,
+      "step": 18959
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2734845075351355e-06,
+      "loss": 0.4741,
+      "step": 18960
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2727817823774746e-06,
+      "loss": 0.4449,
+      "step": 18961
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2720792379821277e-06,
+      "loss": 0.4707,
+      "step": 18962
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.27137687436365e-06,
+      "loss": 0.4526,
+      "step": 18963
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2706746915365875e-06,
+      "loss": 0.396,
+      "step": 18964
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.269972689515485e-06,
+      "loss": 0.467,
+      "step": 18965
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2692708683148814e-06,
+      "loss": 0.3975,
+      "step": 18966
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2685692279493157e-06,
+      "loss": 0.4174,
+      "step": 18967
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2678677684333174e-06,
+      "loss": 0.5161,
+      "step": 18968
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2671664897814184e-06,
+      "loss": 0.4907,
+      "step": 18969
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2664653920081426e-06,
+      "loss": 0.3945,
+      "step": 18970
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2657644751280096e-06,
+      "loss": 0.4138,
+      "step": 18971
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2650637391555388e-06,
+      "loss": 0.4761,
+      "step": 18972
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2643631841052462e-06,
+      "loss": 0.4524,
+      "step": 18973
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2636628099916381e-06,
+      "loss": 0.5776,
+      "step": 18974
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.262962616829223e-06,
+      "loss": 0.4084,
+      "step": 18975
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.262262604632506e-06,
+      "loss": 0.4307,
+      "step": 18976
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2615627734159807e-06,
+      "loss": 0.4397,
+      "step": 18977
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2608631231941482e-06,
+      "loss": 0.4153,
+      "step": 18978
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2601636539814954e-06,
+      "loss": 0.4932,
+      "step": 18979
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2594643657925132e-06,
+      "loss": 0.4731,
+      "step": 18980
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2587652586416822e-06,
+      "loss": 0.4458,
+      "step": 18981
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2580663325434872e-06,
+      "loss": 0.4502,
+      "step": 18982
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.257367587512399e-06,
+      "loss": 0.3992,
+      "step": 18983
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2566690235628965e-06,
+      "loss": 0.4268,
+      "step": 18984
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2559706407094463e-06,
+      "loss": 0.502,
+      "step": 18985
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2552724389665093e-06,
+      "loss": 0.4382,
+      "step": 18986
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2545744183485542e-06,
+      "loss": 0.4321,
+      "step": 18987
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2538765788700313e-06,
+      "loss": 0.4937,
+      "step": 18988
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2531789205454025e-06,
+      "loss": 0.4395,
+      "step": 18989
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2524814433891108e-06,
+      "loss": 0.5156,
+      "step": 18990
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.251784147415609e-06,
+      "loss": 0.478,
+      "step": 18991
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2510870326393332e-06,
+      "loss": 0.4534,
+      "step": 18992
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2503900990747285e-06,
+      "loss": 0.4099,
+      "step": 18993
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.249693346736226e-06,
+      "loss": 0.4272,
+      "step": 18994
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2489967756382604e-06,
+      "loss": 0.4187,
+      "step": 18995
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2483003857952557e-06,
+      "loss": 0.4993,
+      "step": 18996
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2476041772216407e-06,
+      "loss": 0.3903,
+      "step": 18997
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2469081499318304e-06,
+      "loss": 0.5193,
+      "step": 18998
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2462123039402429e-06,
+      "loss": 0.4292,
+      "step": 18999
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2455166392612939e-06,
+      "loss": 0.5127,
+      "step": 19000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.244821155909388e-06,
+      "loss": 0.4563,
+      "step": 19001
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2441258538989353e-06,
+      "loss": 0.4309,
+      "step": 19002
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2434307332443318e-06,
+      "loss": 0.4292,
+      "step": 19003
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2427357939599794e-06,
+      "loss": 0.4492,
+      "step": 19004
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2420410360602686e-06,
+      "loss": 0.4482,
+      "step": 19005
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2413464595595937e-06,
+      "loss": 0.5652,
+      "step": 19006
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2406520644723363e-06,
+      "loss": 0.3733,
+      "step": 19007
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2399578508128829e-06,
+      "loss": 0.4734,
+      "step": 19008
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2392638185956096e-06,
+      "loss": 0.4534,
+      "step": 19009
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.238569967834895e-06,
+      "loss": 0.4634,
+      "step": 19010
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2378762985451064e-06,
+      "loss": 0.4951,
+      "step": 19011
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2371828107406159e-06,
+      "loss": 0.4888,
+      "step": 19012
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2364895044357838e-06,
+      "loss": 0.4358,
+      "step": 19013
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.235796379644969e-06,
+      "loss": 0.4868,
+      "step": 19014
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2351034363825332e-06,
+      "loss": 0.4351,
+      "step": 19015
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2344106746628225e-06,
+      "loss": 0.4983,
+      "step": 19016
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2337180945001926e-06,
+      "loss": 0.4117,
+      "step": 19017
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2330256959089816e-06,
+      "loss": 0.3953,
+      "step": 19018
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2323334789035368e-06,
+      "loss": 0.3965,
+      "step": 19019
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2316414434981906e-06,
+      "loss": 0.4578,
+      "step": 19020
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.230949589707282e-06,
+      "loss": 0.5083,
+      "step": 19021
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2302579175451356e-06,
+      "loss": 0.4731,
+      "step": 19022
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2295664270260822e-06,
+      "loss": 0.5061,
+      "step": 19023
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.228875118164441e-06,
+      "loss": 0.4019,
+      "step": 19024
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2281839909745318e-06,
+      "loss": 0.4727,
+      "step": 19025
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2274930454706712e-06,
+      "loss": 0.478,
+      "step": 19026
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2268022816671676e-06,
+      "loss": 0.4358,
+      "step": 19027
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.226111699578333e-06,
+      "loss": 0.4487,
+      "step": 19028
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2254212992184644e-06,
+      "loss": 0.4424,
+      "step": 19029
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2247310806018675e-06,
+      "loss": 0.415,
+      "step": 19030
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2240410437428352e-06,
+      "loss": 0.4324,
+      "step": 19031
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2233511886556615e-06,
+      "loss": 0.4624,
+      "step": 19032
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2226615153546339e-06,
+      "loss": 0.427,
+      "step": 19033
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.221972023854039e-06,
+      "loss": 0.4778,
+      "step": 19034
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2212827141681539e-06,
+      "loss": 0.4364,
+      "step": 19035
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2205935863112618e-06,
+      "loss": 0.4351,
+      "step": 19036
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.21990464029763e-06,
+      "loss": 0.4191,
+      "step": 19037
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.219215876141534e-06,
+      "loss": 0.5127,
+      "step": 19038
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2185272938572345e-06,
+      "loss": 0.4646,
+      "step": 19039
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.217838893458999e-06,
+      "loss": 0.4265,
+      "step": 19040
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2171506749610817e-06,
+      "loss": 0.4695,
+      "step": 19041
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.21646263837774e-06,
+      "loss": 0.4585,
+      "step": 19042
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2157747837232249e-06,
+      "loss": 0.4783,
+      "step": 19043
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.215087111011779e-06,
+      "loss": 0.4822,
+      "step": 19044
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2143996202576525e-06,
+      "loss": 0.4705,
+      "step": 19045
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2137123114750792e-06,
+      "loss": 0.4331,
+      "step": 19046
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2130251846783003e-06,
+      "loss": 0.3785,
+      "step": 19047
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.212338239881542e-06,
+      "loss": 0.4006,
+      "step": 19048
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2116514770990362e-06,
+      "loss": 0.4988,
+      "step": 19049
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2109648963450093e-06,
+      "loss": 0.427,
+      "step": 19050
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.210278497633678e-06,
+      "loss": 0.4995,
+      "step": 19051
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2095922809792605e-06,
+      "loss": 0.5012,
+      "step": 19052
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2089062463959722e-06,
+      "loss": 0.5103,
+      "step": 19053
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2082203938980198e-06,
+      "loss": 0.4534,
+      "step": 19054
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2075347234996115e-06,
+      "loss": 0.4617,
+      "step": 19055
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2068492352149463e-06,
+      "loss": 0.4224,
+      "step": 19056
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2061639290582273e-06,
+      "loss": 0.4583,
+      "step": 19057
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2054788050436438e-06,
+      "loss": 0.4185,
+      "step": 19058
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2047938631853873e-06,
+      "loss": 0.3894,
+      "step": 19059
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.204109103497647e-06,
+      "loss": 0.4575,
+      "step": 19060
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2034245259946031e-06,
+      "loss": 0.4329,
+      "step": 19061
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2027401306904384e-06,
+      "loss": 0.4539,
+      "step": 19062
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.202055917599324e-06,
+      "loss": 0.4756,
+      "step": 19063
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2013718867354363e-06,
+      "loss": 0.447,
+      "step": 19064
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2006880381129383e-06,
+      "loss": 0.418,
+      "step": 19065
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2000043717459997e-06,
+      "loss": 0.4485,
+      "step": 19066
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1993208876487749e-06,
+      "loss": 0.4442,
+      "step": 19067
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.198637585835427e-06,
+      "loss": 0.4319,
+      "step": 19068
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1979544663201026e-06,
+      "loss": 0.397,
+      "step": 19069
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1972715291169567e-06,
+      "loss": 0.5515,
+      "step": 19070
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1965887742401294e-06,
+      "loss": 0.4724,
+      "step": 19071
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.195906201703767e-06,
+      "loss": 0.5049,
+      "step": 19072
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1952238115220028e-06,
+      "loss": 0.4287,
+      "step": 19073
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1945416037089742e-06,
+      "loss": 0.3916,
+      "step": 19074
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1938595782788088e-06,
+      "loss": 0.4521,
+      "step": 19075
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1931777352456341e-06,
+      "loss": 0.4507,
+      "step": 19076
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1924960746235748e-06,
+      "loss": 0.4941,
+      "step": 19077
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1918145964267458e-06,
+      "loss": 0.4795,
+      "step": 19078
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1911333006692672e-06,
+      "loss": 0.448,
+      "step": 19079
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1904521873652442e-06,
+      "loss": 0.4578,
+      "step": 19080
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1897712565287912e-06,
+      "loss": 0.4695,
+      "step": 19081
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1890905081740046e-06,
+      "loss": 0.3794,
+      "step": 19082
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.188409942314992e-06,
+      "loss": 0.4578,
+      "step": 19083
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1877295589658422e-06,
+      "loss": 0.4612,
+      "step": 19084
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.187049358140654e-06,
+      "loss": 0.4395,
+      "step": 19085
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1863693398535115e-06,
+      "loss": 0.386,
+      "step": 19086
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1856895041185024e-06,
+      "loss": 0.4905,
+      "step": 19087
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1850098509497076e-06,
+      "loss": 0.4348,
+      "step": 19088
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1843303803612006e-06,
+      "loss": 0.4253,
+      "step": 19089
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1836510923670607e-06,
+      "loss": 0.4648,
+      "step": 19090
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1829719869813526e-06,
+      "loss": 0.4761,
+      "step": 19091
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1822930642181463e-06,
+      "loss": 0.4324,
+      "step": 19092
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.181614324091499e-06,
+      "loss": 0.4792,
+      "step": 19093
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1809357666154752e-06,
+      "loss": 0.4631,
+      "step": 19094
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1802573918041249e-06,
+      "loss": 0.4429,
+      "step": 19095
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1795791996715022e-06,
+      "loss": 0.4773,
+      "step": 19096
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1789011902316505e-06,
+      "loss": 0.469,
+      "step": 19097
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1782233634986163e-06,
+      "loss": 0.4861,
+      "step": 19098
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1775457194864359e-06,
+      "loss": 0.4453,
+      "step": 19099
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1768682582091495e-06,
+      "loss": 0.4021,
+      "step": 19100
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1761909796807846e-06,
+      "loss": 0.4194,
+      "step": 19101
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1755138839153702e-06,
+      "loss": 0.45,
+      "step": 19102
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1748369709269336e-06,
+      "loss": 0.4985,
+      "step": 19103
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1741602407294916e-06,
+      "loss": 0.4641,
+      "step": 19104
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1734836933370642e-06,
+      "loss": 0.4485,
+      "step": 19105
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.172807328763661e-06,
+      "loss": 0.4849,
+      "step": 19106
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1721311470232955e-06,
+      "loss": 0.4575,
+      "step": 19107
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1714551481299686e-06,
+      "loss": 0.4426,
+      "step": 19108
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1707793320976846e-06,
+      "loss": 0.4124,
+      "step": 19109
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1701036989404402e-06,
+      "loss": 0.4387,
+      "step": 19110
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1694282486722307e-06,
+      "loss": 0.427,
+      "step": 19111
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.168752981307044e-06,
+      "loss": 0.3984,
+      "step": 19112
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1680778968588702e-06,
+      "loss": 0.479,
+      "step": 19113
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1674029953416865e-06,
+      "loss": 0.4478,
+      "step": 19114
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1667282767694787e-06,
+      "loss": 0.4373,
+      "step": 19115
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1660537411562157e-06,
+      "loss": 0.3998,
+      "step": 19116
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1653793885158736e-06,
+      "loss": 0.4688,
+      "step": 19117
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1647052188624175e-06,
+      "loss": 0.5049,
+      "step": 19118
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.164031232209809e-06,
+      "loss": 0.4546,
+      "step": 19119
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1633574285720129e-06,
+      "loss": 0.3979,
+      "step": 19120
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1626838079629809e-06,
+      "loss": 0.4773,
+      "step": 19121
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.162010370396669e-06,
+      "loss": 0.4519,
+      "step": 19122
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1613371158870223e-06,
+      "loss": 0.4343,
+      "step": 19123
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.160664044447989e-06,
+      "loss": 0.4019,
+      "step": 19124
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1599911560935074e-06,
+      "loss": 0.4812,
+      "step": 19125
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1593184508375143e-06,
+      "loss": 0.364,
+      "step": 19126
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1586459286939467e-06,
+      "loss": 0.4104,
+      "step": 19127
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1579735896767308e-06,
+      "loss": 0.4441,
+      "step": 19128
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1573014337997924e-06,
+      "loss": 0.4807,
+      "step": 19129
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1566294610770579e-06,
+      "loss": 0.4897,
+      "step": 19130
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1559576715224386e-06,
+      "loss": 0.4387,
+      "step": 19131
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1552860651498566e-06,
+      "loss": 0.4189,
+      "step": 19132
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1546146419732173e-06,
+      "loss": 0.4191,
+      "step": 19133
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1539434020064266e-06,
+      "loss": 0.4534,
+      "step": 19134
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1532723452633931e-06,
+      "loss": 0.4722,
+      "step": 19135
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1526014717580093e-06,
+      "loss": 0.4216,
+      "step": 19136
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.151930781504177e-06,
+      "loss": 0.4353,
+      "step": 19137
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1512602745157819e-06,
+      "loss": 0.4763,
+      "step": 19138
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1505899508067176e-06,
+      "loss": 0.407,
+      "step": 19139
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1499198103908638e-06,
+      "loss": 0.4069,
+      "step": 19140
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.149249853282104e-06,
+      "loss": 0.4089,
+      "step": 19141
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1485800794943104e-06,
+      "loss": 0.4275,
+      "step": 19142
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1479104890413618e-06,
+      "loss": 0.4595,
+      "step": 19143
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1472410819371205e-06,
+      "loss": 0.4868,
+      "step": 19144
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1465718581954566e-06,
+      "loss": 0.3569,
+      "step": 19145
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1459028178302279e-06,
+      "loss": 0.4871,
+      "step": 19146
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1452339608552953e-06,
+      "loss": 0.4534,
+      "step": 19147
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1445652872845103e-06,
+      "loss": 0.3485,
+      "step": 19148
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1438967971317216e-06,
+      "loss": 0.4734,
+      "step": 19149
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.143228490410777e-06,
+      "loss": 0.5293,
+      "step": 19150
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1425603671355202e-06,
+      "loss": 0.3896,
+      "step": 19151
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1418924273197863e-06,
+      "loss": 0.4683,
+      "step": 19152
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1412246709774122e-06,
+      "loss": 0.4246,
+      "step": 19153
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1405570981222302e-06,
+      "loss": 0.4202,
+      "step": 19154
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1398897087680626e-06,
+      "loss": 0.4011,
+      "step": 19155
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1392225029287385e-06,
+      "loss": 0.407,
+      "step": 19156
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1385554806180732e-06,
+      "loss": 0.4612,
+      "step": 19157
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1378886418498846e-06,
+      "loss": 0.4133,
+      "step": 19158
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1372219866379818e-06,
+      "loss": 0.4626,
+      "step": 19159
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.136555514996177e-06,
+      "loss": 0.5312,
+      "step": 19160
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1358892269382703e-06,
+      "loss": 0.429,
+      "step": 19161
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1352231224780664e-06,
+      "loss": 0.5034,
+      "step": 19162
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1345572016293594e-06,
+      "loss": 0.4382,
+      "step": 19163
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1338914644059408e-06,
+      "loss": 0.4575,
+      "step": 19164
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1332259108216037e-06,
+      "loss": 0.478,
+      "step": 19165
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1325605408901285e-06,
+      "loss": 0.4778,
+      "step": 19166
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.131895354625301e-06,
+      "loss": 0.4934,
+      "step": 19167
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1312303520408952e-06,
+      "loss": 0.4482,
+      "step": 19168
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1305655331506883e-06,
+      "loss": 0.4514,
+      "step": 19169
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1299008979684479e-06,
+      "loss": 0.4414,
+      "step": 19170
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.129236446507942e-06,
+      "loss": 0.4205,
+      "step": 19171
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1285721787829307e-06,
+      "loss": 0.4521,
+      "step": 19172
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1279080948071764e-06,
+      "loss": 0.4104,
+      "step": 19173
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1272441945944301e-06,
+      "loss": 0.4304,
+      "step": 19174
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1265804781584454e-06,
+      "loss": 0.4626,
+      "step": 19175
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.125916945512967e-06,
+      "loss": 0.4333,
+      "step": 19176
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1252535966717415e-06,
+      "loss": 0.4216,
+      "step": 19177
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1245904316485057e-06,
+      "loss": 0.5095,
+      "step": 19178
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1239274504569952e-06,
+      "loss": 0.4758,
+      "step": 19179
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.123264653110947e-06,
+      "loss": 0.438,
+      "step": 19180
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1226020396240822e-06,
+      "loss": 0.4736,
+      "step": 19181
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.121939610010131e-06,
+      "loss": 0.4971,
+      "step": 19182
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.121277364282809e-06,
+      "loss": 0.4465,
+      "step": 19183
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1206153024558375e-06,
+      "loss": 0.4856,
+      "step": 19184
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1199534245429256e-06,
+      "loss": 0.5027,
+      "step": 19185
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1192917305577865e-06,
+      "loss": 0.4895,
+      "step": 19186
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.118630220514121e-06,
+      "loss": 0.4272,
+      "step": 19187
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.117968894425635e-06,
+      "loss": 0.4661,
+      "step": 19188
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1173077523060216e-06,
+      "loss": 0.4006,
+      "step": 19189
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1166467941689796e-06,
+      "loss": 0.395,
+      "step": 19190
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1159860200281946e-06,
+      "loss": 0.4353,
+      "step": 19191
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.115325429897356e-06,
+      "loss": 0.4419,
+      "step": 19192
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1146650237901458e-06,
+      "loss": 0.4919,
+      "step": 19193
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1140048017202397e-06,
+      "loss": 0.3982,
+      "step": 19194
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1133447637013173e-06,
+      "loss": 0.4492,
+      "step": 19195
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1126849097470449e-06,
+      "loss": 0.3804,
+      "step": 19196
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1120252398710951e-06,
+      "loss": 0.4854,
+      "step": 19197
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1113657540871247e-06,
+      "loss": 0.4529,
+      "step": 19198
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1107064524087986e-06,
+      "loss": 0.437,
+      "step": 19199
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1100473348497699e-06,
+      "loss": 0.4382,
+      "step": 19200
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1093884014236921e-06,
+      "loss": 0.4487,
+      "step": 19201
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1087296521442114e-06,
+      "loss": 0.4131,
+      "step": 19202
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1080710870249734e-06,
+      "loss": 0.4709,
+      "step": 19203
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1074127060796202e-06,
+      "loss": 0.4807,
+      "step": 19204
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1067545093217857e-06,
+      "loss": 0.4399,
+      "step": 19205
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1060964967651033e-06,
+      "loss": 0.4973,
+      "step": 19206
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1054386684232043e-06,
+      "loss": 0.4395,
+      "step": 19207
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1047810243097123e-06,
+      "loss": 0.4824,
+      "step": 19208
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1041235644382475e-06,
+      "loss": 0.5024,
+      "step": 19209
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1034662888224313e-06,
+      "loss": 0.4805,
+      "step": 19210
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1028091974758714e-06,
+      "loss": 0.4685,
+      "step": 19211
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.102152290412185e-06,
+      "loss": 0.4004,
+      "step": 19212
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1014955676449723e-06,
+      "loss": 0.4248,
+      "step": 19213
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.10083902918784e-06,
+      "loss": 0.4094,
+      "step": 19214
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.100182675054382e-06,
+      "loss": 0.4343,
+      "step": 19215
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0995265052581982e-06,
+      "loss": 0.4619,
+      "step": 19216
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0988705198128758e-06,
+      "loss": 0.4795,
+      "step": 19217
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0982147187320035e-06,
+      "loss": 0.4714,
+      "step": 19218
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.097559102029163e-06,
+      "loss": 0.5237,
+      "step": 19219
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.096903669717938e-06,
+      "loss": 0.4441,
+      "step": 19220
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0962484218118997e-06,
+      "loss": 0.3767,
+      "step": 19221
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0955933583246192e-06,
+      "loss": 0.4314,
+      "step": 19222
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0949384792696693e-06,
+      "loss": 0.501,
+      "step": 19223
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0942837846606091e-06,
+      "loss": 0.4414,
+      "step": 19224
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0936292745110032e-06,
+      "loss": 0.4342,
+      "step": 19225
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0929749488344045e-06,
+      "loss": 0.3826,
+      "step": 19226
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.092320807644367e-06,
+      "loss": 0.45,
+      "step": 19227
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0916668509544425e-06,
+      "loss": 0.4885,
+      "step": 19228
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0910130787781715e-06,
+      "loss": 0.4565,
+      "step": 19229
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.090359491129096e-06,
+      "loss": 0.459,
+      "step": 19230
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0897060880207578e-06,
+      "loss": 0.5046,
+      "step": 19231
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0890528694666858e-06,
+      "loss": 0.4626,
+      "step": 19232
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0883998354804127e-06,
+      "loss": 0.4138,
+      "step": 19233
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0877469860754608e-06,
+      "loss": 0.3911,
+      "step": 19234
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0870943212653574e-06,
+      "loss": 0.4822,
+      "step": 19235
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.086441841063618e-06,
+      "loss": 0.5227,
+      "step": 19236
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0857895454837542e-06,
+      "loss": 0.4292,
+      "step": 19237
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0851374345392818e-06,
+      "loss": 0.4431,
+      "step": 19238
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0844855082437033e-06,
+      "loss": 0.3948,
+      "step": 19239
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0838337666105259e-06,
+      "loss": 0.3704,
+      "step": 19240
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0831822096532452e-06,
+      "loss": 0.4451,
+      "step": 19241
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0825308373853593e-06,
+      "loss": 0.4309,
+      "step": 19242
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0818796498203564e-06,
+      "loss": 0.4355,
+      "step": 19243
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0812286469717281e-06,
+      "loss": 0.4976,
+      "step": 19244
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0805778288529545e-06,
+      "loss": 0.481,
+      "step": 19245
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0799271954775204e-06,
+      "loss": 0.4041,
+      "step": 19246
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.079276746858896e-06,
+      "loss": 0.4377,
+      "step": 19247
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0786264830105597e-06,
+      "loss": 0.4607,
+      "step": 19248
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.077976403945975e-06,
+      "loss": 0.4434,
+      "step": 19249
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0773265096786112e-06,
+      "loss": 0.4858,
+      "step": 19250
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0766768002219253e-06,
+      "loss": 0.4812,
+      "step": 19251
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0760272755893775e-06,
+      "loss": 0.4658,
+      "step": 19252
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0753779357944172e-06,
+      "loss": 0.4478,
+      "step": 19253
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.074728780850499e-06,
+      "loss": 0.4419,
+      "step": 19254
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0740798107710626e-06,
+      "loss": 0.4939,
+      "step": 19255
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0734310255695547e-06,
+      "loss": 0.4141,
+      "step": 19256
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0727824252594121e-06,
+      "loss": 0.4231,
+      "step": 19257
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0721340098540666e-06,
+      "loss": 0.4302,
+      "step": 19258
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.071485779366952e-06,
+      "loss": 0.4365,
+      "step": 19259
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0708377338114916e-06,
+      "loss": 0.3823,
+      "step": 19260
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0701898732011108e-06,
+      "loss": 0.4585,
+      "step": 19261
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0695421975492248e-06,
+      "loss": 0.4546,
+      "step": 19262
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0688947068692535e-06,
+      "loss": 0.4465,
+      "step": 19263
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0682474011746024e-06,
+      "loss": 0.4951,
+      "step": 19264
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0676002804786845e-06,
+      "loss": 0.4431,
+      "step": 19265
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0669533447948998e-06,
+      "loss": 0.4268,
+      "step": 19266
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0663065941366479e-06,
+      "loss": 0.5049,
+      "step": 19267
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0656600285173257e-06,
+      "loss": 0.4143,
+      "step": 19268
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0650136479503238e-06,
+      "loss": 0.5032,
+      "step": 19269
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0643674524490334e-06,
+      "loss": 0.3967,
+      "step": 19270
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0637214420268337e-06,
+      "loss": 0.4272,
+      "step": 19271
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0630756166971112e-06,
+      "loss": 0.4066,
+      "step": 19272
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0624299764732372e-06,
+      "loss": 0.4543,
+      "step": 19273
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0617845213685885e-06,
+      "loss": 0.4636,
+      "step": 19274
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0611392513965313e-06,
+      "loss": 0.4822,
+      "step": 19275
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0604941665704327e-06,
+      "loss": 0.5461,
+      "step": 19276
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.059849266903652e-06,
+      "loss": 0.4077,
+      "step": 19277
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0592045524095495e-06,
+      "loss": 0.4216,
+      "step": 19278
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0585600231014759e-06,
+      "loss": 0.4365,
+      "step": 19279
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0579156789927815e-06,
+      "loss": 0.4761,
+      "step": 19280
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0572715200968153e-06,
+      "loss": 0.4927,
+      "step": 19281
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.056627546426915e-06,
+      "loss": 0.4775,
+      "step": 19282
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.055983757996424e-06,
+      "loss": 0.448,
+      "step": 19283
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0553401548186703e-06,
+      "loss": 0.4128,
+      "step": 19284
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0546967369069915e-06,
+      "loss": 0.5142,
+      "step": 19285
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.054053504274708e-06,
+      "loss": 0.4473,
+      "step": 19286
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0534104569351478e-06,
+      "loss": 0.4558,
+      "step": 19287
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0527675949016258e-06,
+      "loss": 0.4443,
+      "step": 19288
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0521249181874615e-06,
+      "loss": 0.564,
+      "step": 19289
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.051482426805963e-06,
+      "loss": 0.446,
+      "step": 19290
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.05084012077044e-06,
+      "loss": 0.4263,
+      "step": 19291
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0501980000941935e-06,
+      "loss": 0.4202,
+      "step": 19292
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0495560647905267e-06,
+      "loss": 0.4707,
+      "step": 19293
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0489143148727333e-06,
+      "loss": 0.4387,
+      "step": 19294
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0482727503541079e-06,
+      "loss": 0.4641,
+      "step": 19295
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0476313712479368e-06,
+      "loss": 0.4448,
+      "step": 19296
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.046990177567504e-06,
+      "loss": 0.4458,
+      "step": 19297
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.046349169326093e-06,
+      "loss": 0.4692,
+      "step": 19298
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0457083465369778e-06,
+      "loss": 0.4304,
+      "step": 19299
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.045067709213433e-06,
+      "loss": 0.4941,
+      "step": 19300
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0444272573687276e-06,
+      "loss": 0.4524,
+      "step": 19301
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0437869910161269e-06,
+      "loss": 0.4905,
+      "step": 19302
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0431469101688907e-06,
+      "loss": 0.407,
+      "step": 19303
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0425070148402793e-06,
+      "loss": 0.4705,
+      "step": 19304
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0418673050435468e-06,
+      "loss": 0.4976,
+      "step": 19305
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.041227780791939e-06,
+      "loss": 0.5132,
+      "step": 19306
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0405884420987067e-06,
+      "loss": 0.4609,
+      "step": 19307
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0399492889770913e-06,
+      "loss": 0.4583,
+      "step": 19308
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0393103214403278e-06,
+      "loss": 0.4585,
+      "step": 19309
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0386715395016556e-06,
+      "loss": 0.449,
+      "step": 19310
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0380329431743042e-06,
+      "loss": 0.4573,
+      "step": 19311
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0373945324714963e-06,
+      "loss": 0.4333,
+      "step": 19312
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0367563074064602e-06,
+      "loss": 0.411,
+      "step": 19313
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.036118267992412e-06,
+      "loss": 0.4717,
+      "step": 19314
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.035480414242569e-06,
+      "loss": 0.4478,
+      "step": 19315
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0348427461701404e-06,
+      "loss": 0.406,
+      "step": 19316
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0342052637883381e-06,
+      "loss": 0.4375,
+      "step": 19317
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0335679671103604e-06,
+      "loss": 0.4194,
+      "step": 19318
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0329308561494122e-06,
+      "loss": 0.4919,
+      "step": 19319
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0322939309186863e-06,
+      "loss": 0.4568,
+      "step": 19320
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0316571914313766e-06,
+      "loss": 0.4597,
+      "step": 19321
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0310206377006705e-06,
+      "loss": 0.4832,
+      "step": 19322
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0303842697397549e-06,
+      "loss": 0.429,
+      "step": 19323
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.029748087561807e-06,
+      "loss": 0.5132,
+      "step": 19324
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0291120911800078e-06,
+      "loss": 0.4077,
+      "step": 19325
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0284762806075288e-06,
+      "loss": 0.4553,
+      "step": 19326
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.027840655857536e-06,
+      "loss": 0.4768,
+      "step": 19327
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0272052169431968e-06,
+      "loss": 0.436,
+      "step": 19328
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.026569963877676e-06,
+      "loss": 0.3896,
+      "step": 19329
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0259348966741268e-06,
+      "loss": 0.5286,
+      "step": 19330
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.025300015345707e-06,
+      "loss": 0.4121,
+      "step": 19331
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0246653199055612e-06,
+      "loss": 0.45,
+      "step": 19332
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0240308103668394e-06,
+      "loss": 0.4379,
+      "step": 19333
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.023396486742686e-06,
+      "loss": 0.4775,
+      "step": 19334
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0227623490462335e-06,
+      "loss": 0.4465,
+      "step": 19335
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0221283972906216e-06,
+      "loss": 0.4773,
+      "step": 19336
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0214946314889773e-06,
+      "loss": 0.5027,
+      "step": 19337
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0208610516544315e-06,
+      "loss": 0.405,
+      "step": 19338
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0202276578001014e-06,
+      "loss": 0.4539,
+      "step": 19339
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.019594449939113e-06,
+      "loss": 0.509,
+      "step": 19340
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0189614280845773e-06,
+      "loss": 0.3845,
+      "step": 19341
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0183285922496045e-06,
+      "loss": 0.3894,
+      "step": 19342
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0176959424473076e-06,
+      "loss": 0.4592,
+      "step": 19343
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.017063478690784e-06,
+      "loss": 0.4507,
+      "step": 19344
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0164312009931387e-06,
+      "loss": 0.4214,
+      "step": 19345
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0157991093674636e-06,
+      "loss": 0.4934,
+      "step": 19346
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0151672038268555e-06,
+      "loss": 0.4915,
+      "step": 19347
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.014535484384398e-06,
+      "loss": 0.4194,
+      "step": 19348
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.01390395105318e-06,
+      "loss": 0.4062,
+      "step": 19349
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0132726038462771e-06,
+      "loss": 0.4536,
+      "step": 19350
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0126414427767716e-06,
+      "loss": 0.4717,
+      "step": 19351
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0120104678577314e-06,
+      "loss": 0.4414,
+      "step": 19352
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0113796791022301e-06,
+      "loss": 0.4136,
+      "step": 19353
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0107490765233286e-06,
+      "loss": 0.4552,
+      "step": 19354
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0101186601340917e-06,
+      "loss": 0.4163,
+      "step": 19355
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.009488429947575e-06,
+      "loss": 0.4771,
+      "step": 19356
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.008858385976831e-06,
+      "loss": 0.4377,
+      "step": 19357
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0082285282349137e-06,
+      "loss": 0.376,
+      "step": 19358
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0075988567348637e-06,
+      "loss": 0.386,
+      "step": 19359
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0069693714897288e-06,
+      "loss": 0.4331,
+      "step": 19360
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0063400725125416e-06,
+      "loss": 0.425,
+      "step": 19361
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0057109598163405e-06,
+      "loss": 0.4082,
+      "step": 19362
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.005082033414153e-06,
+      "loss": 0.4507,
+      "step": 19363
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.00445329331901e-06,
+      "loss": 0.434,
+      "step": 19364
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0038247395439282e-06,
+      "loss": 0.4543,
+      "step": 19365
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0031963721019321e-06,
+      "loss": 0.4716,
+      "step": 19366
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0025681910060326e-06,
+      "loss": 0.4619,
+      "step": 19367
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0019401962692443e-06,
+      "loss": 0.3752,
+      "step": 19368
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0013123879045717e-06,
+      "loss": 0.4639,
+      "step": 19369
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0006847659250207e-06,
+      "loss": 0.4257,
+      "step": 19370
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.000057330343589e-06,
+      "loss": 0.4399,
+      "step": 19371
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.994300811732727e-07,
+      "loss": 0.4312,
+      "step": 19372
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.988030184270647e-07,
+      "loss": 0.4504,
+      "step": 19373
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.981761421179513e-07,
+      "loss": 0.4539,
+      "step": 19374
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.975494522589191e-07,
+      "loss": 0.4341,
+      "step": 19375
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.96922948862945e-07,
+      "loss": 0.4653,
+      "step": 19376
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.9629663194301e-07,
+      "loss": 0.4338,
+      "step": 19377
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.956705015120815e-07,
+      "loss": 0.4397,
+      "step": 19378
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.950445575831336e-07,
+      "loss": 0.501,
+      "step": 19379
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.944188001691258e-07,
+      "loss": 0.5176,
+      "step": 19380
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.93793229283022e-07,
+      "loss": 0.4558,
+      "step": 19381
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.931678449377801e-07,
+      "loss": 0.3984,
+      "step": 19382
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.925426471463517e-07,
+      "loss": 0.4524,
+      "step": 19383
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.919176359216852e-07,
+      "loss": 0.4224,
+      "step": 19384
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.912928112767305e-07,
+      "loss": 0.4233,
+      "step": 19385
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.906681732244273e-07,
+      "loss": 0.4402,
+      "step": 19386
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.900437217777092e-07,
+      "loss": 0.4414,
+      "step": 19387
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.894194569495164e-07,
+      "loss": 0.4927,
+      "step": 19388
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.887953787527738e-07,
+      "loss": 0.3945,
+      "step": 19389
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.88171487200411e-07,
+      "loss": 0.4341,
+      "step": 19390
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.875477823053482e-07,
+      "loss": 0.417,
+      "step": 19391
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.869242640805066e-07,
+      "loss": 0.4917,
+      "step": 19392
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.863009325387962e-07,
+      "loss": 0.4084,
+      "step": 19393
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.856777876931333e-07,
+      "loss": 0.3945,
+      "step": 19394
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.85054829556421e-07,
+      "loss": 0.4646,
+      "step": 19395
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.844320581415645e-07,
+      "loss": 0.4048,
+      "step": 19396
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.838094734614601e-07,
+      "loss": 0.4404,
+      "step": 19397
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.831870755290064e-07,
+      "loss": 0.4365,
+      "step": 19398
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.825648643570917e-07,
+      "loss": 0.4746,
+      "step": 19399
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.819428399586062e-07,
+      "loss": 0.4473,
+      "step": 19400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.813210023464337e-07,
+      "loss": 0.3933,
+      "step": 19401
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.806993515334496e-07,
+      "loss": 0.4202,
+      "step": 19402
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.800778875325356e-07,
+      "loss": 0.4751,
+      "step": 19403
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.794566103565594e-07,
+      "loss": 0.5059,
+      "step": 19404
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.788355200183907e-07,
+      "loss": 0.5022,
+      "step": 19405
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.78214616530896e-07,
+      "loss": 0.4448,
+      "step": 19406
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.775938999069313e-07,
+      "loss": 0.4119,
+      "step": 19407
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.769733701593575e-07,
+      "loss": 0.502,
+      "step": 19408
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.763530273010235e-07,
+      "loss": 0.4961,
+      "step": 19409
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.757328713447811e-07,
+      "loss": 0.4368,
+      "step": 19410
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.751129023034756e-07,
+      "loss": 0.4392,
+      "step": 19411
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.744931201899444e-07,
+      "loss": 0.4421,
+      "step": 19412
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.738735250170295e-07,
+      "loss": 0.4473,
+      "step": 19413
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.732541167975595e-07,
+      "loss": 0.3914,
+      "step": 19414
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.726348955443687e-07,
+      "loss": 0.4277,
+      "step": 19415
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.720158612702802e-07,
+      "loss": 0.4287,
+      "step": 19416
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.713970139881135e-07,
+      "loss": 0.4609,
+      "step": 19417
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.707783537106907e-07,
+      "loss": 0.4725,
+      "step": 19418
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.701598804508216e-07,
+      "loss": 0.4666,
+      "step": 19419
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.695415942213216e-07,
+      "loss": 0.5293,
+      "step": 19420
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.689234950349902e-07,
+      "loss": 0.4404,
+      "step": 19421
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.68305582904636e-07,
+      "loss": 0.4583,
+      "step": 19422
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.676878578430538e-07,
+      "loss": 0.45,
+      "step": 19423
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.670703198630404e-07,
+      "loss": 0.4363,
+      "step": 19424
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.664529689773837e-07,
+      "loss": 0.4551,
+      "step": 19425
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.658358051988748e-07,
+      "loss": 0.425,
+      "step": 19426
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.652188285402918e-07,
+      "loss": 0.4634,
+      "step": 19427
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.646020390144183e-07,
+      "loss": 0.4457,
+      "step": 19428
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.63985436634025e-07,
+      "loss": 0.4192,
+      "step": 19429
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.633690214118885e-07,
+      "loss": 0.427,
+      "step": 19430
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.627527933607717e-07,
+      "loss": 0.4456,
+      "step": 19431
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.621367524934411e-07,
+      "loss": 0.4258,
+      "step": 19432
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.615208988226532e-07,
+      "loss": 0.4348,
+      "step": 19433
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.609052323611667e-07,
+      "loss": 0.4724,
+      "step": 19434
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.60289753121736e-07,
+      "loss": 0.4893,
+      "step": 19435
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.596744611171026e-07,
+      "loss": 0.394,
+      "step": 19436
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.59059356360016e-07,
+      "loss": 0.4653,
+      "step": 19437
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.58444438863214e-07,
+      "loss": 0.5151,
+      "step": 19438
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.57829708639435e-07,
+      "loss": 0.4387,
+      "step": 19439
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.572151657014085e-07,
+      "loss": 0.4561,
+      "step": 19440
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.566008100618673e-07,
+      "loss": 0.4714,
+      "step": 19441
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.559866417335318e-07,
+      "loss": 0.561,
+      "step": 19442
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.55372660729128e-07,
+      "loss": 0.4854,
+      "step": 19443
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.547588670613694e-07,
+      "loss": 0.4172,
+      "step": 19444
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.54145260742968e-07,
+      "loss": 0.4197,
+      "step": 19445
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.53531841786638e-07,
+      "loss": 0.4644,
+      "step": 19446
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.529186102050791e-07,
+      "loss": 0.5063,
+      "step": 19447
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.52305566010997e-07,
+      "loss": 0.4502,
+      "step": 19448
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.516927092170869e-07,
+      "loss": 0.3938,
+      "step": 19449
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.510800398360453e-07,
+      "loss": 0.4949,
+      "step": 19450
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.504675578805589e-07,
+      "loss": 0.4641,
+      "step": 19451
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.498552633633173e-07,
+      "loss": 0.4385,
+      "step": 19452
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.492431562969983e-07,
+      "loss": 0.5063,
+      "step": 19453
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.486312366942851e-07,
+      "loss": 0.446,
+      "step": 19454
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.480195045678464e-07,
+      "loss": 0.4698,
+      "step": 19455
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.474079599303587e-07,
+      "loss": 0.4524,
+      "step": 19456
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.467966027944831e-07,
+      "loss": 0.4253,
+      "step": 19457
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.461854331728859e-07,
+      "loss": 0.421,
+      "step": 19458
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.455744510782261e-07,
+      "loss": 0.4153,
+      "step": 19459
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.449636565231556e-07,
+      "loss": 0.4446,
+      "step": 19460
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.44353049520329e-07,
+      "loss": 0.4624,
+      "step": 19461
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.437426300823904e-07,
+      "loss": 0.396,
+      "step": 19462
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.431323982219864e-07,
+      "loss": 0.4827,
+      "step": 19463
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.425223539517525e-07,
+      "loss": 0.4329,
+      "step": 19464
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.419124972843286e-07,
+      "loss": 0.4119,
+      "step": 19465
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.413028282323411e-07,
+      "loss": 0.4565,
+      "step": 19466
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.406933468084234e-07,
+      "loss": 0.4341,
+      "step": 19467
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.400840530251942e-07,
+      "loss": 0.4839,
+      "step": 19468
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.39474946895278e-07,
+      "loss": 0.408,
+      "step": 19469
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.388660284312878e-07,
+      "loss": 0.4197,
+      "step": 19470
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.382572976458381e-07,
+      "loss": 0.4526,
+      "step": 19471
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.376487545515334e-07,
+      "loss": 0.4028,
+      "step": 19472
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.370403991609834e-07,
+      "loss": 0.4219,
+      "step": 19473
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.364322314867858e-07,
+      "loss": 0.4343,
+      "step": 19474
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.35824251541535e-07,
+      "loss": 0.4883,
+      "step": 19475
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.352164593378288e-07,
+      "loss": 0.4775,
+      "step": 19476
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.346088548882515e-07,
+      "loss": 0.4365,
+      "step": 19477
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.340014382053919e-07,
+      "loss": 0.4644,
+      "step": 19478
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.333942093018266e-07,
+      "loss": 0.4453,
+      "step": 19479
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.32787168190138e-07,
+      "loss": 0.4048,
+      "step": 19480
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.321803148828945e-07,
+      "loss": 0.425,
+      "step": 19481
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.315736493926686e-07,
+      "loss": 0.353,
+      "step": 19482
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.309671717320268e-07,
+      "loss": 0.4805,
+      "step": 19483
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.303608819135279e-07,
+      "loss": 0.4185,
+      "step": 19484
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.297547799497319e-07,
+      "loss": 0.49,
+      "step": 19485
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.2914886585319e-07,
+      "loss": 0.4482,
+      "step": 19486
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.285431396364542e-07,
+      "loss": 0.4272,
+      "step": 19487
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.279376013120722e-07,
+      "loss": 0.417,
+      "step": 19488
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.273322508925831e-07,
+      "loss": 0.4675,
+      "step": 19489
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.267270883905255e-07,
+      "loss": 0.4392,
+      "step": 19490
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.261221138184362e-07,
+      "loss": 0.4189,
+      "step": 19491
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.255173271888429e-07,
+      "loss": 0.4231,
+      "step": 19492
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.249127285142745e-07,
+      "loss": 0.5227,
+      "step": 19493
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.24308317807252e-07,
+      "loss": 0.4297,
+      "step": 19494
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.237040950802955e-07,
+      "loss": 0.5142,
+      "step": 19495
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.231000603459184e-07,
+      "loss": 0.4222,
+      "step": 19496
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.224962136166349e-07,
+      "loss": 0.5017,
+      "step": 19497
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.218925549049473e-07,
+      "loss": 0.4517,
+      "step": 19498
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.212890842233646e-07,
+      "loss": 0.46,
+      "step": 19499
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.206858015843801e-07,
+      "loss": 0.4929,
+      "step": 19500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.200827070004959e-07,
+      "loss": 0.4597,
+      "step": 19501
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.194798004841965e-07,
+      "loss": 0.4358,
+      "step": 19502
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.188770820479765e-07,
+      "loss": 0.3628,
+      "step": 19503
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.182745517043157e-07,
+      "loss": 0.4294,
+      "step": 19504
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.17672209465692e-07,
+      "loss": 0.4695,
+      "step": 19505
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.170700553445866e-07,
+      "loss": 0.3822,
+      "step": 19506
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.164680893534672e-07,
+      "loss": 0.4207,
+      "step": 19507
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.158663115048039e-07,
+      "loss": 0.4951,
+      "step": 19508
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.152647218110632e-07,
+      "loss": 0.4231,
+      "step": 19509
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.14663320284701e-07,
+      "loss": 0.4548,
+      "step": 19510
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.140621069381761e-07,
+      "loss": 0.3948,
+      "step": 19511
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.134610817839428e-07,
+      "loss": 0.4636,
+      "step": 19512
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.128602448344471e-07,
+      "loss": 0.4309,
+      "step": 19513
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.122595961021363e-07,
+      "loss": 0.4507,
+      "step": 19514
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.116591355994475e-07,
+      "loss": 0.4429,
+      "step": 19515
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.110588633388218e-07,
+      "loss": 0.459,
+      "step": 19516
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.104587793326903e-07,
+      "loss": 0.4937,
+      "step": 19517
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.098588835934829e-07,
+      "loss": 0.4197,
+      "step": 19518
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.092591761336255e-07,
+      "loss": 0.387,
+      "step": 19519
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.086596569655359e-07,
+      "loss": 0.489,
+      "step": 19520
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.080603261016363e-07,
+      "loss": 0.4272,
+      "step": 19521
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.074611835543379e-07,
+      "loss": 0.4763,
+      "step": 19522
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.068622293360518e-07,
+      "loss": 0.5334,
+      "step": 19523
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.062634634591805e-07,
+      "loss": 0.4739,
+      "step": 19524
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.056648859361306e-07,
+      "loss": 0.4424,
+      "step": 19525
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.050664967792966e-07,
+      "loss": 0.4382,
+      "step": 19526
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.044682960010754e-07,
+      "loss": 0.4739,
+      "step": 19527
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.038702836138535e-07,
+      "loss": 0.4275,
+      "step": 19528
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.032724596300213e-07,
+      "loss": 0.4075,
+      "step": 19529
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.026748240619576e-07,
+      "loss": 0.425,
+      "step": 19530
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.020773769220448e-07,
+      "loss": 0.4341,
+      "step": 19531
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.01480118222654e-07,
+      "loss": 0.4019,
+      "step": 19532
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.008830479761588e-07,
+      "loss": 0.5378,
+      "step": 19533
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.002861661949213e-07,
+      "loss": 0.4839,
+      "step": 19534
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.996894728913097e-07,
+      "loss": 0.4717,
+      "step": 19535
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.990929680776816e-07,
+      "loss": 0.4661,
+      "step": 19536
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.984966517663896e-07,
+      "loss": 0.4993,
+      "step": 19537
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.97900523969788e-07,
+      "loss": 0.4463,
+      "step": 19538
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.973045847002216e-07,
+      "loss": 0.5234,
+      "step": 19539
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.967088339700369e-07,
+      "loss": 0.4304,
+      "step": 19540
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.961132717915689e-07,
+      "loss": 0.4915,
+      "step": 19541
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.955178981771584e-07,
+      "loss": 0.4287,
+      "step": 19542
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.949227131391325e-07,
+      "loss": 0.5022,
+      "step": 19543
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.943277166898212e-07,
+      "loss": 0.4885,
+      "step": 19544
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.93732908841548e-07,
+      "loss": 0.4514,
+      "step": 19545
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.931382896066332e-07,
+      "loss": 0.4016,
+      "step": 19546
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.925438589973911e-07,
+      "loss": 0.4673,
+      "step": 19547
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.919496170261366e-07,
+      "loss": 0.3928,
+      "step": 19548
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.913555637051774e-07,
+      "loss": 0.4153,
+      "step": 19549
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.907616990468149e-07,
+      "loss": 0.5159,
+      "step": 19550
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.901680230633525e-07,
+      "loss": 0.4099,
+      "step": 19551
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.895745357670849e-07,
+      "loss": 0.5049,
+      "step": 19552
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.889812371703066e-07,
+      "loss": 0.4382,
+      "step": 19553
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.883881272853034e-07,
+      "loss": 0.4849,
+      "step": 19554
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.877952061243634e-07,
+      "loss": 0.4436,
+      "step": 19555
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.872024736997653e-07,
+      "loss": 0.45,
+      "step": 19556
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.866099300237874e-07,
+      "loss": 0.4458,
+      "step": 19557
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.860175751087008e-07,
+      "loss": 0.4495,
+      "step": 19558
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.854254089667757e-07,
+      "loss": 0.4243,
+      "step": 19559
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.848334316102791e-07,
+      "loss": 0.4321,
+      "step": 19560
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.842416430514689e-07,
+      "loss": 0.3916,
+      "step": 19561
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.836500433026063e-07,
+      "loss": 0.4429,
+      "step": 19562
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.830586323759416e-07,
+      "loss": 0.4954,
+      "step": 19563
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.82467410283726e-07,
+      "loss": 0.4812,
+      "step": 19564
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.818763770382044e-07,
+      "loss": 0.4263,
+      "step": 19565
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.812855326516201e-07,
+      "loss": 0.4185,
+      "step": 19566
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.806948771362078e-07,
+      "loss": 0.4268,
+      "step": 19567
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.801044105042056e-07,
+      "loss": 0.4187,
+      "step": 19568
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.795141327678391e-07,
+      "loss": 0.4314,
+      "step": 19569
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.789240439393387e-07,
+      "loss": 0.3931,
+      "step": 19570
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.783341440309235e-07,
+      "loss": 0.4221,
+      "step": 19571
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.777444330548135e-07,
+      "loss": 0.5391,
+      "step": 19572
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.771549110232214e-07,
+      "loss": 0.4602,
+      "step": 19573
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.765655779483606e-07,
+      "loss": 0.3909,
+      "step": 19574
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.759764338424337e-07,
+      "loss": 0.4238,
+      "step": 19575
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.753874787176475e-07,
+      "loss": 0.395,
+      "step": 19576
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.747987125861967e-07,
+      "loss": 0.4207,
+      "step": 19577
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.742101354602795e-07,
+      "loss": 0.5066,
+      "step": 19578
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.736217473520847e-07,
+      "loss": 0.4299,
+      "step": 19579
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.730335482737995e-07,
+      "loss": 0.4158,
+      "step": 19580
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.724455382376084e-07,
+      "loss": 0.4736,
+      "step": 19581
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.718577172556875e-07,
+      "loss": 0.4243,
+      "step": 19582
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.712700853402167e-07,
+      "loss": 0.4404,
+      "step": 19583
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.70682642503362e-07,
+      "loss": 0.4788,
+      "step": 19584
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.700953887572939e-07,
+      "loss": 0.5264,
+      "step": 19585
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.695083241141767e-07,
+      "loss": 0.5046,
+      "step": 19586
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.689214485861663e-07,
+      "loss": 0.5076,
+      "step": 19587
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.683347621854221e-07,
+      "loss": 0.4004,
+      "step": 19588
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.677482649240954e-07,
+      "loss": 0.4546,
+      "step": 19589
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.671619568143319e-07,
+      "loss": 0.3945,
+      "step": 19590
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.665758378682776e-07,
+      "loss": 0.4768,
+      "step": 19591
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.659899080980705e-07,
+      "loss": 0.4563,
+      "step": 19592
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.654041675158487e-07,
+      "loss": 0.4126,
+      "step": 19593
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.648186161337446e-07,
+      "loss": 0.4469,
+      "step": 19594
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.64233253963882e-07,
+      "loss": 0.4707,
+      "step": 19595
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.636480810183911e-07,
+      "loss": 0.4224,
+      "step": 19596
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.630630973093867e-07,
+      "loss": 0.4143,
+      "step": 19597
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.624783028489914e-07,
+      "loss": 0.4561,
+      "step": 19598
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.61893697649312e-07,
+      "loss": 0.4126,
+      "step": 19599
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.613092817224611e-07,
+      "loss": 0.4006,
+      "step": 19600
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.607250550805401e-07,
+      "loss": 0.3967,
+      "step": 19601
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.601410177356528e-07,
+      "loss": 0.4724,
+      "step": 19602
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.595571696998939e-07,
+      "loss": 0.4395,
+      "step": 19603
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.589735109853581e-07,
+      "loss": 0.448,
+      "step": 19604
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.583900416041324e-07,
+      "loss": 0.4084,
+      "step": 19605
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.57806761568305e-07,
+      "loss": 0.4382,
+      "step": 19606
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.572236708899529e-07,
+      "loss": 0.4509,
+      "step": 19607
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.566407695811574e-07,
+      "loss": 0.4158,
+      "step": 19608
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.56058057653989e-07,
+      "loss": 0.4363,
+      "step": 19609
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.554755351205191e-07,
+      "loss": 0.4333,
+      "step": 19610
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.548932019928102e-07,
+      "loss": 0.4214,
+      "step": 19611
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.543110582829272e-07,
+      "loss": 0.4412,
+      "step": 19612
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.537291040029272e-07,
+      "loss": 0.4536,
+      "step": 19613
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.531473391648614e-07,
+      "loss": 0.4658,
+      "step": 19614
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.525657637807827e-07,
+      "loss": 0.4524,
+      "step": 19615
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.519843778627345e-07,
+      "loss": 0.4888,
+      "step": 19616
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.514031814227619e-07,
+      "loss": 0.438,
+      "step": 19617
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.508221744728984e-07,
+      "loss": 0.4395,
+      "step": 19618
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.502413570251822e-07,
+      "loss": 0.4587,
+      "step": 19619
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.496607290916403e-07,
+      "loss": 0.4583,
+      "step": 19620
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.490802906843021e-07,
+      "loss": 0.4253,
+      "step": 19621
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.485000418151868e-07,
+      "loss": 0.5251,
+      "step": 19622
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.479199824963158e-07,
+      "loss": 0.4973,
+      "step": 19623
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.473401127397019e-07,
+      "loss": 0.3936,
+      "step": 19624
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.467604325573542e-07,
+      "loss": 0.4385,
+      "step": 19625
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.461809419612821e-07,
+      "loss": 0.4167,
+      "step": 19626
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.456016409634859e-07,
+      "loss": 0.4451,
+      "step": 19627
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.450225295759673e-07,
+      "loss": 0.4595,
+      "step": 19628
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.444436078107177e-07,
+      "loss": 0.4584,
+      "step": 19629
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.438648756797307e-07,
+      "loss": 0.4807,
+      "step": 19630
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.432863331949914e-07,
+      "loss": 0.3931,
+      "step": 19631
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.427079803684846e-07,
+      "loss": 0.5144,
+      "step": 19632
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.421298172121873e-07,
+      "loss": 0.4438,
+      "step": 19633
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.415518437380776e-07,
+      "loss": 0.4739,
+      "step": 19634
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.40974059958124e-07,
+      "loss": 0.52,
+      "step": 19635
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.403964658842933e-07,
+      "loss": 0.4663,
+      "step": 19636
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.398190615285529e-07,
+      "loss": 0.4702,
+      "step": 19637
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.392418469028585e-07,
+      "loss": 0.4624,
+      "step": 19638
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.386648220191685e-07,
+      "loss": 0.3982,
+      "step": 19639
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.38087986889432e-07,
+      "loss": 0.4094,
+      "step": 19640
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.375113415255986e-07,
+      "loss": 0.46,
+      "step": 19641
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.369348859396098e-07,
+      "loss": 0.4758,
+      "step": 19642
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.363586201434082e-07,
+      "loss": 0.4421,
+      "step": 19643
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.357825441489265e-07,
+      "loss": 0.5193,
+      "step": 19644
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.352066579681006e-07,
+      "loss": 0.4382,
+      "step": 19645
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.346309616128545e-07,
+      "loss": 0.4692,
+      "step": 19646
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.340554550951164e-07,
+      "loss": 0.4734,
+      "step": 19647
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.334801384268009e-07,
+      "loss": 0.4326,
+      "step": 19648
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.329050116198311e-07,
+      "loss": 0.4795,
+      "step": 19649
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.323300746861129e-07,
+      "loss": 0.4458,
+      "step": 19650
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.317553276375601e-07,
+      "loss": 0.49,
+      "step": 19651
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.311807704860742e-07,
+      "loss": 0.4924,
+      "step": 19652
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.306064032435546e-07,
+      "loss": 0.4453,
+      "step": 19653
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.300322259219007e-07,
+      "loss": 0.4968,
+      "step": 19654
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.294582385330019e-07,
+      "loss": 0.5081,
+      "step": 19655
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.288844410887509e-07,
+      "loss": 0.4045,
+      "step": 19656
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.283108336010293e-07,
+      "loss": 0.4392,
+      "step": 19657
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.277374160817197e-07,
+      "loss": 0.4495,
+      "step": 19658
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.271641885426973e-07,
+      "loss": 0.4141,
+      "step": 19659
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.265911509958369e-07,
+      "loss": 0.4009,
+      "step": 19660
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.260183034530056e-07,
+      "loss": 0.4377,
+      "step": 19661
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.254456459260695e-07,
+      "loss": 0.4458,
+      "step": 19662
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.248731784268915e-07,
+      "loss": 0.4463,
+      "step": 19663
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.243009009673253e-07,
+      "loss": 0.4624,
+      "step": 19664
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.237288135592258e-07,
+      "loss": 0.4482,
+      "step": 19665
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.231569162144448e-07,
+      "loss": 0.4204,
+      "step": 19666
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.22585208944825e-07,
+      "loss": 0.4438,
+      "step": 19667
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.220136917622068e-07,
+      "loss": 0.48,
+      "step": 19668
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.21442364678432e-07,
+      "loss": 0.4939,
+      "step": 19669
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.208712277053288e-07,
+      "loss": 0.3958,
+      "step": 19670
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.203002808547312e-07,
+      "loss": 0.5027,
+      "step": 19671
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.197295241384617e-07,
+      "loss": 0.4392,
+      "step": 19672
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.191589575683457e-07,
+      "loss": 0.4739,
+      "step": 19673
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.185885811561978e-07,
+      "loss": 0.4602,
+      "step": 19674
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.180183949138332e-07,
+      "loss": 0.4504,
+      "step": 19675
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.174483988530613e-07,
+      "loss": 0.4902,
+      "step": 19676
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.168785929856904e-07,
+      "loss": 0.5137,
+      "step": 19677
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.163089773235189e-07,
+      "loss": 0.4104,
+      "step": 19678
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.157395518783484e-07,
+      "loss": 0.4475,
+      "step": 19679
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.151703166619707e-07,
+      "loss": 0.4565,
+      "step": 19680
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.146012716861773e-07,
+      "loss": 0.4426,
+      "step": 19681
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.140324169627556e-07,
+      "loss": 0.4211,
+      "step": 19682
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.134637525034838e-07,
+      "loss": 0.366,
+      "step": 19683
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.128952783201449e-07,
+      "loss": 0.4614,
+      "step": 19684
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.123269944245105e-07,
+      "loss": 0.4412,
+      "step": 19685
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.117589008283522e-07,
+      "loss": 0.4392,
+      "step": 19686
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.111909975434385e-07,
+      "loss": 0.499,
+      "step": 19687
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.106232845815276e-07,
+      "loss": 0.4482,
+      "step": 19688
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.100557619543825e-07,
+      "loss": 0.4819,
+      "step": 19689
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.094884296737571e-07,
+      "loss": 0.4717,
+      "step": 19690
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.089212877514007e-07,
+      "loss": 0.4163,
+      "step": 19691
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.08354336199062e-07,
+      "loss": 0.4983,
+      "step": 19692
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.077875750284813e-07,
+      "loss": 0.4512,
+      "step": 19693
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.072210042514017e-07,
+      "loss": 0.4282,
+      "step": 19694
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.066546238795548e-07,
+      "loss": 0.4238,
+      "step": 19695
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.060884339246744e-07,
+      "loss": 0.387,
+      "step": 19696
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.055224343984857e-07,
+      "loss": 0.4458,
+      "step": 19697
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.049566253127117e-07,
+      "loss": 0.4648,
+      "step": 19698
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.043910066790727e-07,
+      "loss": 0.4561,
+      "step": 19699
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.038255785092841e-07,
+      "loss": 0.4006,
+      "step": 19700
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.032603408150575e-07,
+      "loss": 0.417,
+      "step": 19701
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.02695293608099e-07,
+      "loss": 0.443,
+      "step": 19702
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.021304369001137e-07,
+      "loss": 0.4473,
+      "step": 19703
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.015657707028002e-07,
+      "loss": 0.418,
+      "step": 19704
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.010012950278556e-07,
+      "loss": 0.4521,
+      "step": 19705
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.004370098869685e-07,
+      "loss": 0.4829,
+      "step": 19706
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.998729152918306e-07,
+      "loss": 0.4324,
+      "step": 19707
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.993090112541224e-07,
+      "loss": 0.4438,
+      "step": 19708
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.98745297785527e-07,
+      "loss": 0.4526,
+      "step": 19709
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.981817748977161e-07,
+      "loss": 0.4473,
+      "step": 19710
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.976184426023659e-07,
+      "loss": 0.4956,
+      "step": 19711
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.970553009111403e-07,
+      "loss": 0.3931,
+      "step": 19712
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.964923498357058e-07,
+      "loss": 0.4426,
+      "step": 19713
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.959295893877239e-07,
+      "loss": 0.4485,
+      "step": 19714
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.953670195788476e-07,
+      "loss": 0.4119,
+      "step": 19715
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.948046404207322e-07,
+      "loss": 0.4453,
+      "step": 19716
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.942424519250236e-07,
+      "loss": 0.4565,
+      "step": 19717
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.936804541033671e-07,
+      "loss": 0.438,
+      "step": 19718
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.931186469674023e-07,
+      "loss": 0.45,
+      "step": 19719
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.925570305287678e-07,
+      "loss": 0.4822,
+      "step": 19720
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.919956047990918e-07,
+      "loss": 0.4568,
+      "step": 19721
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.914343697900085e-07,
+      "loss": 0.4512,
+      "step": 19722
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.908733255131362e-07,
+      "loss": 0.4385,
+      "step": 19723
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.903124719801014e-07,
+      "loss": 0.4946,
+      "step": 19724
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.897518092025169e-07,
+      "loss": 0.4634,
+      "step": 19725
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.891913371919979e-07,
+      "loss": 0.4719,
+      "step": 19726
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.886310559601518e-07,
+      "loss": 0.4626,
+      "step": 19727
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.880709655185815e-07,
+      "loss": 0.4641,
+      "step": 19728
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.875110658788921e-07,
+      "loss": 0.4009,
+      "step": 19729
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.869513570526765e-07,
+      "loss": 0.3938,
+      "step": 19730
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.863918390515301e-07,
+      "loss": 0.4446,
+      "step": 19731
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.858325118870413e-07,
+      "loss": 0.4009,
+      "step": 19732
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.852733755707953e-07,
+      "loss": 0.4443,
+      "step": 19733
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.847144301143728e-07,
+      "loss": 0.4192,
+      "step": 19734
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.841556755293511e-07,
+      "loss": 0.3911,
+      "step": 19735
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.835971118273033e-07,
+      "loss": 0.4966,
+      "step": 19736
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.830387390198002e-07,
+      "loss": 0.5193,
+      "step": 19737
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.824805571184036e-07,
+      "loss": 0.3889,
+      "step": 19738
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.819225661346763e-07,
+      "loss": 0.4148,
+      "step": 19739
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.813647660801782e-07,
+      "loss": 0.4697,
+      "step": 19740
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.8080715696646e-07,
+      "loss": 0.4607,
+      "step": 19741
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.802497388050733e-07,
+      "loss": 0.521,
+      "step": 19742
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.796925116075604e-07,
+      "loss": 0.3977,
+      "step": 19743
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.791354753854663e-07,
+      "loss": 0.4241,
+      "step": 19744
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.785786301503251e-07,
+      "loss": 0.4578,
+      "step": 19745
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.780219759136742e-07,
+      "loss": 0.4155,
+      "step": 19746
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.774655126870401e-07,
+      "loss": 0.4038,
+      "step": 19747
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.769092404819523e-07,
+      "loss": 0.4314,
+      "step": 19748
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.763531593099271e-07,
+      "loss": 0.4922,
+      "step": 19749
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.757972691824877e-07,
+      "loss": 0.4617,
+      "step": 19750
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.752415701111438e-07,
+      "loss": 0.4958,
+      "step": 19751
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.746860621074093e-07,
+      "loss": 0.4001,
+      "step": 19752
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.741307451827851e-07,
+      "loss": 0.417,
+      "step": 19753
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.735756193487787e-07,
+      "loss": 0.4631,
+      "step": 19754
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.730206846168842e-07,
+      "loss": 0.4238,
+      "step": 19755
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.72465940998598e-07,
+      "loss": 0.4141,
+      "step": 19756
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.719113885054097e-07,
+      "loss": 0.4431,
+      "step": 19757
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.713570271488024e-07,
+      "loss": 0.5105,
+      "step": 19758
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.708028569402637e-07,
+      "loss": 0.3857,
+      "step": 19759
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.702488778912665e-07,
+      "loss": 0.447,
+      "step": 19760
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.696950900132893e-07,
+      "loss": 0.431,
+      "step": 19761
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.691414933177998e-07,
+      "loss": 0.481,
+      "step": 19762
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.685880878162644e-07,
+      "loss": 0.4644,
+      "step": 19763
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.680348735201482e-07,
+      "loss": 0.4297,
+      "step": 19764
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.674818504409065e-07,
+      "loss": 0.4758,
+      "step": 19765
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.669290185899947e-07,
+      "loss": 0.4172,
+      "step": 19766
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.663763779788657e-07,
+      "loss": 0.4214,
+      "step": 19767
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.658239286189617e-07,
+      "loss": 0.4797,
+      "step": 19768
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.6527167052173e-07,
+      "loss": 0.4817,
+      "step": 19769
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.647196036986048e-07,
+      "loss": 0.4377,
+      "step": 19770
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.641677281610249e-07,
+      "loss": 0.4075,
+      "step": 19771
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.636160439204188e-07,
+      "loss": 0.4441,
+      "step": 19772
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.630645509882117e-07,
+      "loss": 0.4534,
+      "step": 19773
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.625132493758303e-07,
+      "loss": 0.4243,
+      "step": 19774
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.619621390946897e-07,
+      "loss": 0.407,
+      "step": 19775
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.614112201562063e-07,
+      "loss": 0.4712,
+      "step": 19776
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.608604925717911e-07,
+      "loss": 0.4429,
+      "step": 19777
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.603099563528515e-07,
+      "loss": 0.4905,
+      "step": 19778
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.597596115107885e-07,
+      "loss": 0.4917,
+      "step": 19779
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.592094580570042e-07,
+      "loss": 0.4902,
+      "step": 19780
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.586594960028904e-07,
+      "loss": 0.4158,
+      "step": 19781
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.581097253598413e-07,
+      "loss": 0.4194,
+      "step": 19782
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.575601461392402e-07,
+      "loss": 0.4236,
+      "step": 19783
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.570107583524744e-07,
+      "loss": 0.4282,
+      "step": 19784
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.564615620109195e-07,
+      "loss": 0.3933,
+      "step": 19785
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.55912557125954e-07,
+      "loss": 0.4524,
+      "step": 19786
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.553637437089468e-07,
+      "loss": 0.4365,
+      "step": 19787
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.548151217712662e-07,
+      "loss": 0.4447,
+      "step": 19788
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.542666913242747e-07,
+      "loss": 0.4988,
+      "step": 19789
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.537184523793317e-07,
+      "loss": 0.5068,
+      "step": 19790
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.531704049477939e-07,
+      "loss": 0.425,
+      "step": 19791
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.526225490410111e-07,
+      "loss": 0.4109,
+      "step": 19792
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.52074884670333e-07,
+      "loss": 0.4944,
+      "step": 19793
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.515274118471005e-07,
+      "loss": 0.457,
+      "step": 19794
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.509801305826547e-07,
+      "loss": 0.4688,
+      "step": 19795
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.504330408883298e-07,
+      "loss": 0.437,
+      "step": 19796
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.4988614277546e-07,
+      "loss": 0.4075,
+      "step": 19797
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.493394362553696e-07,
+      "loss": 0.4297,
+      "step": 19798
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.487929213393852e-07,
+      "loss": 0.4119,
+      "step": 19799
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.482465980388232e-07,
+      "loss": 0.4937,
+      "step": 19800
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.477004663650034e-07,
+      "loss": 0.4407,
+      "step": 19801
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.471545263292357e-07,
+      "loss": 0.4781,
+      "step": 19802
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.466087779428244e-07,
+      "loss": 0.4766,
+      "step": 19803
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.460632212170793e-07,
+      "loss": 0.3987,
+      "step": 19804
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.455178561632959e-07,
+      "loss": 0.4958,
+      "step": 19805
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.449726827927728e-07,
+      "loss": 0.3958,
+      "step": 19806
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.444277011167989e-07,
+      "loss": 0.4592,
+      "step": 19807
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.43882911146665e-07,
+      "loss": 0.4163,
+      "step": 19808
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.433383128936523e-07,
+      "loss": 0.4595,
+      "step": 19809
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.427939063690448e-07,
+      "loss": 0.4583,
+      "step": 19810
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.422496915841137e-07,
+      "loss": 0.4399,
+      "step": 19811
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.417056685501356e-07,
+      "loss": 0.4077,
+      "step": 19812
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.411618372783747e-07,
+      "loss": 0.4763,
+      "step": 19813
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.406181977800986e-07,
+      "loss": 0.4607,
+      "step": 19814
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.400747500665639e-07,
+      "loss": 0.468,
+      "step": 19815
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.395314941490273e-07,
+      "loss": 0.5159,
+      "step": 19816
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.389884300387451e-07,
+      "loss": 0.4346,
+      "step": 19817
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.384455577469607e-07,
+      "loss": 0.3925,
+      "step": 19818
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.379028772849217e-07,
+      "loss": 0.4595,
+      "step": 19819
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.373603886638647e-07,
+      "loss": 0.4812,
+      "step": 19820
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.368180918950296e-07,
+      "loss": 0.4153,
+      "step": 19821
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.362759869896463e-07,
+      "loss": 0.4468,
+      "step": 19822
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.357340739589458e-07,
+      "loss": 0.4211,
+      "step": 19823
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.35192352814148e-07,
+      "loss": 0.4451,
+      "step": 19824
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.346508235664784e-07,
+      "loss": 0.4165,
+      "step": 19825
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.341094862271491e-07,
+      "loss": 0.4675,
+      "step": 19826
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.335683408073769e-07,
+      "loss": 0.5457,
+      "step": 19827
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.330273873183647e-07,
+      "loss": 0.4302,
+      "step": 19828
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.324866257713226e-07,
+      "loss": 0.4604,
+      "step": 19829
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.319460561774472e-07,
+      "loss": 0.4329,
+      "step": 19830
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.314056785479373e-07,
+      "loss": 0.437,
+      "step": 19831
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.308654928939851e-07,
+      "loss": 0.4412,
+      "step": 19832
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.303254992267772e-07,
+      "loss": 0.4509,
+      "step": 19833
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.297856975575013e-07,
+      "loss": 0.4023,
+      "step": 19834
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.292460878973351e-07,
+      "loss": 0.4771,
+      "step": 19835
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.287066702574574e-07,
+      "loss": 0.5417,
+      "step": 19836
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.281674446490383e-07,
+      "loss": 0.4521,
+      "step": 19837
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.27628411083251e-07,
+      "loss": 0.5046,
+      "step": 19838
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.270895695712543e-07,
+      "loss": 0.4155,
+      "step": 19839
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.265509201242138e-07,
+      "loss": 0.4084,
+      "step": 19840
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.260124627532849e-07,
+      "loss": 0.4539,
+      "step": 19841
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.254741974696189e-07,
+      "loss": 0.4971,
+      "step": 19842
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.249361242843656e-07,
+      "loss": 0.4104,
+      "step": 19843
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.243982432086716e-07,
+      "loss": 0.426,
+      "step": 19844
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.238605542536747e-07,
+      "loss": 0.4565,
+      "step": 19845
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.23323057430515e-07,
+      "loss": 0.4409,
+      "step": 19846
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.227857527503235e-07,
+      "loss": 0.4663,
+      "step": 19847
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.222486402242279e-07,
+      "loss": 0.4368,
+      "step": 19848
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.217117198633561e-07,
+      "loss": 0.4163,
+      "step": 19849
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.211749916788247e-07,
+      "loss": 0.4209,
+      "step": 19850
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.206384556817558e-07,
+      "loss": 0.4868,
+      "step": 19851
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.201021118832586e-07,
+      "loss": 0.502,
+      "step": 19852
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.195659602944449e-07,
+      "loss": 0.3784,
+      "step": 19853
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.190300009264161e-07,
+      "loss": 0.4822,
+      "step": 19854
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.184942337902778e-07,
+      "loss": 0.5081,
+      "step": 19855
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.179586588971232e-07,
+      "loss": 0.418,
+      "step": 19856
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.17423276258048e-07,
+      "loss": 0.4949,
+      "step": 19857
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.168880858841388e-07,
+      "loss": 0.4409,
+      "step": 19858
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.163530877864832e-07,
+      "loss": 0.4868,
+      "step": 19859
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.158182819761617e-07,
+      "loss": 0.4329,
+      "step": 19860
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.152836684642494e-07,
+      "loss": 0.4387,
+      "step": 19861
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.147492472618233e-07,
+      "loss": 0.4109,
+      "step": 19862
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.142150183799479e-07,
+      "loss": 0.4507,
+      "step": 19863
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.136809818296908e-07,
+      "loss": 0.4434,
+      "step": 19864
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.131471376221155e-07,
+      "loss": 0.5112,
+      "step": 19865
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.126134857682754e-07,
+      "loss": 0.4602,
+      "step": 19866
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.120800262792249e-07,
+      "loss": 0.4495,
+      "step": 19867
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.115467591660153e-07,
+      "loss": 0.4814,
+      "step": 19868
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.110136844396887e-07,
+      "loss": 0.452,
+      "step": 19869
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.104808021112907e-07,
+      "loss": 0.4646,
+      "step": 19870
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.099481121918528e-07,
+      "loss": 0.4531,
+      "step": 19871
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.094156146924147e-07,
+      "loss": 0.448,
+      "step": 19872
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.08883309624e-07,
+      "loss": 0.4304,
+      "step": 19873
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.083511969976376e-07,
+      "loss": 0.4711,
+      "step": 19874
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.078192768243486e-07,
+      "loss": 0.4775,
+      "step": 19875
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.072875491151488e-07,
+      "loss": 0.3997,
+      "step": 19876
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.067560138810525e-07,
+      "loss": 0.4954,
+      "step": 19877
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.062246711330679e-07,
+      "loss": 0.4211,
+      "step": 19878
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.056935208822035e-07,
+      "loss": 0.4399,
+      "step": 19879
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.051625631394576e-07,
+      "loss": 0.4041,
+      "step": 19880
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.04631797915829e-07,
+      "loss": 0.4231,
+      "step": 19881
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.041012252223112e-07,
+      "loss": 0.4077,
+      "step": 19882
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.035708450698941e-07,
+      "loss": 0.4929,
+      "step": 19883
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.030406574695603e-07,
+      "loss": 0.4618,
+      "step": 19884
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.025106624322964e-07,
+      "loss": 0.4379,
+      "step": 19885
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.019808599690748e-07,
+      "loss": 0.3777,
+      "step": 19886
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.014512500908732e-07,
+      "loss": 0.4092,
+      "step": 19887
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.009218328086575e-07,
+      "loss": 0.4514,
+      "step": 19888
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.003926081333968e-07,
+      "loss": 0.4595,
+      "step": 19889
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.998635760760497e-07,
+      "loss": 0.4534,
+      "step": 19890
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.993347366475756e-07,
+      "loss": 0.4578,
+      "step": 19891
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.988060898589266e-07,
+      "loss": 0.4705,
+      "step": 19892
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.982776357210519e-07,
+      "loss": 0.54,
+      "step": 19893
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.977493742449004e-07,
+      "loss": 0.4563,
+      "step": 19894
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.972213054414101e-07,
+      "loss": 0.4502,
+      "step": 19895
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.966934293215222e-07,
+      "loss": 0.4109,
+      "step": 19896
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.961657458961657e-07,
+      "loss": 0.47,
+      "step": 19897
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.956382551762753e-07,
+      "loss": 0.4014,
+      "step": 19898
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.951109571727721e-07,
+      "loss": 0.5015,
+      "step": 19899
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.945838518965808e-07,
+      "loss": 0.5088,
+      "step": 19900
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.940569393586171e-07,
+      "loss": 0.4397,
+      "step": 19901
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.935302195697968e-07,
+      "loss": 0.5139,
+      "step": 19902
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.930036925410266e-07,
+      "loss": 0.4724,
+      "step": 19903
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.924773582832156e-07,
+      "loss": 0.4497,
+      "step": 19904
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.919512168072628e-07,
+      "loss": 0.4968,
+      "step": 19905
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.914252681240663e-07,
+      "loss": 0.447,
+      "step": 19906
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.908995122445228e-07,
+      "loss": 0.4678,
+      "step": 19907
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.903739491795169e-07,
+      "loss": 0.4529,
+      "step": 19908
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.89848578939939e-07,
+      "loss": 0.5083,
+      "step": 19909
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.893234015366667e-07,
+      "loss": 0.4377,
+      "step": 19910
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.887984169805817e-07,
+      "loss": 0.4419,
+      "step": 19911
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.88273625282554e-07,
+      "loss": 0.4888,
+      "step": 19912
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.877490264534569e-07,
+      "loss": 0.449,
+      "step": 19913
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.872246205041533e-07,
+      "loss": 0.4504,
+      "step": 19914
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.867004074455075e-07,
+      "loss": 0.4636,
+      "step": 19915
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.861763872883743e-07,
+      "loss": 0.4873,
+      "step": 19916
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.856525600436092e-07,
+      "loss": 0.4093,
+      "step": 19917
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.851289257220628e-07,
+      "loss": 0.4321,
+      "step": 19918
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.846054843345784e-07,
+      "loss": 0.49,
+      "step": 19919
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.840822358920007e-07,
+      "loss": 0.4885,
+      "step": 19920
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.835591804051644e-07,
+      "loss": 0.4153,
+      "step": 19921
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.830363178849076e-07,
+      "loss": 0.447,
+      "step": 19922
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.825136483420547e-07,
+      "loss": 0.4885,
+      "step": 19923
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.81991171787435e-07,
+      "loss": 0.4268,
+      "step": 19924
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.814688882318688e-07,
+      "loss": 0.4192,
+      "step": 19925
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.809467976861761e-07,
+      "loss": 0.3689,
+      "step": 19926
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.804249001611662e-07,
+      "loss": 0.4368,
+      "step": 19927
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.79903195667655e-07,
+      "loss": 0.4231,
+      "step": 19928
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.793816842164425e-07,
+      "loss": 0.4729,
+      "step": 19929
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.788603658183346e-07,
+      "loss": 0.3677,
+      "step": 19930
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.783392404841272e-07,
+      "loss": 0.4641,
+      "step": 19931
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.778183082246148e-07,
+      "loss": 0.4414,
+      "step": 19932
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.772975690505857e-07,
+      "loss": 0.4478,
+      "step": 19933
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.767770229728276e-07,
+      "loss": 0.4048,
+      "step": 19934
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.762566700021223e-07,
+      "loss": 0.4553,
+      "step": 19935
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.757365101492452e-07,
+      "loss": 0.4426,
+      "step": 19936
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.752165434249736e-07,
+      "loss": 0.4185,
+      "step": 19937
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.74696769840073e-07,
+      "loss": 0.4658,
+      "step": 19938
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.741771894053139e-07,
+      "loss": 0.4658,
+      "step": 19939
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.736578021314544e-07,
+      "loss": 0.425,
+      "step": 19940
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.731386080292535e-07,
+      "loss": 0.4233,
+      "step": 19941
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.72619607109466e-07,
+      "loss": 0.4165,
+      "step": 19942
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.721007993828399e-07,
+      "loss": 0.4763,
+      "step": 19943
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.715821848601223e-07,
+      "loss": 0.4163,
+      "step": 19944
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.710637635520568e-07,
+      "loss": 0.4396,
+      "step": 19945
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.705455354693768e-07,
+      "loss": 0.3685,
+      "step": 19946
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.700275006228196e-07,
+      "loss": 0.4365,
+      "step": 19947
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.695096590231132e-07,
+      "loss": 0.5032,
+      "step": 19948
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.689920106809844e-07,
+      "loss": 0.4343,
+      "step": 19949
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.684745556071548e-07,
+      "loss": 0.4717,
+      "step": 19950
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.679572938123413e-07,
+      "loss": 0.46,
+      "step": 19951
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.674402253072587e-07,
+      "loss": 0.4902,
+      "step": 19952
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.669233501026151e-07,
+      "loss": 0.4629,
+      "step": 19953
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.664066682091197e-07,
+      "loss": 0.4695,
+      "step": 19954
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.658901796374695e-07,
+      "loss": 0.4399,
+      "step": 19955
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.65373884398367e-07,
+      "loss": 0.4951,
+      "step": 19956
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.648577825025016e-07,
+      "loss": 0.4492,
+      "step": 19957
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.643418739605679e-07,
+      "loss": 0.4495,
+      "step": 19958
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.638261587832473e-07,
+      "loss": 0.4834,
+      "step": 19959
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.633106369812236e-07,
+      "loss": 0.4788,
+      "step": 19960
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.627953085651739e-07,
+      "loss": 0.4768,
+      "step": 19961
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.622801735457729e-07,
+      "loss": 0.4822,
+      "step": 19962
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.617652319336887e-07,
+      "loss": 0.3757,
+      "step": 19963
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.612504837395884e-07,
+      "loss": 0.4456,
+      "step": 19964
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.607359289741344e-07,
+      "loss": 0.47,
+      "step": 19965
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.602215676479807e-07,
+      "loss": 0.439,
+      "step": 19966
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.59707399771784e-07,
+      "loss": 0.5264,
+      "step": 19967
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.59193425356196e-07,
+      "loss": 0.4304,
+      "step": 19968
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.586796444118571e-07,
+      "loss": 0.5081,
+      "step": 19969
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.581660569494119e-07,
+      "loss": 0.3972,
+      "step": 19970
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.576526629794999e-07,
+      "loss": 0.4663,
+      "step": 19971
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.571394625127514e-07,
+      "loss": 0.5024,
+      "step": 19972
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.56626455559799e-07,
+      "loss": 0.5073,
+      "step": 19973
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.561136421312653e-07,
+      "loss": 0.501,
+      "step": 19974
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.55601022237774e-07,
+      "loss": 0.4048,
+      "step": 19975
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.550885958899423e-07,
+      "loss": 0.4883,
+      "step": 19976
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.545763630983837e-07,
+      "loss": 0.4304,
+      "step": 19977
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.540643238737077e-07,
+      "loss": 0.4734,
+      "step": 19978
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.535524782265201e-07,
+      "loss": 0.4021,
+      "step": 19979
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.530408261674237e-07,
+      "loss": 0.3701,
+      "step": 19980
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.525293677070122e-07,
+      "loss": 0.4102,
+      "step": 19981
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.520181028558836e-07,
+      "loss": 0.4253,
+      "step": 19982
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.515070316246241e-07,
+      "loss": 0.4744,
+      "step": 19983
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.50996154023823e-07,
+      "loss": 0.4204,
+      "step": 19984
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.504854700640573e-07,
+      "loss": 0.4639,
+      "step": 19985
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.499749797559085e-07,
+      "loss": 0.4341,
+      "step": 19986
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.494646831099472e-07,
+      "loss": 0.4995,
+      "step": 19987
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.489545801367459e-07,
+      "loss": 0.4521,
+      "step": 19988
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.484446708468661e-07,
+      "loss": 0.4636,
+      "step": 19989
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.47934955250874e-07,
+      "loss": 0.3901,
+      "step": 19990
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.474254333593221e-07,
+      "loss": 0.4939,
+      "step": 19991
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.469161051827688e-07,
+      "loss": 0.4568,
+      "step": 19992
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.464069707317588e-07,
+      "loss": 0.4121,
+      "step": 19993
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.458980300168405e-07,
+      "loss": 0.4258,
+      "step": 19994
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.453892830485565e-07,
+      "loss": 0.491,
+      "step": 19995
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.448807298374404e-07,
+      "loss": 0.4414,
+      "step": 19996
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.443723703940286e-07,
+      "loss": 0.4587,
+      "step": 19997
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.43864204728849e-07,
+      "loss": 0.458,
+      "step": 19998
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.433562328524279e-07,
+      "loss": 0.4336,
+      "step": 19999
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.428484547752856e-07,
+      "loss": 0.4246,
+      "step": 20000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.423408705079415e-07,
+      "loss": 0.4136,
+      "step": 20001
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.418334800609061e-07,
+      "loss": 0.4412,
+      "step": 20002
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.41326283444691e-07,
+      "loss": 0.4517,
+      "step": 20003
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.408192806697988e-07,
+      "loss": 0.3875,
+      "step": 20004
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.403124717467346e-07,
+      "loss": 0.3723,
+      "step": 20005
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.398058566859921e-07,
+      "loss": 0.4866,
+      "step": 20006
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.392994354980686e-07,
+      "loss": 0.4329,
+      "step": 20007
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.387932081934478e-07,
+      "loss": 0.4231,
+      "step": 20008
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.382871747826203e-07,
+      "loss": 0.5037,
+      "step": 20009
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.377813352760642e-07,
+      "loss": 0.4417,
+      "step": 20010
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.372756896842569e-07,
+      "loss": 0.4307,
+      "step": 20011
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.367702380176733e-07,
+      "loss": 0.4624,
+      "step": 20012
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.362649802867804e-07,
+      "loss": 0.4023,
+      "step": 20013
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.357599165020445e-07,
+      "loss": 0.3938,
+      "step": 20014
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.352550466739261e-07,
+      "loss": 0.458,
+      "step": 20015
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.347503708128844e-07,
+      "loss": 0.478,
+      "step": 20016
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.34245888929369e-07,
+      "loss": 0.4502,
+      "step": 20017
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.337416010338304e-07,
+      "loss": 0.4258,
+      "step": 20018
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.332375071367158e-07,
+      "loss": 0.4409,
+      "step": 20019
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.327336072484636e-07,
+      "loss": 0.426,
+      "step": 20020
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.322299013795108e-07,
+      "loss": 0.4573,
+      "step": 20021
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.317263895402937e-07,
+      "loss": 0.467,
+      "step": 20022
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.312230717412371e-07,
+      "loss": 0.4321,
+      "step": 20023
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.307199479927683e-07,
+      "loss": 0.4741,
+      "step": 20024
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.302170183053091e-07,
+      "loss": 0.45,
+      "step": 20025
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.297142826892721e-07,
+      "loss": 0.4412,
+      "step": 20026
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.292117411550758e-07,
+      "loss": 0.4908,
+      "step": 20027
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.28709393713125e-07,
+      "loss": 0.4441,
+      "step": 20028
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.282072403738271e-07,
+      "loss": 0.4714,
+      "step": 20029
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.277052811475792e-07,
+      "loss": 0.4421,
+      "step": 20030
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.272035160447843e-07,
+      "loss": 0.4619,
+      "step": 20031
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.267019450758282e-07,
+      "loss": 0.4519,
+      "step": 20032
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.262005682511052e-07,
+      "loss": 0.4482,
+      "step": 20033
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.256993855809967e-07,
+      "loss": 0.4404,
+      "step": 20034
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.251983970758868e-07,
+      "loss": 0.4395,
+      "step": 20035
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.246976027461471e-07,
+      "loss": 0.4478,
+      "step": 20036
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.24197002602156e-07,
+      "loss": 0.4419,
+      "step": 20037
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.236965966542774e-07,
+      "loss": 0.4468,
+      "step": 20038
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.231963849128797e-07,
+      "loss": 0.4199,
+      "step": 20039
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.226963673883212e-07,
+      "loss": 0.4551,
+      "step": 20040
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.221965440909583e-07,
+      "loss": 0.4729,
+      "step": 20041
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.216969150311459e-07,
+      "loss": 0.382,
+      "step": 20042
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.2119748021923e-07,
+      "loss": 0.4463,
+      "step": 20043
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.206982396655558e-07,
+      "loss": 0.4229,
+      "step": 20044
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.201991933804652e-07,
+      "loss": 0.3699,
+      "step": 20045
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.19700341374293e-07,
+      "loss": 0.4409,
+      "step": 20046
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.19201683657371e-07,
+      "loss": 0.4558,
+      "step": 20047
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.187032202400322e-07,
+      "loss": 0.4617,
+      "step": 20048
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.182049511325949e-07,
+      "loss": 0.429,
+      "step": 20049
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.177068763453853e-07,
+      "loss": 0.4368,
+      "step": 20050
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.17208995888714e-07,
+      "loss": 0.439,
+      "step": 20051
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.167113097728983e-07,
+      "loss": 0.4539,
+      "step": 20052
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.162138180082433e-07,
+      "loss": 0.4043,
+      "step": 20053
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.157165206050553e-07,
+      "loss": 0.417,
+      "step": 20054
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.152194175736337e-07,
+      "loss": 0.4438,
+      "step": 20055
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.147225089242725e-07,
+      "loss": 0.4849,
+      "step": 20056
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.142257946672692e-07,
+      "loss": 0.5083,
+      "step": 20057
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.137292748129065e-07,
+      "loss": 0.4277,
+      "step": 20058
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.132329493714717e-07,
+      "loss": 0.4553,
+      "step": 20059
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.127368183532434e-07,
+      "loss": 0.4844,
+      "step": 20060
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.12240881768501e-07,
+      "loss": 0.4773,
+      "step": 20061
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.11745139627511e-07,
+      "loss": 0.3789,
+      "step": 20062
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.11249591940547e-07,
+      "loss": 0.4241,
+      "step": 20063
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.107542387178678e-07,
+      "loss": 0.4873,
+      "step": 20064
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.102590799697383e-07,
+      "loss": 0.4619,
+      "step": 20065
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.097641157064116e-07,
+      "loss": 0.415,
+      "step": 20066
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.092693459381405e-07,
+      "loss": 0.5225,
+      "step": 20067
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.087747706751724e-07,
+      "loss": 0.4502,
+      "step": 20068
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.082803899277523e-07,
+      "loss": 0.4121,
+      "step": 20069
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.077862037061177e-07,
+      "loss": 0.4148,
+      "step": 20070
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.07292212020506e-07,
+      "loss": 0.4133,
+      "step": 20071
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.067984148811501e-07,
+      "loss": 0.446,
+      "step": 20072
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.063048122982762e-07,
+      "loss": 0.5254,
+      "step": 20073
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.058114042821095e-07,
+      "loss": 0.4241,
+      "step": 20074
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.053181908428663e-07,
+      "loss": 0.4666,
+      "step": 20075
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.048251719907672e-07,
+      "loss": 0.4645,
+      "step": 20076
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.043323477360196e-07,
+      "loss": 0.4575,
+      "step": 20077
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.038397180888333e-07,
+      "loss": 0.4426,
+      "step": 20078
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.033472830594111e-07,
+      "loss": 0.4358,
+      "step": 20079
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.028550426579538e-07,
+      "loss": 0.5273,
+      "step": 20080
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.023629968946543e-07,
+      "loss": 0.502,
+      "step": 20081
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.018711457797078e-07,
+      "loss": 0.4849,
+      "step": 20082
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.013794893232993e-07,
+      "loss": 0.4504,
+      "step": 20083
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.00888027535611e-07,
+      "loss": 0.4626,
+      "step": 20084
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.003967604268257e-07,
+      "loss": 0.4424,
+      "step": 20085
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.999056880071152e-07,
+      "loss": 0.4812,
+      "step": 20086
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.994148102866537e-07,
+      "loss": 0.5002,
+      "step": 20087
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.989241272756064e-07,
+      "loss": 0.4512,
+      "step": 20088
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.984336389841394e-07,
+      "loss": 0.4297,
+      "step": 20089
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.979433454224082e-07,
+      "loss": 0.519,
+      "step": 20090
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.974532466005712e-07,
+      "loss": 0.4788,
+      "step": 20091
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.969633425287757e-07,
+      "loss": 0.4368,
+      "step": 20092
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.964736332171738e-07,
+      "loss": 0.3494,
+      "step": 20093
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.959841186759053e-07,
+      "loss": 0.394,
+      "step": 20094
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.954947989151094e-07,
+      "loss": 0.4229,
+      "step": 20095
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.95005673944924e-07,
+      "loss": 0.4705,
+      "step": 20096
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.945167437754751e-07,
+      "loss": 0.4941,
+      "step": 20097
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.940280084168948e-07,
+      "loss": 0.3896,
+      "step": 20098
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.935394678793027e-07,
+      "loss": 0.4475,
+      "step": 20099
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.930511221728196e-07,
+      "loss": 0.4243,
+      "step": 20100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.925629713075575e-07,
+      "loss": 0.4575,
+      "step": 20101
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.920750152936305e-07,
+      "loss": 0.4233,
+      "step": 20102
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.915872541411439e-07,
+      "loss": 0.3621,
+      "step": 20103
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.910996878602005e-07,
+      "loss": 0.4626,
+      "step": 20104
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.90612316460899e-07,
+      "loss": 0.4512,
+      "step": 20105
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.901251399533347e-07,
+      "loss": 0.4458,
+      "step": 20106
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.896381583475963e-07,
+      "loss": 0.4583,
+      "step": 20107
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.891513716537734e-07,
+      "loss": 0.4585,
+      "step": 20108
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.886647798819456e-07,
+      "loss": 0.4612,
+      "step": 20109
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.881783830421939e-07,
+      "loss": 0.4609,
+      "step": 20110
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.876921811445901e-07,
+      "loss": 0.4556,
+      "step": 20111
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.872061741992075e-07,
+      "loss": 0.4127,
+      "step": 20112
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.86720362216111e-07,
+      "loss": 0.4841,
+      "step": 20113
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.86234745205363e-07,
+      "loss": 0.4719,
+      "step": 20114
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.857493231770217e-07,
+      "loss": 0.4426,
+      "step": 20115
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.852640961411415e-07,
+      "loss": 0.4058,
+      "step": 20116
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.847790641077733e-07,
+      "loss": 0.4315,
+      "step": 20117
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.842942270869623e-07,
+      "loss": 0.4094,
+      "step": 20118
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.838095850887526e-07,
+      "loss": 0.4355,
+      "step": 20119
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.833251381231786e-07,
+      "loss": 0.4604,
+      "step": 20120
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.828408862002777e-07,
+      "loss": 0.4827,
+      "step": 20121
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.823568293300807e-07,
+      "loss": 0.4421,
+      "step": 20122
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.818729675226098e-07,
+      "loss": 0.511,
+      "step": 20123
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.81389300787889e-07,
+      "loss": 0.3779,
+      "step": 20124
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.809058291359382e-07,
+      "loss": 0.4268,
+      "step": 20125
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.80422552576767e-07,
+      "loss": 0.5,
+      "step": 20126
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.79939471120391e-07,
+      "loss": 0.4434,
+      "step": 20127
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.794565847768108e-07,
+      "loss": 0.4648,
+      "step": 20128
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.789738935560296e-07,
+      "loss": 0.4917,
+      "step": 20129
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.784913974680473e-07,
+      "loss": 0.4138,
+      "step": 20130
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.780090965228535e-07,
+      "loss": 0.4155,
+      "step": 20131
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.775269907304426e-07,
+      "loss": 0.4426,
+      "step": 20132
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.770450801007965e-07,
+      "loss": 0.4309,
+      "step": 20133
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.765633646438995e-07,
+      "loss": 0.4133,
+      "step": 20134
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.760818443697258e-07,
+      "loss": 0.4377,
+      "step": 20135
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.756005192882519e-07,
+      "loss": 0.4336,
+      "step": 20136
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.751193894094454e-07,
+      "loss": 0.5249,
+      "step": 20137
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.746384547432738e-07,
+      "loss": 0.4292,
+      "step": 20138
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.741577152996958e-07,
+      "loss": 0.5273,
+      "step": 20139
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.736771710886713e-07,
+      "loss": 0.5037,
+      "step": 20140
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.7319682212015e-07,
+      "loss": 0.4662,
+      "step": 20141
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.727166684040863e-07,
+      "loss": 0.4277,
+      "step": 20142
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.722367099504211e-07,
+      "loss": 0.4253,
+      "step": 20143
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.717569467690965e-07,
+      "loss": 0.4578,
+      "step": 20144
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.712773788700498e-07,
+      "loss": 0.4817,
+      "step": 20145
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.707980062632146e-07,
+      "loss": 0.4509,
+      "step": 20146
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.703188289585193e-07,
+      "loss": 0.4941,
+      "step": 20147
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.698398469658883e-07,
+      "loss": 0.4254,
+      "step": 20148
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.693610602952449e-07,
+      "loss": 0.4504,
+      "step": 20149
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.688824689565032e-07,
+      "loss": 0.4453,
+      "step": 20150
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.684040729595785e-07,
+      "loss": 0.4268,
+      "step": 20151
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.679258723143765e-07,
+      "loss": 0.4038,
+      "step": 20152
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.674478670308047e-07,
+      "loss": 0.4027,
+      "step": 20153
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.66970057118762e-07,
+      "loss": 0.4399,
+      "step": 20154
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.664924425881469e-07,
+      "loss": 0.4161,
+      "step": 20155
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.660150234488493e-07,
+      "loss": 0.4424,
+      "step": 20156
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.655377997107603e-07,
+      "loss": 0.5044,
+      "step": 20157
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.65060771383763e-07,
+      "loss": 0.3882,
+      "step": 20158
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.645839384777374e-07,
+      "loss": 0.4827,
+      "step": 20159
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.641073010025622e-07,
+      "loss": 0.4563,
+      "step": 20160
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.636308589681072e-07,
+      "loss": 0.4282,
+      "step": 20161
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.631546123842424e-07,
+      "loss": 0.4468,
+      "step": 20162
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.626785612608299e-07,
+      "loss": 0.4009,
+      "step": 20163
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.622027056077329e-07,
+      "loss": 0.4351,
+      "step": 20164
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.617270454348045e-07,
+      "loss": 0.3889,
+      "step": 20165
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.612515807518992e-07,
+      "loss": 0.426,
+      "step": 20166
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.607763115688636e-07,
+      "loss": 0.407,
+      "step": 20167
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.60301237895543e-07,
+      "loss": 0.4988,
+      "step": 20168
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.598263597417753e-07,
+      "loss": 0.49,
+      "step": 20169
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.59351677117399e-07,
+      "loss": 0.4387,
+      "step": 20170
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.588771900322432e-07,
+      "loss": 0.4448,
+      "step": 20171
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.584028984961376e-07,
+      "loss": 0.4485,
+      "step": 20172
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.579288025189066e-07,
+      "loss": 0.4287,
+      "step": 20173
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.574549021103681e-07,
+      "loss": 0.4275,
+      "step": 20174
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.569811972803385e-07,
+      "loss": 0.4778,
+      "step": 20175
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.565076880386289e-07,
+      "loss": 0.4111,
+      "step": 20176
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.560343743950492e-07,
+      "loss": 0.4208,
+      "step": 20177
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.555612563593982e-07,
+      "loss": 0.4392,
+      "step": 20178
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.550883339414803e-07,
+      "loss": 0.5457,
+      "step": 20179
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.546156071510877e-07,
+      "loss": 0.4858,
+      "step": 20180
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.541430759980137e-07,
+      "loss": 0.4176,
+      "step": 20181
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.536707404920428e-07,
+      "loss": 0.4279,
+      "step": 20182
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.531986006429623e-07,
+      "loss": 0.4592,
+      "step": 20183
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.527266564605471e-07,
+      "loss": 0.4236,
+      "step": 20184
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.522549079545769e-07,
+      "loss": 0.4937,
+      "step": 20185
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.517833551348185e-07,
+      "loss": 0.4082,
+      "step": 20186
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.513119980110415e-07,
+      "loss": 0.4259,
+      "step": 20187
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.508408365930085e-07,
+      "loss": 0.4468,
+      "step": 20188
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.50369870890477e-07,
+      "loss": 0.3921,
+      "step": 20189
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.498991009132037e-07,
+      "loss": 0.4771,
+      "step": 20190
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.494285266709376e-07,
+      "loss": 0.436,
+      "step": 20191
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.489581481734274e-07,
+      "loss": 0.4058,
+      "step": 20192
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.484879654304143e-07,
+      "loss": 0.4304,
+      "step": 20193
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.480179784516382e-07,
+      "loss": 0.4573,
+      "step": 20194
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.475481872468324e-07,
+      "loss": 0.3894,
+      "step": 20195
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.470785918257293e-07,
+      "loss": 0.425,
+      "step": 20196
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.466091921980521e-07,
+      "loss": 0.4204,
+      "step": 20197
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.461399883735252e-07,
+      "loss": 0.3884,
+      "step": 20198
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.456709803618698e-07,
+      "loss": 0.4214,
+      "step": 20199
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.452021681727948e-07,
+      "loss": 0.4285,
+      "step": 20200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.447335518160135e-07,
+      "loss": 0.4595,
+      "step": 20201
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.442651313012326e-07,
+      "loss": 0.4819,
+      "step": 20202
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.437969066381543e-07,
+      "loss": 0.4592,
+      "step": 20203
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.433288778364742e-07,
+      "loss": 0.4968,
+      "step": 20204
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.428610449058891e-07,
+      "loss": 0.4446,
+      "step": 20205
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.423934078560867e-07,
+      "loss": 0.3953,
+      "step": 20206
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.419259666967547e-07,
+      "loss": 0.489,
+      "step": 20207
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.414587214375733e-07,
+      "loss": 0.4851,
+      "step": 20208
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.409916720882225e-07,
+      "loss": 0.3989,
+      "step": 20209
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.405248186583744e-07,
+      "loss": 0.4805,
+      "step": 20210
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.400581611576994e-07,
+      "loss": 0.4331,
+      "step": 20211
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.395916995958617e-07,
+      "loss": 0.4182,
+      "step": 20212
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.391254339825258e-07,
+      "loss": 0.439,
+      "step": 20213
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.386593643273464e-07,
+      "loss": 0.4255,
+      "step": 20214
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.381934906399788e-07,
+      "loss": 0.5068,
+      "step": 20215
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.377278129300701e-07,
+      "loss": 0.4033,
+      "step": 20216
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.372623312072689e-07,
+      "loss": 0.4331,
+      "step": 20217
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.367970454812143e-07,
+      "loss": 0.4644,
+      "step": 20218
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.363319557615432e-07,
+      "loss": 0.4336,
+      "step": 20219
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.35867062057891e-07,
+      "loss": 0.4424,
+      "step": 20220
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.354023643798834e-07,
+      "loss": 0.4224,
+      "step": 20221
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.349378627371471e-07,
+      "loss": 0.4556,
+      "step": 20222
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.344735571393056e-07,
+      "loss": 0.416,
+      "step": 20223
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.340094475959712e-07,
+      "loss": 0.4893,
+      "step": 20224
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.335455341167594e-07,
+      "loss": 0.4866,
+      "step": 20225
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.330818167112806e-07,
+      "loss": 0.4487,
+      "step": 20226
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.326182953891357e-07,
+      "loss": 0.4285,
+      "step": 20227
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.321549701599293e-07,
+      "loss": 0.4365,
+      "step": 20228
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.316918410332539e-07,
+      "loss": 0.4565,
+      "step": 20229
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.312289080187061e-07,
+      "loss": 0.4045,
+      "step": 20230
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.307661711258704e-07,
+      "loss": 0.4546,
+      "step": 20231
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.30303630364336e-07,
+      "loss": 0.4895,
+      "step": 20232
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.298412857436796e-07,
+      "loss": 0.512,
+      "step": 20233
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.29379137273478e-07,
+      "loss": 0.4487,
+      "step": 20234
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.289171849633045e-07,
+      "loss": 0.4629,
+      "step": 20235
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.28455428822725e-07,
+      "loss": 0.4551,
+      "step": 20236
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.279938688613073e-07,
+      "loss": 0.4546,
+      "step": 20237
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.275325050886082e-07,
+      "loss": 0.4025,
+      "step": 20238
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.270713375141856e-07,
+      "loss": 0.4351,
+      "step": 20239
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.266103661475896e-07,
+      "loss": 0.4817,
+      "step": 20240
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.261495909983716e-07,
+      "loss": 0.4614,
+      "step": 20241
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.256890120760705e-07,
+      "loss": 0.3792,
+      "step": 20242
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.252286293902298e-07,
+      "loss": 0.4241,
+      "step": 20243
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.247684429503841e-07,
+      "loss": 0.3949,
+      "step": 20244
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.243084527660647e-07,
+      "loss": 0.4307,
+      "step": 20245
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.238486588467984e-07,
+      "loss": 0.4863,
+      "step": 20246
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.233890612021108e-07,
+      "loss": 0.4434,
+      "step": 20247
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.22929659841519e-07,
+      "loss": 0.4731,
+      "step": 20248
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.224704547745385e-07,
+      "loss": 0.4629,
+      "step": 20249
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.22011446010684e-07,
+      "loss": 0.4365,
+      "step": 20250
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.215526335594579e-07,
+      "loss": 0.429,
+      "step": 20251
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.210940174303669e-07,
+      "loss": 0.4363,
+      "step": 20252
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.206355976329081e-07,
+      "loss": 0.4724,
+      "step": 20253
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.201773741765781e-07,
+      "loss": 0.4371,
+      "step": 20254
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.197193470708661e-07,
+      "loss": 0.4697,
+      "step": 20255
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.192615163252601e-07,
+      "loss": 0.47,
+      "step": 20256
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.188038819492425e-07,
+      "loss": 0.4387,
+      "step": 20257
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.183464439522934e-07,
+      "loss": 0.4714,
+      "step": 20258
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.178892023438842e-07,
+      "loss": 0.4526,
+      "step": 20259
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.174321571334884e-07,
+      "loss": 0.5183,
+      "step": 20260
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.169753083305717e-07,
+      "loss": 0.3984,
+      "step": 20261
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.165186559445967e-07,
+      "loss": 0.4241,
+      "step": 20262
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.160621999850224e-07,
+      "loss": 0.4929,
+      "step": 20263
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.156059404613012e-07,
+      "loss": 0.4421,
+      "step": 20264
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.151498773828856e-07,
+      "loss": 0.491,
+      "step": 20265
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.146940107592191e-07,
+      "loss": 0.3975,
+      "step": 20266
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.142383405997475e-07,
+      "loss": 0.4148,
+      "step": 20267
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.137828669139045e-07,
+      "loss": 0.4592,
+      "step": 20268
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.13327589711129e-07,
+      "loss": 0.4302,
+      "step": 20269
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.128725090008458e-07,
+      "loss": 0.3849,
+      "step": 20270
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.12417624792485e-07,
+      "loss": 0.4255,
+      "step": 20271
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.11962937095466e-07,
+      "loss": 0.4109,
+      "step": 20272
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.115084459192077e-07,
+      "loss": 0.3933,
+      "step": 20273
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.110541512731227e-07,
+      "loss": 0.4573,
+      "step": 20274
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.106000531666211e-07,
+      "loss": 0.4294,
+      "step": 20275
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.101461516091088e-07,
+      "loss": 0.4819,
+      "step": 20276
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.096924466099862e-07,
+      "loss": 0.4402,
+      "step": 20277
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.092389381786522e-07,
+      "loss": 0.4292,
+      "step": 20278
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.087856263244972e-07,
+      "loss": 0.4417,
+      "step": 20279
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.083325110569148e-07,
+      "loss": 0.5146,
+      "step": 20280
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.078795923852864e-07,
+      "loss": 0.4241,
+      "step": 20281
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.074268703189955e-07,
+      "loss": 0.425,
+      "step": 20282
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.069743448674158e-07,
+      "loss": 0.4517,
+      "step": 20283
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.065220160399242e-07,
+      "loss": 0.4167,
+      "step": 20284
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.060698838458866e-07,
+      "loss": 0.4939,
+      "step": 20285
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.056179482946711e-07,
+      "loss": 0.4362,
+      "step": 20286
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.051662093956333e-07,
+      "loss": 0.4934,
+      "step": 20287
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.047146671581349e-07,
+      "loss": 0.4185,
+      "step": 20288
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.04263321591525e-07,
+      "loss": 0.4648,
+      "step": 20289
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.038121727051548e-07,
+      "loss": 0.4526,
+      "step": 20290
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.033612205083671e-07,
+      "loss": 0.4241,
+      "step": 20291
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.02910465010501e-07,
+      "loss": 0.4783,
+      "step": 20292
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.024599062208957e-07,
+      "loss": 0.416,
+      "step": 20293
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.020095441488804e-07,
+      "loss": 0.4509,
+      "step": 20294
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.015593788037864e-07,
+      "loss": 0.4421,
+      "step": 20295
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.011094101949344e-07,
+      "loss": 0.4321,
+      "step": 20296
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.006596383316475e-07,
+      "loss": 0.4185,
+      "step": 20297
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.002100632232387e-07,
+      "loss": 0.4492,
+      "step": 20298
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.997606848790215e-07,
+      "loss": 0.4558,
+      "step": 20299
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.99311503308304e-07,
+      "loss": 0.426,
+      "step": 20300
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.988625185203888e-07,
+      "loss": 0.5081,
+      "step": 20301
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.984137305245751e-07,
+      "loss": 0.4619,
+      "step": 20302
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.97965139330161e-07,
+      "loss": 0.4277,
+      "step": 20303
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.975167449464347e-07,
+      "loss": 0.4167,
+      "step": 20304
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.970685473826864e-07,
+      "loss": 0.4675,
+      "step": 20305
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.966205466481988e-07,
+      "loss": 0.4814,
+      "step": 20306
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.961727427522478e-07,
+      "loss": 0.4585,
+      "step": 20307
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.957251357041126e-07,
+      "loss": 0.4404,
+      "step": 20308
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.952777255130614e-07,
+      "loss": 0.4053,
+      "step": 20309
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.948305121883645e-07,
+      "loss": 0.458,
+      "step": 20310
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.943834957392813e-07,
+      "loss": 0.5022,
+      "step": 20311
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.93936676175073e-07,
+      "loss": 0.4231,
+      "step": 20312
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.934900535049913e-07,
+      "loss": 0.4722,
+      "step": 20313
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.93043627738291e-07,
+      "loss": 0.4241,
+      "step": 20314
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.925973988842158e-07,
+      "loss": 0.4609,
+      "step": 20315
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.921513669520094e-07,
+      "loss": 0.4291,
+      "step": 20316
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9170553195091e-07,
+      "loss": 0.4839,
+      "step": 20317
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.912598938901525e-07,
+      "loss": 0.4255,
+      "step": 20318
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.908144527789649e-07,
+      "loss": 0.4634,
+      "step": 20319
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.903692086265765e-07,
+      "loss": 0.467,
+      "step": 20320
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.899241614422079e-07,
+      "loss": 0.4482,
+      "step": 20321
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.894793112350759e-07,
+      "loss": 0.4331,
+      "step": 20322
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.890346580143945e-07,
+      "loss": 0.4016,
+      "step": 20323
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.885902017893773e-07,
+      "loss": 0.4382,
+      "step": 20324
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.88145942569227e-07,
+      "loss": 0.4146,
+      "step": 20325
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.877018803631439e-07,
+      "loss": 0.4175,
+      "step": 20326
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.872580151803308e-07,
+      "loss": 0.5081,
+      "step": 20327
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.868143470299758e-07,
+      "loss": 0.4861,
+      "step": 20328
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.863708759212715e-07,
+      "loss": 0.4746,
+      "step": 20329
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.859276018634018e-07,
+      "loss": 0.4413,
+      "step": 20330
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.854845248655493e-07,
+      "loss": 0.4658,
+      "step": 20331
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.8504164493689e-07,
+      "loss": 0.469,
+      "step": 20332
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.845989620865988e-07,
+      "loss": 0.4229,
+      "step": 20333
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.841564763238416e-07,
+      "loss": 0.4626,
+      "step": 20334
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.837141876577867e-07,
+      "loss": 0.5334,
+      "step": 20335
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.832720960975934e-07,
+      "loss": 0.4167,
+      "step": 20336
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.828302016524177e-07,
+      "loss": 0.4612,
+      "step": 20337
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.823885043314147e-07,
+      "loss": 0.5112,
+      "step": 20338
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.819470041437313e-07,
+      "loss": 0.4766,
+      "step": 20339
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.815057010985125e-07,
+      "loss": 0.4082,
+      "step": 20340
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.810645952048975e-07,
+      "loss": 0.4429,
+      "step": 20341
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.80623686472026e-07,
+      "loss": 0.4302,
+      "step": 20342
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.801829749090259e-07,
+      "loss": 0.4531,
+      "step": 20343
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.797424605250289e-07,
+      "loss": 0.4497,
+      "step": 20344
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.793021433291578e-07,
+      "loss": 0.5056,
+      "step": 20345
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.788620233305341e-07,
+      "loss": 0.4001,
+      "step": 20346
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.784221005382706e-07,
+      "loss": 0.4736,
+      "step": 20347
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.779823749614821e-07,
+      "loss": 0.4203,
+      "step": 20348
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.775428466092746e-07,
+      "loss": 0.4062,
+      "step": 20349
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.771035154907544e-07,
+      "loss": 0.4338,
+      "step": 20350
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.766643816150174e-07,
+      "loss": 0.4141,
+      "step": 20351
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7622544499116186e-07,
+      "loss": 0.4563,
+      "step": 20352
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7578670562827945e-07,
+      "loss": 0.4294,
+      "step": 20353
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.753481635354551e-07,
+      "loss": 0.364,
+      "step": 20354
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7490981872177486e-07,
+      "loss": 0.4456,
+      "step": 20355
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.74471671196316e-07,
+      "loss": 0.4495,
+      "step": 20356
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.740337209681545e-07,
+      "loss": 0.479,
+      "step": 20357
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7359596804636086e-07,
+      "loss": 0.4446,
+      "step": 20358
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.731584124400035e-07,
+      "loss": 0.4182,
+      "step": 20359
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.727210541581428e-07,
+      "loss": 0.4985,
+      "step": 20360
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.722838932098406e-07,
+      "loss": 0.5559,
+      "step": 20361
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.718469296041484e-07,
+      "loss": 0.4714,
+      "step": 20362
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7141016335012003e-07,
+      "loss": 0.4509,
+      "step": 20363
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7097359445679836e-07,
+      "loss": 0.4058,
+      "step": 20364
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7053722293322947e-07,
+      "loss": 0.4778,
+      "step": 20365
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7010104878844945e-07,
+      "loss": 0.4807,
+      "step": 20366
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.696650720314921e-07,
+      "loss": 0.4438,
+      "step": 20367
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.692292926713904e-07,
+      "loss": 0.4353,
+      "step": 20368
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.687937107171658e-07,
+      "loss": 0.4341,
+      "step": 20369
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.683583261778457e-07,
+      "loss": 0.4688,
+      "step": 20370
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.679231390624428e-07,
+      "loss": 0.4019,
+      "step": 20371
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6748814937997545e-07,
+      "loss": 0.5137,
+      "step": 20372
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.670533571394509e-07,
+      "loss": 0.4399,
+      "step": 20373
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6661876234987526e-07,
+      "loss": 0.5005,
+      "step": 20374
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6618436502024913e-07,
+      "loss": 0.4395,
+      "step": 20375
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6575016515957086e-07,
+      "loss": 0.4829,
+      "step": 20376
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6531616277683545e-07,
+      "loss": 0.4717,
+      "step": 20377
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6488235788102797e-07,
+      "loss": 0.4629,
+      "step": 20378
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.644487504811379e-07,
+      "loss": 0.4771,
+      "step": 20379
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.640153405861447e-07,
+      "loss": 0.4895,
+      "step": 20380
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6358212820502566e-07,
+      "loss": 0.4829,
+      "step": 20381
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.631491133467514e-07,
+      "loss": 0.416,
+      "step": 20382
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6271629602029466e-07,
+      "loss": 0.4578,
+      "step": 20383
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6228367623461614e-07,
+      "loss": 0.4316,
+      "step": 20384
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.618512539986797e-07,
+      "loss": 0.3896,
+      "step": 20385
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6141902932144046e-07,
+      "loss": 0.4329,
+      "step": 20386
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6098700221185123e-07,
+      "loss": 0.4167,
+      "step": 20387
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.605551726788582e-07,
+      "loss": 0.3916,
+      "step": 20388
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6012354073140977e-07,
+      "loss": 0.4639,
+      "step": 20389
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.596921063784432e-07,
+      "loss": 0.498,
+      "step": 20390
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5926086962889584e-07,
+      "loss": 0.3938,
+      "step": 20391
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.588298304916972e-07,
+      "loss": 0.4714,
+      "step": 20392
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5839898897577895e-07,
+      "loss": 0.4604,
+      "step": 20393
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5796834509006183e-07,
+      "loss": 0.4131,
+      "step": 20394
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.575378988434676e-07,
+      "loss": 0.4261,
+      "step": 20395
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.571076502449112e-07,
+      "loss": 0.4971,
+      "step": 20396
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5667759930330234e-07,
+      "loss": 0.4102,
+      "step": 20397
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.562477460275516e-07,
+      "loss": 0.4456,
+      "step": 20398
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5581809042655967e-07,
+      "loss": 0.5227,
+      "step": 20399
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5538863250922606e-07,
+      "loss": 0.4714,
+      "step": 20400
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.549593722844492e-07,
+      "loss": 0.4451,
+      "step": 20401
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.545303097611153e-07,
+      "loss": 0.551,
+      "step": 20402
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.54101444948114e-07,
+      "loss": 0.438,
+      "step": 20403
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5367277785432815e-07,
+      "loss": 0.4436,
+      "step": 20404
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.532443084886362e-07,
+      "loss": 0.4758,
+      "step": 20405
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5281603685991327e-07,
+      "loss": 0.4248,
+      "step": 20406
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.523879629770278e-07,
+      "loss": 0.3979,
+      "step": 20407
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.519600868488505e-07,
+      "loss": 0.4058,
+      "step": 20408
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5153240848423875e-07,
+      "loss": 0.4348,
+      "step": 20409
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5110492789205427e-07,
+      "loss": 0.4333,
+      "step": 20410
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.506776450811512e-07,
+      "loss": 0.4053,
+      "step": 20411
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5025056006037684e-07,
+      "loss": 0.4321,
+      "step": 20412
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4982367283857963e-07,
+      "loss": 0.4119,
+      "step": 20413
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4939698342460036e-07,
+      "loss": 0.3916,
+      "step": 20414
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.489704918272775e-07,
+      "loss": 0.3899,
+      "step": 20415
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.485441980554439e-07,
+      "loss": 0.5024,
+      "step": 20416
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.481181021179304e-07,
+      "loss": 0.4353,
+      "step": 20417
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4769220402356097e-07,
+      "loss": 0.3772,
+      "step": 20418
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4726650378115965e-07,
+      "loss": 0.4475,
+      "step": 20419
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4684100139953943e-07,
+      "loss": 0.4543,
+      "step": 20420
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4641569688751885e-07,
+      "loss": 0.4006,
+      "step": 20421
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.459905902539019e-07,
+      "loss": 0.417,
+      "step": 20422
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4556568150749713e-07,
+      "loss": 0.488,
+      "step": 20423
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.451409706571042e-07,
+      "loss": 0.4146,
+      "step": 20424
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4471645771152037e-07,
+      "loss": 0.4727,
+      "step": 20425
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.442921426795366e-07,
+      "loss": 0.3965,
+      "step": 20426
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.438680255699435e-07,
+      "loss": 0.4033,
+      "step": 20427
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.434441063915251e-07,
+      "loss": 0.4207,
+      "step": 20428
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.430203851530612e-07,
+      "loss": 0.5107,
+      "step": 20429
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.425968618633292e-07,
+      "loss": 0.4436,
+      "step": 20430
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4217353653109973e-07,
+      "loss": 0.4441,
+      "step": 20431
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.417504091651437e-07,
+      "loss": 0.4495,
+      "step": 20432
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4132747977422174e-07,
+      "loss": 0.4138,
+      "step": 20433
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.409047483670958e-07,
+      "loss": 0.4456,
+      "step": 20434
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.404822149525212e-07,
+      "loss": 0.4155,
+      "step": 20435
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.400598795392497e-07,
+      "loss": 0.4773,
+      "step": 20436
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.396377421360287e-07,
+      "loss": 0.4292,
+      "step": 20437
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3921580275160246e-07,
+      "loss": 0.4187,
+      "step": 20438
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.387940613947095e-07,
+      "loss": 0.4187,
+      "step": 20439
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.383725180740861e-07,
+      "loss": 0.5059,
+      "step": 20440
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.37951172798462e-07,
+      "loss": 0.4541,
+      "step": 20441
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3753002557656465e-07,
+      "loss": 0.4368,
+      "step": 20442
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.371090764171193e-07,
+      "loss": 0.428,
+      "step": 20443
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.366883253288412e-07,
+      "loss": 0.4553,
+      "step": 20444
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.362677723204489e-07,
+      "loss": 0.4802,
+      "step": 20445
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3584741740064884e-07,
+      "loss": 0.5447,
+      "step": 20446
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3542726057815175e-07,
+      "loss": 0.489,
+      "step": 20447
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3500730186165743e-07,
+      "loss": 0.4375,
+      "step": 20448
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.345875412598666e-07,
+      "loss": 0.4233,
+      "step": 20449
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.341679787814712e-07,
+      "loss": 0.4866,
+      "step": 20450
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3374861443516215e-07,
+      "loss": 0.4229,
+      "step": 20451
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3332944822962575e-07,
+      "loss": 0.45,
+      "step": 20452
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.32910480173544e-07,
+      "loss": 0.4719,
+      "step": 20453
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3249171027559544e-07,
+      "loss": 0.4314,
+      "step": 20454
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.320731385444521e-07,
+      "loss": 0.4302,
+      "step": 20455
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3165476498878476e-07,
+      "loss": 0.4263,
+      "step": 20456
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.312365896172588e-07,
+      "loss": 0.4504,
+      "step": 20457
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.3081861243853716e-07,
+      "loss": 0.5042,
+      "step": 20458
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.304008334612741e-07,
+      "loss": 0.4075,
+      "step": 20459
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.299832526941261e-07,
+      "loss": 0.4856,
+      "step": 20460
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2956587014573837e-07,
+      "loss": 0.4277,
+      "step": 20461
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2914868582476064e-07,
+      "loss": 0.4282,
+      "step": 20462
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.287316997398294e-07,
+      "loss": 0.45,
+      "step": 20463
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.283149118995855e-07,
+      "loss": 0.4978,
+      "step": 20464
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.278983223126587e-07,
+      "loss": 0.4233,
+      "step": 20465
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2748193098767877e-07,
+      "loss": 0.4104,
+      "step": 20466
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2706573793326876e-07,
+      "loss": 0.4697,
+      "step": 20467
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2664974315805295e-07,
+      "loss": 0.4473,
+      "step": 20468
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.262339466706422e-07,
+      "loss": 0.4514,
+      "step": 20469
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.258183484796541e-07,
+      "loss": 0.4053,
+      "step": 20470
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.254029485936928e-07,
+      "loss": 0.4788,
+      "step": 20471
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2498774702136257e-07,
+      "loss": 0.4614,
+      "step": 20472
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.245727437712666e-07,
+      "loss": 0.4473,
+      "step": 20473
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2415793885199565e-07,
+      "loss": 0.4268,
+      "step": 20474
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2374333227214626e-07,
+      "loss": 0.5095,
+      "step": 20475
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2332892404030154e-07,
+      "loss": 0.4949,
+      "step": 20476
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2291471416504804e-07,
+      "loss": 0.4634,
+      "step": 20477
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2250070265496435e-07,
+      "loss": 0.4797,
+      "step": 20478
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2208688951862475e-07,
+      "loss": 0.4543,
+      "step": 20479
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.216732747646013e-07,
+      "loss": 0.4014,
+      "step": 20480
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2125985840146155e-07,
+      "loss": 0.4768,
+      "step": 20481
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2084664043776537e-07,
+      "loss": 0.3926,
+      "step": 20482
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2043362088207583e-07,
+      "loss": 0.4788,
+      "step": 20483
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2002079974294394e-07,
+      "loss": 0.4495,
+      "step": 20484
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.19608177028924e-07,
+      "loss": 0.4587,
+      "step": 20485
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1919575274855906e-07,
+      "loss": 0.4441,
+      "step": 20486
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1878352691039125e-07,
+      "loss": 0.5146,
+      "step": 20487
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1837149952296263e-07,
+      "loss": 0.4209,
+      "step": 20488
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1795967059480193e-07,
+      "loss": 0.3896,
+      "step": 20489
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.175480401344445e-07,
+      "loss": 0.4475,
+      "step": 20490
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1713660815041137e-07,
+      "loss": 0.427,
+      "step": 20491
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.16725374651229e-07,
+      "loss": 0.5159,
+      "step": 20492
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.163143396454117e-07,
+      "loss": 0.3938,
+      "step": 20493
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.159035031414738e-07,
+      "loss": 0.439,
+      "step": 20494
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.154928651479251e-07,
+      "loss": 0.429,
+      "step": 20495
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1508242567327106e-07,
+      "loss": 0.4258,
+      "step": 20496
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1467218472601267e-07,
+      "loss": 0.4832,
+      "step": 20497
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.142621423146476e-07,
+      "loss": 0.4067,
+      "step": 20498
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1385229844766786e-07,
+      "loss": 0.4114,
+      "step": 20499
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1344265313356224e-07,
+      "loss": 0.4194,
+      "step": 20500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1303320638081624e-07,
+      "loss": 0.478,
+      "step": 20501
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.126239581979108e-07,
+      "loss": 0.4644,
+      "step": 20502
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1221490859332024e-07,
+      "loss": 0.4556,
+      "step": 20503
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1180605757552116e-07,
+      "loss": 0.3557,
+      "step": 20504
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1139740515297786e-07,
+      "loss": 0.4104,
+      "step": 20505
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1098895133415583e-07,
+      "loss": 0.437,
+      "step": 20506
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1058069612751716e-07,
+      "loss": 0.4189,
+      "step": 20507
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.10172639541514e-07,
+      "loss": 0.4343,
+      "step": 20508
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.097647815846029e-07,
+      "loss": 0.45,
+      "step": 20509
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.093571222652271e-07,
+      "loss": 0.4299,
+      "step": 20510
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0894966159183315e-07,
+      "loss": 0.4536,
+      "step": 20511
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.085423995728588e-07,
+      "loss": 0.4236,
+      "step": 20512
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.081353362167406e-07,
+      "loss": 0.4263,
+      "step": 20513
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.077284715319108e-07,
+      "loss": 0.4749,
+      "step": 20514
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0732180552679357e-07,
+      "loss": 0.4773,
+      "step": 20515
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0691533820981453e-07,
+      "loss": 0.4443,
+      "step": 20516
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.065090695893903e-07,
+      "loss": 0.4636,
+      "step": 20517
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0610299967393964e-07,
+      "loss": 0.4529,
+      "step": 20518
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.05697128471868e-07,
+      "loss": 0.3806,
+      "step": 20519
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0529145599158657e-07,
+      "loss": 0.4795,
+      "step": 20520
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0488598224149524e-07,
+      "loss": 0.4861,
+      "step": 20521
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.04480707229995e-07,
+      "loss": 0.4573,
+      "step": 20522
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0407563096547587e-07,
+      "loss": 0.405,
+      "step": 20523
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.036707534563322e-07,
+      "loss": 0.4348,
+      "step": 20524
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0326607471094623e-07,
+      "loss": 0.4187,
+      "step": 20525
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.028615947377035e-07,
+      "loss": 0.4089,
+      "step": 20526
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.024573135449794e-07,
+      "loss": 0.4136,
+      "step": 20527
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0205323114114957e-07,
+      "loss": 0.4241,
+      "step": 20528
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.016493475345806e-07,
+      "loss": 0.4988,
+      "step": 20529
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.012456627336392e-07,
+      "loss": 0.4685,
+      "step": 20530
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.008421767466886e-07,
+      "loss": 0.418,
+      "step": 20531
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.004388895820843e-07,
+      "loss": 0.4268,
+      "step": 20532
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0003580124817974e-07,
+      "loss": 0.4172,
+      "step": 20533
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9963291175332263e-07,
+      "loss": 0.4128,
+      "step": 20534
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9923022110585965e-07,
+      "loss": 0.4561,
+      "step": 20535
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9882772931413074e-07,
+      "loss": 0.4478,
+      "step": 20536
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9842543638647256e-07,
+      "loss": 0.4197,
+      "step": 20537
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.980233423312174e-07,
+      "loss": 0.4031,
+      "step": 20538
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9762144715669415e-07,
+      "loss": 0.4277,
+      "step": 20539
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.972197508712261e-07,
+      "loss": 0.4646,
+      "step": 20540
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.968182534831355e-07,
+      "loss": 0.4895,
+      "step": 20541
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.964169550007346e-07,
+      "loss": 0.4949,
+      "step": 20542
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.96015855432339e-07,
+      "loss": 0.4819,
+      "step": 20543
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.956149547862553e-07,
+      "loss": 0.4658,
+      "step": 20544
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.952142530707859e-07,
+      "loss": 0.4365,
+      "step": 20545
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9481375029423174e-07,
+      "loss": 0.4224,
+      "step": 20546
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.944134464648863e-07,
+      "loss": 0.4939,
+      "step": 20547
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.940133415910441e-07,
+      "loss": 0.4663,
+      "step": 20548
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.936134356809895e-07,
+      "loss": 0.3855,
+      "step": 20549
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.932137287430071e-07,
+      "loss": 0.5498,
+      "step": 20550
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9281422078537466e-07,
+      "loss": 0.5129,
+      "step": 20551
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.924149118163689e-07,
+      "loss": 0.3933,
+      "step": 20552
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.920158018442566e-07,
+      "loss": 0.4397,
+      "step": 20553
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.916168908773088e-07,
+      "loss": 0.5054,
+      "step": 20554
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9121817892378566e-07,
+      "loss": 0.4033,
+      "step": 20555
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.908196659919461e-07,
+      "loss": 0.3875,
+      "step": 20556
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.904213520900424e-07,
+      "loss": 0.4211,
+      "step": 20557
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9002323722632904e-07,
+      "loss": 0.4702,
+      "step": 20558
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.896253214090484e-07,
+      "loss": 0.4253,
+      "step": 20559
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.892276046464416e-07,
+      "loss": 0.4177,
+      "step": 20560
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8883008694674986e-07,
+      "loss": 0.4229,
+      "step": 20561
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.884327683182032e-07,
+      "loss": 0.3962,
+      "step": 20562
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.88035648769034e-07,
+      "loss": 0.4758,
+      "step": 20563
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8763872830746564e-07,
+      "loss": 0.4448,
+      "step": 20564
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.872420069417204e-07,
+      "loss": 0.4648,
+      "step": 20565
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.868454846800129e-07,
+      "loss": 0.4656,
+      "step": 20566
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.86449161530561e-07,
+      "loss": 0.4685,
+      "step": 20567
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8605303750156807e-07,
+      "loss": 0.4763,
+      "step": 20568
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.856571126012432e-07,
+      "loss": 0.4307,
+      "step": 20569
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8526138683778414e-07,
+      "loss": 0.4028,
+      "step": 20570
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.848658602193889e-07,
+      "loss": 0.5112,
+      "step": 20571
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.844705327542486e-07,
+      "loss": 0.4104,
+      "step": 20572
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.840754044505524e-07,
+      "loss": 0.4502,
+      "step": 20573
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.836804753164847e-07,
+      "loss": 0.4326,
+      "step": 20574
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8328574536022454e-07,
+      "loss": 0.3896,
+      "step": 20575
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.828912145899477e-07,
+      "loss": 0.437,
+      "step": 20576
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.824968830138265e-07,
+      "loss": 0.4429,
+      "step": 20577
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8210275064002766e-07,
+      "loss": 0.4165,
+      "step": 20578
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8170881747671585e-07,
+      "loss": 0.426,
+      "step": 20579
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8131508353205007e-07,
+      "loss": 0.4226,
+      "step": 20580
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8092154881418485e-07,
+      "loss": 0.4768,
+      "step": 20581
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.805282133312715e-07,
+      "loss": 0.4836,
+      "step": 20582
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.801350770914569e-07,
+      "loss": 0.4531,
+      "step": 20583
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7974214010288556e-07,
+      "loss": 0.4546,
+      "step": 20584
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.793494023736932e-07,
+      "loss": 0.4365,
+      "step": 20585
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.789568639120178e-07,
+      "loss": 0.4036,
+      "step": 20586
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7856452472598614e-07,
+      "loss": 0.5034,
+      "step": 20587
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7817238482372845e-07,
+      "loss": 0.4712,
+      "step": 20588
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.777804442133637e-07,
+      "loss": 0.4287,
+      "step": 20589
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.773887029030099e-07,
+      "loss": 0.4507,
+      "step": 20590
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.769971609007839e-07,
+      "loss": 0.4905,
+      "step": 20591
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7660581821479247e-07,
+      "loss": 0.4434,
+      "step": 20592
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.762146748531437e-07,
+      "loss": 0.4507,
+      "step": 20593
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7582373082393766e-07,
+      "loss": 0.4592,
+      "step": 20594
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.754329861352723e-07,
+      "loss": 0.4531,
+      "step": 20595
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7504244079524023e-07,
+      "loss": 0.3911,
+      "step": 20596
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7465209481193143e-07,
+      "loss": 0.5444,
+      "step": 20597
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7426194819342955e-07,
+      "loss": 0.4048,
+      "step": 20598
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.738720009478192e-07,
+      "loss": 0.3818,
+      "step": 20599
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7348225308317165e-07,
+      "loss": 0.4258,
+      "step": 20600
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7309270460756496e-07,
+      "loss": 0.4209,
+      "step": 20601
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.727033555290627e-07,
+      "loss": 0.4365,
+      "step": 20602
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7231420585573387e-07,
+      "loss": 0.4712,
+      "step": 20603
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.719252555956343e-07,
+      "loss": 0.4039,
+      "step": 20604
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7153650475682423e-07,
+      "loss": 0.4463,
+      "step": 20605
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7114795334735167e-07,
+      "loss": 0.4116,
+      "step": 20606
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.707596013752679e-07,
+      "loss": 0.4946,
+      "step": 20607
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7037144884861543e-07,
+      "loss": 0.4731,
+      "step": 20608
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6998349577543337e-07,
+      "loss": 0.5193,
+      "step": 20609
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.695957421637586e-07,
+      "loss": 0.5098,
+      "step": 20610
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.692081880216192e-07,
+      "loss": 0.4607,
+      "step": 20611
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6882083335704757e-07,
+      "loss": 0.4468,
+      "step": 20612
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6843367817806176e-07,
+      "loss": 0.4402,
+      "step": 20613
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6804672249268316e-07,
+      "loss": 0.4631,
+      "step": 20614
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.676599663089253e-07,
+      "loss": 0.4487,
+      "step": 20615
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6727340963480186e-07,
+      "loss": 0.4651,
+      "step": 20616
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6688705247831524e-07,
+      "loss": 0.4277,
+      "step": 20617
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.665008948474713e-07,
+      "loss": 0.4307,
+      "step": 20618
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.66114936750267e-07,
+      "loss": 0.469,
+      "step": 20619
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.657291781946959e-07,
+      "loss": 0.4739,
+      "step": 20620
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.653436191887494e-07,
+      "loss": 0.4817,
+      "step": 20621
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6495825974040997e-07,
+      "loss": 0.4331,
+      "step": 20622
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6457309985766465e-07,
+      "loss": 0.3784,
+      "step": 20623
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6418813954848587e-07,
+      "loss": 0.4326,
+      "step": 20624
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6380337882085173e-07,
+      "loss": 0.458,
+      "step": 20625
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6341881768272804e-07,
+      "loss": 0.4507,
+      "step": 20626
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.630344561420818e-07,
+      "loss": 0.4189,
+      "step": 20627
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6265029420687323e-07,
+      "loss": 0.4783,
+      "step": 20628
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6226633188506055e-07,
+      "loss": 0.458,
+      "step": 20629
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6188256918459507e-07,
+      "loss": 0.4573,
+      "step": 20630
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.614990061134249e-07,
+      "loss": 0.4844,
+      "step": 20631
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6111564267949816e-07,
+      "loss": 0.5012,
+      "step": 20632
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6073247889075183e-07,
+      "loss": 0.4192,
+      "step": 20633
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.603495147551239e-07,
+      "loss": 0.4683,
+      "step": 20634
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.5996675028054375e-07,
+      "loss": 0.4648,
+      "step": 20635
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.595841854749438e-07,
+      "loss": 0.4946,
+      "step": 20636
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.592018203462444e-07,
+      "loss": 0.406,
+      "step": 20637
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.58819654902367e-07,
+      "loss": 0.4163,
+      "step": 20638
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5843768915122646e-07,
+      "loss": 0.4104,
+      "step": 20639
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5805592310073524e-07,
+      "loss": 0.4275,
+      "step": 20640
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5767435675879813e-07,
+      "loss": 0.4487,
+      "step": 20641
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5729299013332107e-07,
+      "loss": 0.4153,
+      "step": 20642
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.569118232322022e-07,
+      "loss": 0.4573,
+      "step": 20643
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5653085606333627e-07,
+      "loss": 0.4497,
+      "step": 20644
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.561500886346136e-07,
+      "loss": 0.4048,
+      "step": 20645
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.557695209539225e-07,
+      "loss": 0.4817,
+      "step": 20646
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5538915302914313e-07,
+      "loss": 0.4114,
+      "step": 20647
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5500898486815596e-07,
+      "loss": 0.4602,
+      "step": 20648
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.546290164788335e-07,
+      "loss": 0.439,
+      "step": 20649
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5424924786904623e-07,
+      "loss": 0.427,
+      "step": 20650
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5386967904666223e-07,
+      "loss": 0.5059,
+      "step": 20651
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.534903100195397e-07,
+      "loss": 0.4595,
+      "step": 20652
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.53111140795539e-07,
+      "loss": 0.3977,
+      "step": 20653
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.527321713825127e-07,
+      "loss": 0.4351,
+      "step": 20654
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.52353401788309e-07,
+      "loss": 0.4297,
+      "step": 20655
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.519748320207772e-07,
+      "loss": 0.4429,
+      "step": 20656
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.515964620877532e-07,
+      "loss": 0.3657,
+      "step": 20657
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5121829199707857e-07,
+      "loss": 0.4885,
+      "step": 20658
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5084032175658257e-07,
+      "loss": 0.439,
+      "step": 20659
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5046255137409556e-07,
+      "loss": 0.4401,
+      "step": 20660
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.500849808574425e-07,
+      "loss": 0.5183,
+      "step": 20661
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.497076102144426e-07,
+      "loss": 0.5425,
+      "step": 20662
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.49330439452914e-07,
+      "loss": 0.397,
+      "step": 20663
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4895346858066723e-07,
+      "loss": 0.4709,
+      "step": 20664
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.485766976055105e-07,
+      "loss": 0.399,
+      "step": 20665
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4820012653524747e-07,
+      "loss": 0.4684,
+      "step": 20666
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4782375537767867e-07,
+      "loss": 0.4585,
+      "step": 20667
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4744758414059887e-07,
+      "loss": 0.4443,
+      "step": 20668
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4707161283179966e-07,
+      "loss": 0.4158,
+      "step": 20669
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.466958414590704e-07,
+      "loss": 0.4221,
+      "step": 20670
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4632027003019044e-07,
+      "loss": 0.4126,
+      "step": 20671
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4594489855294233e-07,
+      "loss": 0.4517,
+      "step": 20672
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.455697270350977e-07,
+      "loss": 0.4541,
+      "step": 20673
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.451947554844304e-07,
+      "loss": 0.4885,
+      "step": 20674
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4481998390870407e-07,
+      "loss": 0.502,
+      "step": 20675
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4444541231568375e-07,
+      "loss": 0.4294,
+      "step": 20676
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4407104071312646e-07,
+      "loss": 0.4902,
+      "step": 20677
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.436968691087872e-07,
+      "loss": 0.4069,
+      "step": 20678
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4332289751041523e-07,
+      "loss": 0.4487,
+      "step": 20679
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.429491259257556e-07,
+      "loss": 0.4912,
+      "step": 20680
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4257555436255086e-07,
+      "loss": 0.4365,
+      "step": 20681
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.422021828285405e-07,
+      "loss": 0.4199,
+      "step": 20682
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4182901133145484e-07,
+      "loss": 0.4209,
+      "step": 20683
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4145603987902455e-07,
+      "loss": 0.4863,
+      "step": 20684
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4108326847897556e-07,
+      "loss": 0.4626,
+      "step": 20685
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4071069713902725e-07,
+      "loss": 0.4138,
+      "step": 20686
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.40338325866898e-07,
+      "loss": 0.4336,
+      "step": 20687
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.399661546703004e-07,
+      "loss": 0.459,
+      "step": 20688
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3959418355694165e-07,
+      "loss": 0.4478,
+      "step": 20689
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.392224125345278e-07,
+      "loss": 0.4321,
+      "step": 20690
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.388508416107583e-07,
+      "loss": 0.3779,
+      "step": 20691
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.384794707933292e-07,
+      "loss": 0.4409,
+      "step": 20692
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.381083000899332e-07,
+      "loss": 0.4092,
+      "step": 20693
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.377373295082587e-07,
+      "loss": 0.5083,
+      "step": 20694
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3736655905598604e-07,
+      "loss": 0.46,
+      "step": 20695
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.369959887407992e-07,
+      "loss": 0.4375,
+      "step": 20696
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.366256185703709e-07,
+      "loss": 0.4305,
+      "step": 20697
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3625544855237393e-07,
+      "loss": 0.4448,
+      "step": 20698
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.358854786944732e-07,
+      "loss": 0.4454,
+      "step": 20699
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.355157090043337e-07,
+      "loss": 0.4478,
+      "step": 20700
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3514613948961385e-07,
+      "loss": 0.5242,
+      "step": 20701
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.347767701579685e-07,
+      "loss": 0.4534,
+      "step": 20702
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3440760101704714e-07,
+      "loss": 0.4298,
+      "step": 20703
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3403863207449814e-07,
+      "loss": 0.4485,
+      "step": 20704
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3366986333796205e-07,
+      "loss": 0.4788,
+      "step": 20705
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3330129481507824e-07,
+      "loss": 0.4883,
+      "step": 20706
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.329329265134784e-07,
+      "loss": 0.3926,
+      "step": 20707
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3256475844079426e-07,
+      "loss": 0.4504,
+      "step": 20708
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3219679060465195e-07,
+      "loss": 0.4817,
+      "step": 20709
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3182902301267196e-07,
+      "loss": 0.4854,
+      "step": 20710
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3146145567247154e-07,
+      "loss": 0.4683,
+      "step": 20711
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.310940885916647e-07,
+      "loss": 0.5085,
+      "step": 20712
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.307269217778608e-07,
+      "loss": 0.4243,
+      "step": 20713
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.303599552386627e-07,
+      "loss": 0.438,
+      "step": 20714
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.299931889816732e-07,
+      "loss": 0.448,
+      "step": 20715
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2962662301448845e-07,
+      "loss": 0.4198,
+      "step": 20716
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2926025734470123e-07,
+      "loss": 0.4204,
+      "step": 20717
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2889409197989773e-07,
+      "loss": 0.4175,
+      "step": 20718
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.285281269276652e-07,
+      "loss": 0.4563,
+      "step": 20719
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2816236219558207e-07,
+      "loss": 0.4229,
+      "step": 20720
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2779679779122553e-07,
+      "loss": 0.4646,
+      "step": 20721
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.274314337221651e-07,
+      "loss": 0.4399,
+      "step": 20722
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2706626999596923e-07,
+      "loss": 0.4653,
+      "step": 20723
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2670130662020293e-07,
+      "loss": 0.4885,
+      "step": 20724
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.263365436024235e-07,
+      "loss": 0.4161,
+      "step": 20725
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.259719809501882e-07,
+      "loss": 0.4199,
+      "step": 20726
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.256076186710444e-07,
+      "loss": 0.4229,
+      "step": 20727
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2524345677254266e-07,
+      "loss": 0.5437,
+      "step": 20728
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.248794952622236e-07,
+      "loss": 0.4272,
+      "step": 20729
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.245157341476279e-07,
+      "loss": 0.4744,
+      "step": 20730
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2415217343628623e-07,
+      "loss": 0.4714,
+      "step": 20731
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.237888131357314e-07,
+      "loss": 0.4851,
+      "step": 20732
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2342565325349074e-07,
+      "loss": 0.4697,
+      "step": 20733
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.230626937970838e-07,
+      "loss": 0.449,
+      "step": 20734
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2269993477403006e-07,
+      "loss": 0.4136,
+      "step": 20735
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2233737619184023e-07,
+      "loss": 0.418,
+      "step": 20736
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2197501805802833e-07,
+      "loss": 0.4255,
+      "step": 20737
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.21612860380095e-07,
+      "loss": 0.364,
+      "step": 20738
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.212509031655453e-07,
+      "loss": 0.4041,
+      "step": 20739
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.208891464218733e-07,
+      "loss": 0.4441,
+      "step": 20740
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.205275901565752e-07,
+      "loss": 0.4294,
+      "step": 20741
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.201662343771361e-07,
+      "loss": 0.4177,
+      "step": 20742
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.198050790910423e-07,
+      "loss": 0.4817,
+      "step": 20743
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1944412430577443e-07,
+      "loss": 0.4641,
+      "step": 20744
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.190833700288087e-07,
+      "loss": 0.4705,
+      "step": 20745
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.187228162676159e-07,
+      "loss": 0.4717,
+      "step": 20746
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.183624630296667e-07,
+      "loss": 0.4543,
+      "step": 20747
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.180023103224217e-07,
+      "loss": 0.4805,
+      "step": 20748
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.176423581533428e-07,
+      "loss": 0.4631,
+      "step": 20749
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.172826065298851e-07,
+      "loss": 0.4841,
+      "step": 20750
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.169230554594993e-07,
+      "loss": 0.4885,
+      "step": 20751
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1656370494963396e-07,
+      "loss": 0.4597,
+      "step": 20752
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1620455500772975e-07,
+      "loss": 0.4226,
+      "step": 20753
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1584560564122734e-07,
+      "loss": 0.5156,
+      "step": 20754
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.154868568575609e-07,
+      "loss": 0.4456,
+      "step": 20755
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1512830866416213e-07,
+      "loss": 0.4673,
+      "step": 20756
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1476996106845516e-07,
+      "loss": 0.5232,
+      "step": 20757
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.144118140778629e-07,
+      "loss": 0.4175,
+      "step": 20758
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1405386769980616e-07,
+      "loss": 0.4102,
+      "step": 20759
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.136961219416956e-07,
+      "loss": 0.3582,
+      "step": 20760
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1333857681094203e-07,
+      "loss": 0.4568,
+      "step": 20761
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.129812323149528e-07,
+      "loss": 0.4326,
+      "step": 20762
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1262408846112644e-07,
+      "loss": 0.3831,
+      "step": 20763
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1226714525686374e-07,
+      "loss": 0.4106,
+      "step": 20764
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.119104027095543e-07,
+      "loss": 0.4487,
+      "step": 20765
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1155386082659e-07,
+      "loss": 0.3723,
+      "step": 20766
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1119751961535385e-07,
+      "loss": 0.4585,
+      "step": 20767
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.108413790832265e-07,
+      "loss": 0.4812,
+      "step": 20768
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1048543923758665e-07,
+      "loss": 0.4646,
+      "step": 20769
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.101297000858039e-07,
+      "loss": 0.469,
+      "step": 20770
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.09774161635249e-07,
+      "loss": 0.4233,
+      "step": 20771
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.094188238932838e-07,
+      "loss": 0.3823,
+      "step": 20772
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.090636868672703e-07,
+      "loss": 0.3994,
+      "step": 20773
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0870875056456253e-07,
+      "loss": 0.4141,
+      "step": 20774
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.083540149925135e-07,
+      "loss": 0.4639,
+      "step": 20775
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0799948015846845e-07,
+      "loss": 0.5264,
+      "step": 20776
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.076451460697749e-07,
+      "loss": 0.4351,
+      "step": 20777
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.072910127337669e-07,
+      "loss": 0.521,
+      "step": 20778
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0693708015778314e-07,
+      "loss": 0.5002,
+      "step": 20779
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0658334834915207e-07,
+      "loss": 0.4883,
+      "step": 20780
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.062298173152023e-07,
+      "loss": 0.3853,
+      "step": 20781
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0587648706325467e-07,
+      "loss": 0.4436,
+      "step": 20782
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0552335760063e-07,
+      "loss": 0.4895,
+      "step": 20783
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.051704289346391e-07,
+      "loss": 0.4392,
+      "step": 20784
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.048177010725928e-07,
+      "loss": 0.5212,
+      "step": 20785
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0446517402179965e-07,
+      "loss": 0.4612,
+      "step": 20786
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.041128477895594e-07,
+      "loss": 0.4612,
+      "step": 20787
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.037607223831696e-07,
+      "loss": 0.4409,
+      "step": 20788
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.034087978099232e-07,
+      "loss": 0.426,
+      "step": 20789
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0305707407711106e-07,
+      "loss": 0.4856,
+      "step": 20790
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.027055511920163e-07,
+      "loss": 0.4709,
+      "step": 20791
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0235422916192193e-07,
+      "loss": 0.4661,
+      "step": 20792
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0200310799410214e-07,
+      "loss": 0.4521,
+      "step": 20793
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0165218769583227e-07,
+      "loss": 0.4485,
+      "step": 20794
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0130146827437754e-07,
+      "loss": 0.4524,
+      "step": 20795
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0095094973700554e-07,
+      "loss": 0.4929,
+      "step": 20796
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.00600632090976e-07,
+      "loss": 0.458,
+      "step": 20797
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0025051534354086e-07,
+      "loss": 0.4373,
+      "step": 20798
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.999005995019577e-07,
+      "loss": 0.4417,
+      "step": 20799
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.995508845734685e-07,
+      "loss": 0.3912,
+      "step": 20800
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9920137056532183e-07,
+      "loss": 0.3945,
+      "step": 20801
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.988520574847531e-07,
+      "loss": 0.4246,
+      "step": 20802
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.985029453389998e-07,
+      "loss": 0.4604,
+      "step": 20803
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9815403413529067e-07,
+      "loss": 0.4417,
+      "step": 20804
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9780532388085536e-07,
+      "loss": 0.4604,
+      "step": 20805
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9745681458291484e-07,
+      "loss": 0.4341,
+      "step": 20806
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.971085062486878e-07,
+      "loss": 0.4229,
+      "step": 20807
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9676039888538843e-07,
+      "loss": 0.4438,
+      "step": 20808
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.964124925002265e-07,
+      "loss": 0.4109,
+      "step": 20809
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9606478710041075e-07,
+      "loss": 0.3779,
+      "step": 20810
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.957172826931387e-07,
+      "loss": 0.4829,
+      "step": 20811
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.953699792856113e-07,
+      "loss": 0.437,
+      "step": 20812
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9502287688502053e-07,
+      "loss": 0.3594,
+      "step": 20813
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.946759754985573e-07,
+      "loss": 0.4609,
+      "step": 20814
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.943292751334037e-07,
+      "loss": 0.4768,
+      "step": 20815
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.939827757967451e-07,
+      "loss": 0.4419,
+      "step": 20816
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.936364774957545e-07,
+      "loss": 0.397,
+      "step": 20817
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9329038023760635e-07,
+      "loss": 0.3599,
+      "step": 20818
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9294448402946816e-07,
+      "loss": 0.3708,
+      "step": 20819
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.925987888785065e-07,
+      "loss": 0.4822,
+      "step": 20820
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.922532947918788e-07,
+      "loss": 0.4211,
+      "step": 20821
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9190800177674396e-07,
+      "loss": 0.4805,
+      "step": 20822
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.915629098402517e-07,
+      "loss": 0.4165,
+      "step": 20823
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9121801898955084e-07,
+      "loss": 0.5039,
+      "step": 20824
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9087332923178333e-07,
+      "loss": 0.4341,
+      "step": 20825
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9052884057409026e-07,
+      "loss": 0.4636,
+      "step": 20826
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9018455302360693e-07,
+      "loss": 0.4208,
+      "step": 20827
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8984046658746214e-07,
+      "loss": 0.431,
+      "step": 20828
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8949658127278567e-07,
+      "loss": 0.4736,
+      "step": 20829
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8915289708669747e-07,
+      "loss": 0.4219,
+      "step": 20830
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8880941403631955e-07,
+      "loss": 0.4441,
+      "step": 20831
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8846613212876297e-07,
+      "loss": 0.3912,
+      "step": 20832
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8812305137113973e-07,
+      "loss": 0.3597,
+      "step": 20833
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8778017177055417e-07,
+      "loss": 0.4316,
+      "step": 20834
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8743749333411066e-07,
+      "loss": 0.4451,
+      "step": 20835
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8709501606890565e-07,
+      "loss": 0.4536,
+      "step": 20836
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8675273998203246e-07,
+      "loss": 0.468,
+      "step": 20837
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.864106650805798e-07,
+      "loss": 0.4399,
+      "step": 20838
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8606879137163536e-07,
+      "loss": 0.5654,
+      "step": 20839
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.857271188622779e-07,
+      "loss": 0.4226,
+      "step": 20840
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.853856475595851e-07,
+      "loss": 0.4607,
+      "step": 20841
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.850443774706302e-07,
+      "loss": 0.4766,
+      "step": 20842
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8470330860248086e-07,
+      "loss": 0.447,
+      "step": 20843
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8436244096220147e-07,
+      "loss": 0.4618,
+      "step": 20844
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8402177455685296e-07,
+      "loss": 0.3375,
+      "step": 20845
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.836813093934909e-07,
+      "loss": 0.4783,
+      "step": 20846
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.833410454791674e-07,
+      "loss": 0.4241,
+      "step": 20847
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8300098282093014e-07,
+      "loss": 0.4971,
+      "step": 20848
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8266112142582126e-07,
+      "loss": 0.4062,
+      "step": 20849
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.823214613008829e-07,
+      "loss": 0.4873,
+      "step": 20850
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.819820024531472e-07,
+      "loss": 0.4846,
+      "step": 20851
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.816427448896486e-07,
+      "loss": 0.4336,
+      "step": 20852
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.813036886174103e-07,
+      "loss": 0.4565,
+      "step": 20853
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8096483364345896e-07,
+      "loss": 0.4158,
+      "step": 20854
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.806261799748089e-07,
+      "loss": 0.4739,
+      "step": 20855
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.802877276184779e-07,
+      "loss": 0.4497,
+      "step": 20856
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7994947658147474e-07,
+      "loss": 0.4783,
+      "step": 20857
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.79611426870805e-07,
+      "loss": 0.4595,
+      "step": 20858
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.792735784934708e-07,
+      "loss": 0.4133,
+      "step": 20859
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.78935931456471e-07,
+      "loss": 0.469,
+      "step": 20860
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.785984857667967e-07,
+      "loss": 0.4509,
+      "step": 20861
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7826124143143894e-07,
+      "loss": 0.4055,
+      "step": 20862
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7792419845738326e-07,
+      "loss": 0.4155,
+      "step": 20863
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7758735685160966e-07,
+      "loss": 0.4771,
+      "step": 20864
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7725071662109583e-07,
+      "loss": 0.4216,
+      "step": 20865
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7691427777281286e-07,
+      "loss": 0.4041,
+      "step": 20866
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7657804031373194e-07,
+      "loss": 0.4136,
+      "step": 20867
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7624200425081403e-07,
+      "loss": 0.4543,
+      "step": 20868
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7590616959102255e-07,
+      "loss": 0.4902,
+      "step": 20869
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.755705363413108e-07,
+      "loss": 0.4751,
+      "step": 20870
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.752351045086321e-07,
+      "loss": 0.5007,
+      "step": 20871
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7489987409993424e-07,
+      "loss": 0.4036,
+      "step": 20872
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.745648451221594e-07,
+      "loss": 0.5125,
+      "step": 20873
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7423001758224764e-07,
+      "loss": 0.4086,
+      "step": 20874
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.738953914871334e-07,
+      "loss": 0.4482,
+      "step": 20875
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.735609668437489e-07,
+      "loss": 0.4136,
+      "step": 20876
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.732267436590197e-07,
+      "loss": 0.3855,
+      "step": 20877
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7289272193986916e-07,
+      "loss": 0.4873,
+      "step": 20878
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.725589016932151e-07,
+      "loss": 0.458,
+      "step": 20879
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7222528292597306e-07,
+      "loss": 0.4719,
+      "step": 20880
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7189186564505087e-07,
+      "loss": 0.4424,
+      "step": 20881
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.715586498573564e-07,
+      "loss": 0.4504,
+      "step": 20882
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7122563556978956e-07,
+      "loss": 0.3923,
+      "step": 20883
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7089282278925046e-07,
+      "loss": 0.3978,
+      "step": 20884
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.705602115226302e-07,
+      "loss": 0.4707,
+      "step": 20885
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7022780177681786e-07,
+      "loss": 0.4138,
+      "step": 20886
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.698955935587011e-07,
+      "loss": 0.4463,
+      "step": 20887
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.695635868751578e-07,
+      "loss": 0.5063,
+      "step": 20888
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.69231781733067e-07,
+      "loss": 0.4136,
+      "step": 20889
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.689001781392997e-07,
+      "loss": 0.481,
+      "step": 20890
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6856877610072496e-07,
+      "loss": 0.4258,
+      "step": 20891
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.682375756242062e-07,
+      "loss": 0.5339,
+      "step": 20892
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6790657671660447e-07,
+      "loss": 0.4302,
+      "step": 20893
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.675757793847744e-07,
+      "loss": 0.4585,
+      "step": 20894
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6724518363556827e-07,
+      "loss": 0.4214,
+      "step": 20895
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6691478947583284e-07,
+      "loss": 0.4641,
+      "step": 20896
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6658459691241366e-07,
+      "loss": 0.4204,
+      "step": 20897
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.662546059521465e-07,
+      "loss": 0.4607,
+      "step": 20898
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.659248166018691e-07,
+      "loss": 0.4154,
+      "step": 20899
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.655952288684094e-07,
+      "loss": 0.4375,
+      "step": 20900
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6526584275859746e-07,
+      "loss": 0.3696,
+      "step": 20901
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.649366582792534e-07,
+      "loss": 0.3918,
+      "step": 20902
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6460767543719624e-07,
+      "loss": 0.478,
+      "step": 20903
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6427889423923937e-07,
+      "loss": 0.4604,
+      "step": 20904
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.639503146921918e-07,
+      "loss": 0.4602,
+      "step": 20905
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.636219368028625e-07,
+      "loss": 0.4475,
+      "step": 20906
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6329376057804943e-07,
+      "loss": 0.408,
+      "step": 20907
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6296578602455156e-07,
+      "loss": 0.4634,
+      "step": 20908
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6263801314916236e-07,
+      "loss": 0.4324,
+      "step": 20909
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.623104419586708e-07,
+      "loss": 0.4539,
+      "step": 20910
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6198307245986043e-07,
+      "loss": 0.4087,
+      "step": 20911
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.616559046595124e-07,
+      "loss": 0.425,
+      "step": 20912
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6132893856440467e-07,
+      "loss": 0.4382,
+      "step": 20913
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6100217418130737e-07,
+      "loss": 0.4026,
+      "step": 20914
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.606756115169895e-07,
+      "loss": 0.3967,
+      "step": 20915
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.603492505782157e-07,
+      "loss": 0.4333,
+      "step": 20916
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.60023091371745e-07,
+      "loss": 0.4241,
+      "step": 20917
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5969713390433304e-07,
+      "loss": 0.4202,
+      "step": 20918
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5937137818273115e-07,
+      "loss": 0.5364,
+      "step": 20919
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5904582421368616e-07,
+      "loss": 0.4395,
+      "step": 20920
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5872047200394156e-07,
+      "loss": 0.3992,
+      "step": 20921
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.583953215602364e-07,
+      "loss": 0.4504,
+      "step": 20922
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5807037288930525e-07,
+      "loss": 0.439,
+      "step": 20923
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5774562599787835e-07,
+      "loss": 0.4478,
+      "step": 20924
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5742108089268245e-07,
+      "loss": 0.4893,
+      "step": 20925
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.57096737580439e-07,
+      "loss": 0.4224,
+      "step": 20926
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5677259606786686e-07,
+      "loss": 0.449,
+      "step": 20927
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.564486563616786e-07,
+      "loss": 0.4175,
+      "step": 20928
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5612491846858434e-07,
+      "loss": 0.4761,
+      "step": 20929
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5580138239529094e-07,
+      "loss": 0.4102,
+      "step": 20930
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5547804814849644e-07,
+      "loss": 0.4429,
+      "step": 20931
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.551549157349009e-07,
+      "loss": 0.4319,
+      "step": 20932
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.548319851611958e-07,
+      "loss": 0.4995,
+      "step": 20933
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.545092564340701e-07,
+      "loss": 0.4314,
+      "step": 20934
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.541867295602085e-07,
+      "loss": 0.5056,
+      "step": 20935
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5386440454629015e-07,
+      "loss": 0.472,
+      "step": 20936
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5354228139899407e-07,
+      "loss": 0.5037,
+      "step": 20937
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.532203601249883e-07,
+      "loss": 0.4634,
+      "step": 20938
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5289864073094304e-07,
+      "loss": 0.519,
+      "step": 20939
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.52577123223523e-07,
+      "loss": 0.4441,
+      "step": 20940
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5225580760938504e-07,
+      "loss": 0.4816,
+      "step": 20941
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.519346938951861e-07,
+      "loss": 0.4917,
+      "step": 20942
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5161378208757635e-07,
+      "loss": 0.4456,
+      "step": 20943
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5129307219320276e-07,
+      "loss": 0.457,
+      "step": 20944
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5097256421870884e-07,
+      "loss": 0.4573,
+      "step": 20945
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.506522581707316e-07,
+      "loss": 0.4421,
+      "step": 20946
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5033215405590785e-07,
+      "loss": 0.4797,
+      "step": 20947
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.500122518808645e-07,
+      "loss": 0.4521,
+      "step": 20948
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.496925516522297e-07,
+      "loss": 0.429,
+      "step": 20949
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4937305337662476e-07,
+      "loss": 0.4548,
+      "step": 20950
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.490537570606677e-07,
+      "loss": 0.488,
+      "step": 20951
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.487346627109699e-07,
+      "loss": 0.4673,
+      "step": 20952
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4841577033414386e-07,
+      "loss": 0.4187,
+      "step": 20953
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.480970799367921e-07,
+      "loss": 0.478,
+      "step": 20954
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4777859152551706e-07,
+      "loss": 0.4358,
+      "step": 20955
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.474603051069135e-07,
+      "loss": 0.5156,
+      "step": 20956
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.471422206875762e-07,
+      "loss": 0.4768,
+      "step": 20957
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.468243382740909e-07,
+      "loss": 0.5137,
+      "step": 20958
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4650665787304464e-07,
+      "loss": 0.4407,
+      "step": 20959
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.461891794910154e-07,
+      "loss": 0.4253,
+      "step": 20960
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.458719031345802e-07,
+      "loss": 0.4487,
+      "step": 20961
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.455548288103082e-07,
+      "loss": 0.4636,
+      "step": 20962
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.452379565247698e-07,
+      "loss": 0.4338,
+      "step": 20963
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4492128628452737e-07,
+      "loss": 0.4211,
+      "step": 20964
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.44604818096138e-07,
+      "loss": 0.4004,
+      "step": 20965
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.442885519661586e-07,
+      "loss": 0.4734,
+      "step": 20966
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4397248790113956e-07,
+      "loss": 0.4033,
+      "step": 20967
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.436566259076278e-07,
+      "loss": 0.5854,
+      "step": 20968
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4334096599216263e-07,
+      "loss": 0.427,
+      "step": 20969
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4302550816128644e-07,
+      "loss": 0.429,
+      "step": 20970
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.427102524215297e-07,
+      "loss": 0.4448,
+      "step": 20971
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4239519877942484e-07,
+      "loss": 0.4414,
+      "step": 20972
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4208034724149454e-07,
+      "loss": 0.4526,
+      "step": 20973
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.417656978142635e-07,
+      "loss": 0.468,
+      "step": 20974
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4145125050424545e-07,
+      "loss": 0.4263,
+      "step": 20975
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4113700531795514e-07,
+      "loss": 0.4089,
+      "step": 20976
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4082296226190185e-07,
+      "loss": 0.4766,
+      "step": 20977
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.405091213425892e-07,
+      "loss": 0.3662,
+      "step": 20978
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.401954825665187e-07,
+      "loss": 0.4451,
+      "step": 20979
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3988204594018403e-07,
+      "loss": 0.4402,
+      "step": 20980
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3956881147008115e-07,
+      "loss": 0.4509,
+      "step": 20981
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.392557791626937e-07,
+      "loss": 0.3887,
+      "step": 20982
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3894294902450985e-07,
+      "loss": 0.4497,
+      "step": 20983
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3863032106200447e-07,
+      "loss": 0.4639,
+      "step": 20984
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.383178952816567e-07,
+      "loss": 0.4049,
+      "step": 20985
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3800567168993482e-07,
+      "loss": 0.4277,
+      "step": 20986
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3769365029330804e-07,
+      "loss": 0.4919,
+      "step": 20987
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3738183109823676e-07,
+      "loss": 0.5083,
+      "step": 20988
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3707021411118136e-07,
+      "loss": 0.4058,
+      "step": 20989
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3675879933859668e-07,
+      "loss": 0.4192,
+      "step": 20990
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.364475867869309e-07,
+      "loss": 0.511,
+      "step": 20991
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.361365764626311e-07,
+      "loss": 0.4656,
+      "step": 20992
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3582576837213878e-07,
+      "loss": 0.4009,
+      "step": 20993
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.355151625218921e-07,
+      "loss": 0.4392,
+      "step": 20994
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3520475891832373e-07,
+      "loss": 0.4202,
+      "step": 20995
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3489455756786406e-07,
+      "loss": 0.3911,
+      "step": 20996
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3458455847693685e-07,
+      "loss": 0.4104,
+      "step": 20997
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3427476165196362e-07,
+      "loss": 0.4434,
+      "step": 20998
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3396516709936034e-07,
+      "loss": 0.4243,
+      "step": 20999
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3365577482553968e-07,
+      "loss": 0.4875,
+      "step": 21000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3334658483691096e-07,
+      "loss": 0.4253,
+      "step": 21001
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3303759713987795e-07,
+      "loss": 0.4595,
+      "step": 21002
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3272881174083883e-07,
+      "loss": 0.4487,
+      "step": 21003
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3242022864619185e-07,
+      "loss": 0.4387,
+      "step": 21004
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3211184786232632e-07,
+      "loss": 0.4771,
+      "step": 21005
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3180366939563047e-07,
+      "loss": 0.4346,
+      "step": 21006
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3149569325248921e-07,
+      "loss": 0.3877,
+      "step": 21007
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3118791943927744e-07,
+      "loss": 0.4375,
+      "step": 21008
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3088034796237446e-07,
+      "loss": 0.418,
+      "step": 21009
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3057297882814634e-07,
+      "loss": 0.5205,
+      "step": 21010
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.302658120429635e-07,
+      "loss": 0.5447,
+      "step": 21011
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.299588476131853e-07,
+      "loss": 0.4114,
+      "step": 21012
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2965208554517115e-07,
+      "loss": 0.4355,
+      "step": 21013
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2934552584527482e-07,
+      "loss": 0.4658,
+      "step": 21014
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2903916851984566e-07,
+      "loss": 0.4812,
+      "step": 21015
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2873301357522858e-07,
+      "loss": 0.4692,
+      "step": 21016
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2842706101776524e-07,
+      "loss": 0.4956,
+      "step": 21017
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2812131085379274e-07,
+      "loss": 0.4033,
+      "step": 21018
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2781576308964493e-07,
+      "loss": 0.439,
+      "step": 21019
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2751041773164894e-07,
+      "loss": 0.3662,
+      "step": 21020
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2720527478612976e-07,
+      "loss": 0.4949,
+      "step": 21021
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2690033425940783e-07,
+      "loss": 0.4246,
+      "step": 21022
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2659559615779813e-07,
+      "loss": 0.405,
+      "step": 21023
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.262910604876145e-07,
+      "loss": 0.4329,
+      "step": 21024
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.259867272551619e-07,
+      "loss": 0.4712,
+      "step": 21025
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2568259646674751e-07,
+      "loss": 0.4011,
+      "step": 21026
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2537866812866627e-07,
+      "loss": 0.4629,
+      "step": 21027
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.250749422472176e-07,
+      "loss": 0.3999,
+      "step": 21028
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.247714188286898e-07,
+      "loss": 0.4949,
+      "step": 21029
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2446809787937007e-07,
+      "loss": 0.4192,
+      "step": 21030
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2416497940554116e-07,
+      "loss": 0.4412,
+      "step": 21031
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2386206341348137e-07,
+      "loss": 0.4297,
+      "step": 21032
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.235593499094646e-07,
+      "loss": 0.4519,
+      "step": 21033
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2325683889976135e-07,
+      "loss": 0.4514,
+      "step": 21034
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2295453039063774e-07,
+      "loss": 0.4232,
+      "step": 21035
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2265242438835323e-07,
+      "loss": 0.4307,
+      "step": 21036
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.223505208991661e-07,
+      "loss": 0.4131,
+      "step": 21037
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2204881992933137e-07,
+      "loss": 0.4236,
+      "step": 21038
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2174732148509625e-07,
+      "loss": 0.4363,
+      "step": 21039
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2144602557270578e-07,
+      "loss": 0.4492,
+      "step": 21040
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.211449321984005e-07,
+      "loss": 0.4072,
+      "step": 21041
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2084404136841765e-07,
+      "loss": 0.4734,
+      "step": 21042
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2054335308898889e-07,
+      "loss": 0.4631,
+      "step": 21043
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2024286736634036e-07,
+      "loss": 0.4707,
+      "step": 21044
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1994258420669933e-07,
+      "loss": 0.4563,
+      "step": 21045
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1964250361628304e-07,
+      "loss": 0.4575,
+      "step": 21046
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.193426256013076e-07,
+      "loss": 0.4878,
+      "step": 21047
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1904295016798362e-07,
+      "loss": 0.4114,
+      "step": 21048
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1874347732251944e-07,
+      "loss": 0.468,
+      "step": 21049
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.184442070711179e-07,
+      "loss": 0.4243,
+      "step": 21050
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1814513941997516e-07,
+      "loss": 0.4463,
+      "step": 21051
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.178462743752885e-07,
+      "loss": 0.4298,
+      "step": 21052
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1754761194324515e-07,
+      "loss": 0.4546,
+      "step": 21053
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.172491521300346e-07,
+      "loss": 0.4822,
+      "step": 21054
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.169508949418364e-07,
+      "loss": 0.5127,
+      "step": 21055
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1665284038483002e-07,
+      "loss": 0.459,
+      "step": 21056
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.163549884651861e-07,
+      "loss": 0.4331,
+      "step": 21057
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1605733918907633e-07,
+      "loss": 0.4736,
+      "step": 21058
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1575989256266472e-07,
+      "loss": 0.4434,
+      "step": 21059
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1546264859211298e-07,
+      "loss": 0.4336,
+      "step": 21060
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1516560728357616e-07,
+      "loss": 0.4309,
+      "step": 21061
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1486876864320828e-07,
+      "loss": 0.4275,
+      "step": 21062
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.145721326771566e-07,
+      "loss": 0.4685,
+      "step": 21063
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1427569939156623e-07,
+      "loss": 0.4622,
+      "step": 21064
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1397946879257669e-07,
+      "loss": 0.4268,
+      "step": 21065
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.13683440886322e-07,
+      "loss": 0.4626,
+      "step": 21066
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1338761567893607e-07,
+      "loss": 0.4587,
+      "step": 21067
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1309199317654516e-07,
+      "loss": 0.4365,
+      "step": 21068
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1279657338527326e-07,
+      "loss": 0.5085,
+      "step": 21069
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.125013563112366e-07,
+      "loss": 0.4412,
+      "step": 21070
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1220634196055244e-07,
+      "loss": 0.4185,
+      "step": 21071
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1191153033932933e-07,
+      "loss": 0.3831,
+      "step": 21072
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1161692145367674e-07,
+      "loss": 0.449,
+      "step": 21073
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1132251530969206e-07,
+      "loss": 0.4338,
+      "step": 21074
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1102831191347817e-07,
+      "loss": 0.4431,
+      "step": 21075
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1073431127112463e-07,
+      "loss": 0.4541,
+      "step": 21076
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1044051338872328e-07,
+      "loss": 0.48,
+      "step": 21077
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1014691827235813e-07,
+      "loss": 0.4066,
+      "step": 21078
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0985352592811204e-07,
+      "loss": 0.4504,
+      "step": 21079
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.095603363620602e-07,
+      "loss": 0.4888,
+      "step": 21080
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.092673495802755e-07,
+      "loss": 0.4526,
+      "step": 21081
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0897456558882645e-07,
+      "loss": 0.4929,
+      "step": 21082
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.086819843937782e-07,
+      "loss": 0.4653,
+      "step": 21083
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.083896060011903e-07,
+      "loss": 0.5107,
+      "step": 21084
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0809743041711795e-07,
+      "loss": 0.4468,
+      "step": 21085
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0780545764761407e-07,
+      "loss": 0.3982,
+      "step": 21086
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0751368769872382e-07,
+      "loss": 0.4585,
+      "step": 21087
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.072221205764946e-07,
+      "loss": 0.4854,
+      "step": 21088
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0693075628696158e-07,
+      "loss": 0.4539,
+      "step": 21089
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0663959483616103e-07,
+      "loss": 0.4028,
+      "step": 21090
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.063486362301248e-07,
+      "loss": 0.4189,
+      "step": 21091
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0605788047487697e-07,
+      "loss": 0.4182,
+      "step": 21092
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.057673275764416e-07,
+      "loss": 0.4363,
+      "step": 21093
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0547697754083607e-07,
+      "loss": 0.498,
+      "step": 21094
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.051868303740756e-07,
+      "loss": 0.4407,
+      "step": 21095
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0489688608216762e-07,
+      "loss": 0.4375,
+      "step": 21096
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0460714467111952e-07,
+      "loss": 0.3879,
+      "step": 21097
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0431760614693096e-07,
+      "loss": 0.3962,
+      "step": 21098
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0402827051560047e-07,
+      "loss": 0.4324,
+      "step": 21099
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0373913778311882e-07,
+      "loss": 0.4248,
+      "step": 21100
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0345020795547788e-07,
+      "loss": 0.4688,
+      "step": 21101
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.031614810386595e-07,
+      "loss": 0.4729,
+      "step": 21102
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0287295703864452e-07,
+      "loss": 0.4331,
+      "step": 21103
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.025846359614092e-07,
+      "loss": 0.3899,
+      "step": 21104
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0229651781292547e-07,
+      "loss": 0.4526,
+      "step": 21105
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0200860259916078e-07,
+      "loss": 0.4116,
+      "step": 21106
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.017208903260792e-07,
+      "loss": 0.4749,
+      "step": 21107
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0143338099963827e-07,
+      "loss": 0.4841,
+      "step": 21108
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0114607462579538e-07,
+      "loss": 0.4149,
+      "step": 21109
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0085897121050025e-07,
+      "loss": 0.5337,
+      "step": 21110
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.005720707596981e-07,
+      "loss": 0.4683,
+      "step": 21111
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0028537327933418e-07,
+      "loss": 0.415,
+      "step": 21112
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9999887877534373e-07,
+      "loss": 0.4653,
+      "step": 21113
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9971258725366206e-07,
+      "loss": 0.4373,
+      "step": 21114
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9942649872021992e-07,
+      "loss": 0.4531,
+      "step": 21115
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9914061318094148e-07,
+      "loss": 0.4404,
+      "step": 21116
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9885493064174866e-07,
+      "loss": 0.4255,
+      "step": 21117
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9856945110856006e-07,
+      "loss": 0.4492,
+      "step": 21118
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9828417458728656e-07,
+      "loss": 0.4668,
+      "step": 21119
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.979991010838378e-07,
+      "loss": 0.4465,
+      "step": 21120
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9771423060411799e-07,
+      "loss": 0.4153,
+      "step": 21121
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9742956315402907e-07,
+      "loss": 0.4209,
+      "step": 21122
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9714509873946407e-07,
+      "loss": 0.3999,
+      "step": 21123
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9686083736631835e-07,
+      "loss": 0.47,
+      "step": 21124
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9657677904047822e-07,
+      "loss": 0.4243,
+      "step": 21125
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9629292376782683e-07,
+      "loss": 0.4497,
+      "step": 21126
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9600927155424388e-07,
+      "loss": 0.438,
+      "step": 21127
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9572582240560356e-07,
+      "loss": 0.4309,
+      "step": 21128
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9544257632777897e-07,
+      "loss": 0.425,
+      "step": 21129
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9515953332663428e-07,
+      "loss": 0.5161,
+      "step": 21130
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9487669340803483e-07,
+      "loss": 0.4954,
+      "step": 21131
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9459405657783703e-07,
+      "loss": 0.4783,
+      "step": 21132
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.943116228418951e-07,
+      "loss": 0.4731,
+      "step": 21133
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9402939220605875e-07,
+      "loss": 0.4329,
+      "step": 21134
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9374736467617562e-07,
+      "loss": 0.4434,
+      "step": 21135
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.934655402580843e-07,
+      "loss": 0.4019,
+      "step": 21136
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9318391895762457e-07,
+      "loss": 0.5015,
+      "step": 21137
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9290250078062734e-07,
+      "loss": 0.443,
+      "step": 21138
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.926212857329235e-07,
+      "loss": 0.5022,
+      "step": 21139
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9234027382033616e-07,
+      "loss": 0.4363,
+      "step": 21140
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9205946504868733e-07,
+      "loss": 0.4348,
+      "step": 21141
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9177885942379126e-07,
+      "loss": 0.4536,
+      "step": 21142
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9149845695146218e-07,
+      "loss": 0.4343,
+      "step": 21143
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9121825763750655e-07,
+      "loss": 0.4492,
+      "step": 21144
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9093826148772753e-07,
+      "loss": 0.3855,
+      "step": 21145
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9065846850792602e-07,
+      "loss": 0.5073,
+      "step": 21146
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9037887870389626e-07,
+      "loss": 0.4456,
+      "step": 21147
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.900994920814292e-07,
+      "loss": 0.4807,
+      "step": 21148
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.898203086463124e-07,
+      "loss": 0.4178,
+      "step": 21149
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8954132840432794e-07,
+      "loss": 0.3617,
+      "step": 21150
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.892625513612534e-07,
+      "loss": 0.4635,
+      "step": 21151
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8898397752286414e-07,
+      "loss": 0.4053,
+      "step": 21152
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8870560689493112e-07,
+      "loss": 0.4895,
+      "step": 21153
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.884274394832164e-07,
+      "loss": 0.4282,
+      "step": 21154
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8814947529348538e-07,
+      "loss": 0.4702,
+      "step": 21155
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8787171433149232e-07,
+      "loss": 0.4482,
+      "step": 21156
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8759415660299263e-07,
+      "loss": 0.4454,
+      "step": 21157
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8731680211373393e-07,
+      "loss": 0.4756,
+      "step": 21158
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8703965086946274e-07,
+      "loss": 0.4553,
+      "step": 21159
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8676270287591669e-07,
+      "loss": 0.4568,
+      "step": 21160
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8648595813883452e-07,
+      "loss": 0.3977,
+      "step": 21161
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.862094166639472e-07,
+      "loss": 0.4561,
+      "step": 21162
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.859330784569824e-07,
+      "loss": 0.4194,
+      "step": 21163
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8565694352366437e-07,
+      "loss": 0.4521,
+      "step": 21164
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.853810118697119e-07,
+      "loss": 0.4285,
+      "step": 21165
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8510528350084045e-07,
+      "loss": 0.4182,
+      "step": 21166
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8482975842276096e-07,
+      "loss": 0.4006,
+      "step": 21167
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8455443664118223e-07,
+      "loss": 0.4502,
+      "step": 21168
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.84279318161803e-07,
+      "loss": 0.4368,
+      "step": 21169
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.840044029903254e-07,
+      "loss": 0.3713,
+      "step": 21170
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8372969113244044e-07,
+      "loss": 0.3955,
+      "step": 21171
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.834551825938402e-07,
+      "loss": 0.3794,
+      "step": 21172
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8318087738021018e-07,
+      "loss": 0.4551,
+      "step": 21173
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8290677549723136e-07,
+      "loss": 0.4668,
+      "step": 21174
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.826328769505814e-07,
+      "loss": 0.481,
+      "step": 21175
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8235918174593358e-07,
+      "loss": 0.4026,
+      "step": 21176
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.820856898889556e-07,
+      "loss": 0.4578,
+      "step": 21177
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8181240138531398e-07,
+      "loss": 0.4849,
+      "step": 21178
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8153931624066755e-07,
+      "loss": 0.4047,
+      "step": 21179
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8126643446067405e-07,
+      "loss": 0.5254,
+      "step": 21180
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8099375605098334e-07,
+      "loss": 0.4756,
+      "step": 21181
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.807212810172465e-07,
+      "loss": 0.4119,
+      "step": 21182
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8044900936510567e-07,
+      "loss": 0.4558,
+      "step": 21183
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8017694110019856e-07,
+      "loss": 0.4829,
+      "step": 21184
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.799050762281629e-07,
+      "loss": 0.5112,
+      "step": 21185
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7963341475462747e-07,
+      "loss": 0.4927,
+      "step": 21186
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7936195668522116e-07,
+      "loss": 0.4473,
+      "step": 21187
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7909070202556388e-07,
+      "loss": 0.4431,
+      "step": 21188
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7881965078127784e-07,
+      "loss": 0.4534,
+      "step": 21189
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7854880295797406e-07,
+      "loss": 0.4292,
+      "step": 21190
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7827815856126253e-07,
+      "loss": 0.4722,
+      "step": 21191
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7800771759675096e-07,
+      "loss": 0.47,
+      "step": 21192
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7773748007003932e-07,
+      "loss": 0.4089,
+      "step": 21193
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.774674459867254e-07,
+      "loss": 0.4541,
+      "step": 21194
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7719761535240242e-07,
+      "loss": 0.436,
+      "step": 21195
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7692798817265932e-07,
+      "loss": 0.4282,
+      "step": 21196
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.766585644530805e-07,
+      "loss": 0.4543,
+      "step": 21197
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7638934419924593e-07,
+      "loss": 0.4463,
+      "step": 21198
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7612032741673223e-07,
+      "loss": 0.4448,
+      "step": 21199
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7585151411111168e-07,
+      "loss": 0.4656,
+      "step": 21200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7558290428795088e-07,
+      "loss": 0.4116,
+      "step": 21201
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.753144979528154e-07,
+      "loss": 0.4287,
+      "step": 21202
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7504629511126302e-07,
+      "loss": 0.4534,
+      "step": 21203
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7477829576885042e-07,
+      "loss": 0.4895,
+      "step": 21204
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7451049993112646e-07,
+      "loss": 0.5215,
+      "step": 21205
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7424290760363894e-07,
+      "loss": 0.4795,
+      "step": 21206
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7397551879193009e-07,
+      "loss": 0.4478,
+      "step": 21207
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7370833350153993e-07,
+      "loss": 0.458,
+      "step": 21208
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7344135173799958e-07,
+      "loss": 0.4163,
+      "step": 21209
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.731745735068413e-07,
+      "loss": 0.4719,
+      "step": 21210
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7290799881358843e-07,
+      "loss": 0.4131,
+      "step": 21211
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7264162766376436e-07,
+      "loss": 0.4258,
+      "step": 21212
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7237546006288686e-07,
+      "loss": 0.4636,
+      "step": 21213
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7210949601646597e-07,
+      "loss": 0.4564,
+      "step": 21214
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7184373553001287e-07,
+      "loss": 0.4062,
+      "step": 21215
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7157817860903203e-07,
+      "loss": 0.426,
+      "step": 21216
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7131282525902238e-07,
+      "loss": 0.4524,
+      "step": 21217
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7104767548548175e-07,
+      "loss": 0.4233,
+      "step": 21218
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7078272929390015e-07,
+      "loss": 0.4541,
+      "step": 21219
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.705179866897666e-07,
+      "loss": 0.439,
+      "step": 21220
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7025344767856444e-07,
+      "loss": 0.4613,
+      "step": 21221
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6998911226577264e-07,
+      "loss": 0.4307,
+      "step": 21222
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6972498045686792e-07,
+      "loss": 0.4165,
+      "step": 21223
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6946105225731812e-07,
+      "loss": 0.4634,
+      "step": 21224
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6919732767259112e-07,
+      "loss": 0.4863,
+      "step": 21225
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6893380670815028e-07,
+      "loss": 0.3955,
+      "step": 21226
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6867048936945241e-07,
+      "loss": 0.457,
+      "step": 21227
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.684073756619531e-07,
+      "loss": 0.4832,
+      "step": 21228
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6814446559110021e-07,
+      "loss": 0.4377,
+      "step": 21229
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.678817591623405e-07,
+      "loss": 0.3752,
+      "step": 21230
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.67619256381113e-07,
+      "loss": 0.4204,
+      "step": 21231
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6735695725285883e-07,
+      "loss": 0.4673,
+      "step": 21232
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6709486178300816e-07,
+      "loss": 0.4392,
+      "step": 21233
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6683296997698995e-07,
+      "loss": 0.5007,
+      "step": 21234
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6657128184022765e-07,
+      "loss": 0.4844,
+      "step": 21235
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6630979737814467e-07,
+      "loss": 0.4009,
+      "step": 21236
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.660485165961534e-07,
+      "loss": 0.4241,
+      "step": 21237
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6578743949966837e-07,
+      "loss": 0.4614,
+      "step": 21238
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6552656609409523e-07,
+      "loss": 0.4233,
+      "step": 21239
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6526589638483858e-07,
+      "loss": 0.46,
+      "step": 21240
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6500543037729633e-07,
+      "loss": 0.4888,
+      "step": 21241
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6474516807686415e-07,
+      "loss": 0.4495,
+      "step": 21242
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6448510948893215e-07,
+      "loss": 0.5066,
+      "step": 21243
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6422525461888827e-07,
+      "loss": 0.4099,
+      "step": 21244
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6396560347211265e-07,
+      "loss": 0.4263,
+      "step": 21245
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6370615605398544e-07,
+      "loss": 0.4102,
+      "step": 21246
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6344691236987898e-07,
+      "loss": 0.4155,
+      "step": 21247
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6318787242516343e-07,
+      "loss": 0.4603,
+      "step": 21248
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.629290362252034e-07,
+      "loss": 0.4065,
+      "step": 21249
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6267040377536125e-07,
+      "loss": 0.4294,
+      "step": 21250
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.624119750809927e-07,
+      "loss": 0.4956,
+      "step": 21251
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6215375014745126e-07,
+      "loss": 0.3816,
+      "step": 21252
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6189572898008488e-07,
+      "loss": 0.4587,
+      "step": 21253
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6163791158423815e-07,
+      "loss": 0.4788,
+      "step": 21254
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6138029796525124e-07,
+      "loss": 0.4563,
+      "step": 21255
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6112288812845988e-07,
+      "loss": 0.4692,
+      "step": 21256
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.608656820791954e-07,
+      "loss": 0.468,
+      "step": 21257
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.606086798227857e-07,
+      "loss": 0.4387,
+      "step": 21258
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6035188136455215e-07,
+      "loss": 0.3827,
+      "step": 21259
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6009528670981711e-07,
+      "loss": 0.5166,
+      "step": 21260
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.598388958638919e-07,
+      "loss": 0.458,
+      "step": 21261
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5958270883208894e-07,
+      "loss": 0.4379,
+      "step": 21262
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5932672561971285e-07,
+      "loss": 0.4077,
+      "step": 21263
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.590709462320683e-07,
+      "loss": 0.4119,
+      "step": 21264
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5881537067444996e-07,
+      "loss": 0.3948,
+      "step": 21265
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.585599989521547e-07,
+      "loss": 0.4866,
+      "step": 21266
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5830483107046934e-07,
+      "loss": 0.4729,
+      "step": 21267
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5804986703467974e-07,
+      "loss": 0.4634,
+      "step": 21268
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5779510685006827e-07,
+      "loss": 0.4939,
+      "step": 21269
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5754055052190854e-07,
+      "loss": 0.4771,
+      "step": 21270
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5728619805547518e-07,
+      "loss": 0.4155,
+      "step": 21271
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5703204945603734e-07,
+      "loss": 0.4078,
+      "step": 21272
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5677810472885857e-07,
+      "loss": 0.4651,
+      "step": 21273
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.565243638791969e-07,
+      "loss": 0.4478,
+      "step": 21274
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.562708269123092e-07,
+      "loss": 0.4202,
+      "step": 21275
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5601749383344688e-07,
+      "loss": 0.4399,
+      "step": 21276
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5576436464785682e-07,
+      "loss": 0.4941,
+      "step": 21277
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.555114393607815e-07,
+      "loss": 0.4116,
+      "step": 21278
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5525871797746118e-07,
+      "loss": 0.4723,
+      "step": 21279
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5500620050312943e-07,
+      "loss": 0.4822,
+      "step": 21280
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.547538869430154e-07,
+      "loss": 0.427,
+      "step": 21281
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5450177730234717e-07,
+      "loss": 0.4695,
+      "step": 21282
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5424987158634496e-07,
+      "loss": 0.4224,
+      "step": 21283
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5399816980022686e-07,
+      "loss": 0.4048,
+      "step": 21284
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5374667194920757e-07,
+      "loss": 0.4612,
+      "step": 21285
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.53495378038494e-07,
+      "loss": 0.4216,
+      "step": 21286
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5324428807329316e-07,
+      "loss": 0.4565,
+      "step": 21287
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.529934020588042e-07,
+      "loss": 0.4407,
+      "step": 21288
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5274272000022295e-07,
+      "loss": 0.4171,
+      "step": 21289
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5249224190274414e-07,
+      "loss": 0.4385,
+      "step": 21290
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5224196777155252e-07,
+      "loss": 0.489,
+      "step": 21291
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5199189761183508e-07,
+      "loss": 0.501,
+      "step": 21292
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5174203142877098e-07,
+      "loss": 0.4485,
+      "step": 21293
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5149236922753275e-07,
+      "loss": 0.3999,
+      "step": 21294
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5124291101329513e-07,
+      "loss": 0.4675,
+      "step": 21295
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.509936567912229e-07,
+      "loss": 0.4338,
+      "step": 21296
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.507446065664786e-07,
+      "loss": 0.4207,
+      "step": 21297
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.504957603442214e-07,
+      "loss": 0.4167,
+      "step": 21298
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5024711812960503e-07,
+      "loss": 0.4612,
+      "step": 21299
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4999867992778083e-07,
+      "loss": 0.4666,
+      "step": 21300
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4975044574389364e-07,
+      "loss": 0.4102,
+      "step": 21301
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4950241558308486e-07,
+      "loss": 0.4551,
+      "step": 21302
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4925458945049155e-07,
+      "loss": 0.4111,
+      "step": 21303
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.490069673512462e-07,
+      "loss": 0.4509,
+      "step": 21304
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.487595492904803e-07,
+      "loss": 0.4478,
+      "step": 21305
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4851233527331532e-07,
+      "loss": 0.512,
+      "step": 21306
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4826532530487492e-07,
+      "loss": 0.4449,
+      "step": 21307
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4801851939027167e-07,
+      "loss": 0.4524,
+      "step": 21308
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4777191753462038e-07,
+      "loss": 0.4971,
+      "step": 21309
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.475255197430281e-07,
+      "loss": 0.4014,
+      "step": 21310
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4727932602059735e-07,
+      "loss": 0.4688,
+      "step": 21311
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4703333637242855e-07,
+      "loss": 0.4253,
+      "step": 21312
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.467875508036165e-07,
+      "loss": 0.4714,
+      "step": 21313
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4654196931925046e-07,
+      "loss": 0.4209,
+      "step": 21314
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.462965919244197e-07,
+      "loss": 0.4314,
+      "step": 21315
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4605141862420457e-07,
+      "loss": 0.5083,
+      "step": 21316
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.458064494236855e-07,
+      "loss": 0.4973,
+      "step": 21317
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4556168432793284e-07,
+      "loss": 0.4067,
+      "step": 21318
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.453171233420192e-07,
+      "loss": 0.3949,
+      "step": 21319
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4507276647100944e-07,
+      "loss": 0.4194,
+      "step": 21320
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4482861371996393e-07,
+      "loss": 0.4641,
+      "step": 21321
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4458466509394087e-07,
+      "loss": 0.5024,
+      "step": 21322
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4434092059799066e-07,
+      "loss": 0.3904,
+      "step": 21323
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4409738023716592e-07,
+      "loss": 0.4028,
+      "step": 21324
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4385404401650704e-07,
+      "loss": 0.4766,
+      "step": 21325
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.436109119410567e-07,
+      "loss": 0.4592,
+      "step": 21326
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4336798401584972e-07,
+      "loss": 0.4592,
+      "step": 21327
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4312526024591767e-07,
+      "loss": 0.4263,
+      "step": 21328
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.428827406362876e-07,
+      "loss": 0.47,
+      "step": 21329
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.426404251919844e-07,
+      "loss": 0.4983,
+      "step": 21330
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4239831391802405e-07,
+      "loss": 0.522,
+      "step": 21331
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4215640681942476e-07,
+      "loss": 0.4475,
+      "step": 21332
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4191470390119588e-07,
+      "loss": 0.4771,
+      "step": 21333
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4167320516834116e-07,
+      "loss": 0.4578,
+      "step": 21334
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4143191062586548e-07,
+      "loss": 0.4385,
+      "step": 21335
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4119082027876597e-07,
+      "loss": 0.4451,
+      "step": 21336
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4094993413203528e-07,
+      "loss": 0.4143,
+      "step": 21337
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.407092521906639e-07,
+      "loss": 0.4497,
+      "step": 21338
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.404687744596367e-07,
+      "loss": 0.4082,
+      "step": 21339
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4022850094393414e-07,
+      "loss": 0.437,
+      "step": 21340
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3998843164853337e-07,
+      "loss": 0.3965,
+      "step": 21341
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3974856657840595e-07,
+      "loss": 0.411,
+      "step": 21342
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3950890573852129e-07,
+      "loss": 0.4514,
+      "step": 21343
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.392694491338431e-07,
+      "loss": 0.4473,
+      "step": 21344
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.390301967693297e-07,
+      "loss": 0.509,
+      "step": 21345
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3879114864993936e-07,
+      "loss": 0.5471,
+      "step": 21346
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3855230478062032e-07,
+      "loss": 0.4097,
+      "step": 21347
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3831366516632194e-07,
+      "loss": 0.3994,
+      "step": 21348
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3807522981198474e-07,
+      "loss": 0.4448,
+      "step": 21349
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.378369987225503e-07,
+      "loss": 0.4797,
+      "step": 21350
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3759897190295024e-07,
+      "loss": 0.4475,
+      "step": 21351
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3736114935811618e-07,
+      "loss": 0.4434,
+      "step": 21352
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3712353109297417e-07,
+      "loss": 0.4583,
+      "step": 21353
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.368861171124447e-07,
+      "loss": 0.4451,
+      "step": 21354
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.366489074214461e-07,
+      "loss": 0.4165,
+      "step": 21355
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3641190202489108e-07,
+      "loss": 0.3923,
+      "step": 21356
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3617510092768904e-07,
+      "loss": 0.4517,
+      "step": 21357
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3593850413474495e-07,
+      "loss": 0.48,
+      "step": 21358
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3570211165095714e-07,
+      "loss": 0.4258,
+      "step": 21359
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3546592348122611e-07,
+      "loss": 0.5068,
+      "step": 21360
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3522993963044018e-07,
+      "loss": 0.4392,
+      "step": 21361
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3499416010348876e-07,
+      "loss": 0.4399,
+      "step": 21362
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3475858490525463e-07,
+      "loss": 0.4453,
+      "step": 21363
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.345232140406172e-07,
+      "loss": 0.4644,
+      "step": 21364
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3428804751445256e-07,
+      "loss": 0.4441,
+      "step": 21365
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.340530853316313e-07,
+      "loss": 0.4614,
+      "step": 21366
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3381832749701951e-07,
+      "loss": 0.4817,
+      "step": 21367
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3358377401547883e-07,
+      "loss": 0.4937,
+      "step": 21368
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3334942489186875e-07,
+      "loss": 0.4358,
+      "step": 21369
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3311528013104425e-07,
+      "loss": 0.4185,
+      "step": 21370
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3288133973785257e-07,
+      "loss": 0.4478,
+      "step": 21371
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3264760371714203e-07,
+      "loss": 0.4446,
+      "step": 21372
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.324140720737499e-07,
+      "loss": 0.4209,
+      "step": 21373
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3218074481251676e-07,
+      "loss": 0.4561,
+      "step": 21374
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3194762193827426e-07,
+      "loss": 0.4221,
+      "step": 21375
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.317147034558508e-07,
+      "loss": 0.4385,
+      "step": 21376
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3148198937007028e-07,
+      "loss": 0.5122,
+      "step": 21377
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.312494796857544e-07,
+      "loss": 0.4358,
+      "step": 21378
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3101717440771598e-07,
+      "loss": 0.4426,
+      "step": 21379
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3078507354077007e-07,
+      "loss": 0.4275,
+      "step": 21380
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.305531770897217e-07,
+      "loss": 0.4507,
+      "step": 21381
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3032148505937593e-07,
+      "loss": 0.4724,
+      "step": 21382
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3008999745452888e-07,
+      "loss": 0.4617,
+      "step": 21383
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2985871427997787e-07,
+      "loss": 0.4304,
+      "step": 21384
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.296276355405124e-07,
+      "loss": 0.4082,
+      "step": 21385
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.293967612409186e-07,
+      "loss": 0.3882,
+      "step": 21386
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2916609138597824e-07,
+      "loss": 0.4956,
+      "step": 21387
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.289356259804708e-07,
+      "loss": 0.4873,
+      "step": 21388
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2870536502916587e-07,
+      "loss": 0.4336,
+      "step": 21389
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2847530853683733e-07,
+      "loss": 0.4263,
+      "step": 21390
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2824545650824805e-07,
+      "loss": 0.4685,
+      "step": 21391
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2801580894815758e-07,
+      "loss": 0.4199,
+      "step": 21392
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.277863658613243e-07,
+      "loss": 0.4607,
+      "step": 21393
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.275571272525e-07,
+      "loss": 0.4,
+      "step": 21394
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.27328093126432e-07,
+      "loss": 0.4148,
+      "step": 21395
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2709926348786649e-07,
+      "loss": 0.4951,
+      "step": 21396
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.268706383415408e-07,
+      "loss": 0.4656,
+      "step": 21397
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2664221769219109e-07,
+      "loss": 0.4971,
+      "step": 21398
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2641400154454918e-07,
+      "loss": 0.4697,
+      "step": 21399
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2618598990334018e-07,
+      "loss": 0.4209,
+      "step": 21400
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.259581827732892e-07,
+      "loss": 0.4037,
+      "step": 21401
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2573058015911243e-07,
+      "loss": 0.4172,
+      "step": 21402
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2550318206552613e-07,
+      "loss": 0.4451,
+      "step": 21403
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2527598849723767e-07,
+      "loss": 0.4338,
+      "step": 21404
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2504899945895654e-07,
+      "loss": 0.4412,
+      "step": 21405
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2482221495538126e-07,
+      "loss": 0.4573,
+      "step": 21406
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2459563499120919e-07,
+      "loss": 0.4661,
+      "step": 21407
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2436925957113434e-07,
+      "loss": 0.4214,
+      "step": 21408
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2414308869984515e-07,
+      "loss": 0.4924,
+      "step": 21409
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.239171223820268e-07,
+      "loss": 0.4648,
+      "step": 21410
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2369136062235775e-07,
+      "loss": 0.4309,
+      "step": 21411
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.234658034255165e-07,
+      "loss": 0.4761,
+      "step": 21412
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2324045079617376e-07,
+      "loss": 0.426,
+      "step": 21413
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2301530273899797e-07,
+      "loss": 0.4526,
+      "step": 21414
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.227903592586499e-07,
+      "loss": 0.4233,
+      "step": 21415
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2256562035979248e-07,
+      "loss": 0.4338,
+      "step": 21416
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.223410860470775e-07,
+      "loss": 0.4004,
+      "step": 21417
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2211675632515686e-07,
+      "loss": 0.4199,
+      "step": 21418
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2189263119867677e-07,
+      "loss": 0.4185,
+      "step": 21419
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2166871067228027e-07,
+      "loss": 0.4333,
+      "step": 21420
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2144499475060356e-07,
+      "loss": 0.4907,
+      "step": 21421
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2122148343828076e-07,
+      "loss": 0.3621,
+      "step": 21422
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2099817673994375e-07,
+      "loss": 0.4722,
+      "step": 21423
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2077507466021544e-07,
+      "loss": 0.4597,
+      "step": 21424
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.205521772037177e-07,
+      "loss": 0.3611,
+      "step": 21425
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2032948437506574e-07,
+      "loss": 0.428,
+      "step": 21426
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2010699617887477e-07,
+      "loss": 0.4368,
+      "step": 21427
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1988471261974998e-07,
+      "loss": 0.4058,
+      "step": 21428
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.196626337022988e-07,
+      "loss": 0.4448,
+      "step": 21429
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1944075943111755e-07,
+      "loss": 0.4225,
+      "step": 21430
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1921908981080477e-07,
+      "loss": 0.4783,
+      "step": 21431
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1899762484595013e-07,
+      "loss": 0.4502,
+      "step": 21432
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1877636454114105e-07,
+      "loss": 0.5234,
+      "step": 21433
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1855530890095946e-07,
+      "loss": 0.4846,
+      "step": 21434
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.183344579299861e-07,
+      "loss": 0.4438,
+      "step": 21435
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1811381163279511e-07,
+      "loss": 0.4697,
+      "step": 21436
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1789337001395396e-07,
+      "loss": 0.4114,
+      "step": 21437
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1767313307803119e-07,
+      "loss": 0.4736,
+      "step": 21438
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.174531008295865e-07,
+      "loss": 0.4705,
+      "step": 21439
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1723327327317957e-07,
+      "loss": 0.4441,
+      "step": 21440
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1701365041336233e-07,
+      "loss": 0.4692,
+      "step": 21441
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1679423225468334e-07,
+      "loss": 0.381,
+      "step": 21442
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1657501880168676e-07,
+      "loss": 0.4309,
+      "step": 21443
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1635601005891561e-07,
+      "loss": 0.4514,
+      "step": 21444
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1613720603090295e-07,
+      "loss": 0.4695,
+      "step": 21445
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1591860672218292e-07,
+      "loss": 0.416,
+      "step": 21446
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.15700212137283e-07,
+      "loss": 0.4678,
+      "step": 21447
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1548202228072624e-07,
+      "loss": 0.4312,
+      "step": 21448
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1526403715703127e-07,
+      "loss": 0.4653,
+      "step": 21449
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1504625677071446e-07,
+      "loss": 0.4561,
+      "step": 21450
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1482868112628554e-07,
+      "loss": 0.4292,
+      "step": 21451
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.14611310228252e-07,
+      "loss": 0.4167,
+      "step": 21452
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1439414408111471e-07,
+      "loss": 0.4011,
+      "step": 21453
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1417718268937229e-07,
+      "loss": 0.397,
+      "step": 21454
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1396042605752e-07,
+      "loss": 0.4587,
+      "step": 21455
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.137438741900454e-07,
+      "loss": 0.4209,
+      "step": 21456
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1352752709143489e-07,
+      "loss": 0.4395,
+      "step": 21457
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1331138476616932e-07,
+      "loss": 0.3892,
+      "step": 21458
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1309544721872512e-07,
+      "loss": 0.4731,
+      "step": 21459
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1287971445357537e-07,
+      "loss": 0.4048,
+      "step": 21460
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1266418647518873e-07,
+      "loss": 0.4514,
+      "step": 21461
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1244886328802828e-07,
+      "loss": 0.4277,
+      "step": 21462
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1223374489655492e-07,
+      "loss": 0.4617,
+      "step": 21463
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1201883130522396e-07,
+      "loss": 0.5186,
+      "step": 21464
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1180412251848626e-07,
+      "loss": 0.4131,
+      "step": 21465
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1158961854079054e-07,
+      "loss": 0.4082,
+      "step": 21466
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1137531937657764e-07,
+      "loss": 0.4429,
+      "step": 21467
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1116122503028736e-07,
+      "loss": 0.4297,
+      "step": 21468
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1094733550635395e-07,
+      "loss": 0.4614,
+      "step": 21469
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1073365080920829e-07,
+      "loss": 0.4561,
+      "step": 21470
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1052017094327461e-07,
+      "loss": 0.4204,
+      "step": 21471
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1030689591297494e-07,
+      "loss": 0.4751,
+      "step": 21472
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1009382572272908e-07,
+      "loss": 0.4575,
+      "step": 21473
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0988096037694796e-07,
+      "loss": 0.5244,
+      "step": 21474
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0966829988004025e-07,
+      "loss": 0.4529,
+      "step": 21475
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0945584423641353e-07,
+      "loss": 0.4058,
+      "step": 21476
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0924359345046431e-07,
+      "loss": 0.4663,
+      "step": 21477
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.090315475265924e-07,
+      "loss": 0.4536,
+      "step": 21478
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.088197064691876e-07,
+      "loss": 0.4265,
+      "step": 21479
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0860807028263865e-07,
+      "loss": 0.4446,
+      "step": 21480
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0839663897132868e-07,
+      "loss": 0.4678,
+      "step": 21481
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0818541253963754e-07,
+      "loss": 0.5583,
+      "step": 21482
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0797439099193952e-07,
+      "loss": 0.4155,
+      "step": 21483
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0776357433260443e-07,
+      "loss": 0.4919,
+      "step": 21484
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0755296256600101e-07,
+      "loss": 0.425,
+      "step": 21485
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0734255569649022e-07,
+      "loss": 0.4153,
+      "step": 21486
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0713235372843078e-07,
+      "loss": 0.4529,
+      "step": 21487
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0692235666617589e-07,
+      "loss": 0.498,
+      "step": 21488
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0671256451407542e-07,
+      "loss": 0.5105,
+      "step": 21489
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0650297727647363e-07,
+      "loss": 0.4502,
+      "step": 21490
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0629359495771374e-07,
+      "loss": 0.4758,
+      "step": 21491
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0608441756213117e-07,
+      "loss": 0.3879,
+      "step": 21492
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0587544509405912e-07,
+      "loss": 0.4504,
+      "step": 21493
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0566667755782412e-07,
+      "loss": 0.478,
+      "step": 21494
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0545811495775382e-07,
+      "loss": 0.4604,
+      "step": 21495
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0524975729816367e-07,
+      "loss": 0.396,
+      "step": 21496
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0504160458337354e-07,
+      "loss": 0.4392,
+      "step": 21497
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0483365681769108e-07,
+      "loss": 0.5049,
+      "step": 21498
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.046259140054262e-07,
+      "loss": 0.3931,
+      "step": 21499
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0441837615088102e-07,
+      "loss": 0.4326,
+      "step": 21500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0421104325835208e-07,
+      "loss": 0.5129,
+      "step": 21501
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0400391533213706e-07,
+      "loss": 0.6033,
+      "step": 21502
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0379699237652474e-07,
+      "loss": 0.46,
+      "step": 21503
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0359027439580061e-07,
+      "loss": 0.4587,
+      "step": 21504
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0338376139424566e-07,
+      "loss": 0.4104,
+      "step": 21505
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0317745337613872e-07,
+      "loss": 0.4954,
+      "step": 21506
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.029713503457519e-07,
+      "loss": 0.4814,
+      "step": 21507
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0276545230735514e-07,
+      "loss": 0.5098,
+      "step": 21508
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0255975926521167e-07,
+      "loss": 0.416,
+      "step": 21509
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0235427122358366e-07,
+      "loss": 0.338,
+      "step": 21510
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0214898818672658e-07,
+      "loss": 0.4753,
+      "step": 21511
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0194391015889149e-07,
+      "loss": 0.3838,
+      "step": 21512
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.017390371443272e-07,
+      "loss": 0.4081,
+      "step": 21513
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0153436914727588e-07,
+      "loss": 0.4229,
+      "step": 21514
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0132990617197858e-07,
+      "loss": 0.4141,
+      "step": 21515
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0112564822266746e-07,
+      "loss": 0.4597,
+      "step": 21516
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0092159530357692e-07,
+      "loss": 0.4385,
+      "step": 21517
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0071774741892914e-07,
+      "loss": 0.5173,
+      "step": 21518
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0051410457295074e-07,
+      "loss": 0.4363,
+      "step": 21519
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0031066676985612e-07,
+      "loss": 0.4421,
+      "step": 21520
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.001074340138608e-07,
+      "loss": 0.4204,
+      "step": 21521
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.990440630917364e-08,
+      "loss": 0.4351,
+      "step": 21522
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.970158365999904e-08,
+      "loss": 0.4551,
+      "step": 21523
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.949896607054033e-08,
+      "loss": 0.4856,
+      "step": 21524
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.92965535449919e-08,
+      "loss": 0.5049,
+      "step": 21525
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.90943460875482e-08,
+      "loss": 0.4709,
+      "step": 21526
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.889234370239475e-08,
+      "loss": 0.4639,
+      "step": 21527
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.869054639371823e-08,
+      "loss": 0.4312,
+      "step": 21528
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.848895416569637e-08,
+      "loss": 0.5,
+      "step": 21529
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.828756702250697e-08,
+      "loss": 0.4263,
+      "step": 21530
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.80863849683189e-08,
+      "loss": 0.4316,
+      "step": 21531
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.788540800729996e-08,
+      "loss": 0.4907,
+      "step": 21532
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.768463614361234e-08,
+      "loss": 0.5046,
+      "step": 21533
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.748406938141719e-08,
+      "loss": 0.5278,
+      "step": 21534
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.728370772486451e-08,
+      "loss": 0.438,
+      "step": 21535
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.708355117810764e-08,
+      "loss": 0.4231,
+      "step": 21536
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.688359974529105e-08,
+      "loss": 0.4717,
+      "step": 21537
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.668385343055586e-08,
+      "loss": 0.4177,
+      "step": 21538
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.64843122380399e-08,
+      "loss": 0.4299,
+      "step": 21539
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.62849761718776e-08,
+      "loss": 0.4441,
+      "step": 21540
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.608584523619458e-08,
+      "loss": 0.4714,
+      "step": 21541
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.588691943511752e-08,
+      "loss": 0.4824,
+      "step": 21542
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.568819877276536e-08,
+      "loss": 0.511,
+      "step": 21543
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.548968325325481e-08,
+      "loss": 0.4336,
+      "step": 21544
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.529137288069812e-08,
+      "loss": 0.4363,
+      "step": 21545
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.509326765920202e-08,
+      "loss": 0.4485,
+      "step": 21546
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.4895367592871e-08,
+      "loss": 0.48,
+      "step": 21547
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.469767268580177e-08,
+      "loss": 0.4792,
+      "step": 21548
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.450018294208995e-08,
+      "loss": 0.4226,
+      "step": 21549
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.430289836582785e-08,
+      "loss": 0.4614,
+      "step": 21550
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.410581896109993e-08,
+      "loss": 0.4771,
+      "step": 21551
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.390894473198742e-08,
+      "loss": 0.4304,
+      "step": 21552
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.371227568257146e-08,
+      "loss": 0.4541,
+      "step": 21553
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.351581181692215e-08,
+      "loss": 0.4514,
+      "step": 21554
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.331955313910957e-08,
+      "loss": 0.4287,
+      "step": 21555
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.312349965320045e-08,
+      "loss": 0.4407,
+      "step": 21556
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.292765136325266e-08,
+      "loss": 0.3975,
+      "step": 21557
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.273200827332519e-08,
+      "loss": 0.4631,
+      "step": 21558
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.253657038746922e-08,
+      "loss": 0.4641,
+      "step": 21559
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.234133770973264e-08,
+      "loss": 0.5383,
+      "step": 21560
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.214631024415998e-08,
+      "loss": 0.4233,
+      "step": 21561
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.195148799479026e-08,
+      "loss": 0.4321,
+      "step": 21562
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.175687096565689e-08,
+      "loss": 0.3799,
+      "step": 21563
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.156245916079443e-08,
+      "loss": 0.4707,
+      "step": 21564
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.136825258422744e-08,
+      "loss": 0.444,
+      "step": 21565
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.117425123997937e-08,
+      "loss": 0.4114,
+      "step": 21566
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.098045513206699e-08,
+      "loss": 0.4324,
+      "step": 21567
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.0786864264506e-08,
+      "loss": 0.4587,
+      "step": 21568
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.059347864130541e-08,
+      "loss": 0.4397,
+      "step": 21569
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.040029826646979e-08,
+      "loss": 0.4866,
+      "step": 21570
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.02073231440026e-08,
+      "loss": 0.4543,
+      "step": 21571
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.001455327789843e-08,
+      "loss": 0.447,
+      "step": 21572
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.982198867215074e-08,
+      "loss": 0.4709,
+      "step": 21573
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.96296293307497e-08,
+      "loss": 0.4548,
+      "step": 21574
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.943747525767654e-08,
+      "loss": 0.467,
+      "step": 21575
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.924552645691365e-08,
+      "loss": 0.4897,
+      "step": 21576
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.905378293243671e-08,
+      "loss": 0.4883,
+      "step": 21577
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.886224468821591e-08,
+      "loss": 0.4312,
+      "step": 21578
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.867091172822029e-08,
+      "loss": 0.3723,
+      "step": 21579
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.847978405641112e-08,
+      "loss": 0.4175,
+      "step": 21580
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.828886167674744e-08,
+      "loss": 0.4656,
+      "step": 21581
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.809814459318389e-08,
+      "loss": 0.3896,
+      "step": 21582
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.790763280967063e-08,
+      "loss": 0.4895,
+      "step": 21583
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.771732633015451e-08,
+      "loss": 0.4653,
+      "step": 21584
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.752722515857571e-08,
+      "loss": 0.4507,
+      "step": 21585
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.733732929887328e-08,
+      "loss": 0.429,
+      "step": 21586
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.714763875497856e-08,
+      "loss": 0.4609,
+      "step": 21587
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.695815353082282e-08,
+      "loss": 0.4526,
+      "step": 21588
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.676887363032849e-08,
+      "loss": 0.5034,
+      "step": 21589
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.657979905741798e-08,
+      "loss": 0.4377,
+      "step": 21590
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.639092981600595e-08,
+      "loss": 0.4089,
+      "step": 21591
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.62022659100048e-08,
+      "loss": 0.4275,
+      "step": 21592
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.601380734332254e-08,
+      "loss": 0.4392,
+      "step": 21593
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.58255541198627e-08,
+      "loss": 0.5325,
+      "step": 21594
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.56375062435244e-08,
+      "loss": 0.428,
+      "step": 21595
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.544966371820229e-08,
+      "loss": 0.4929,
+      "step": 21596
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.52620265477877e-08,
+      "loss": 0.4065,
+      "step": 21597
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.507459473616641e-08,
+      "loss": 0.4058,
+      "step": 21598
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.488736828721978e-08,
+      "loss": 0.4285,
+      "step": 21599
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.47003472048269e-08,
+      "loss": 0.4548,
+      "step": 21600
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.451353149286246e-08,
+      "loss": 0.4307,
+      "step": 21601
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.432692115519337e-08,
+      "loss": 0.4568,
+      "step": 21602
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.414051619568764e-08,
+      "loss": 0.4324,
+      "step": 21603
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.39543166182033e-08,
+      "loss": 0.4546,
+      "step": 21604
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.376832242659838e-08,
+      "loss": 0.4214,
+      "step": 21605
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.358253362472535e-08,
+      "loss": 0.4185,
+      "step": 21606
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.339695021643335e-08,
+      "loss": 0.4404,
+      "step": 21607
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.321157220556375e-08,
+      "loss": 0.5078,
+      "step": 21608
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.302639959595904e-08,
+      "loss": 0.4238,
+      "step": 21609
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.284143239145171e-08,
+      "loss": 0.4446,
+      "step": 21610
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.265667059587535e-08,
+      "loss": 0.4561,
+      "step": 21611
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.247211421305468e-08,
+      "loss": 0.4412,
+      "step": 21612
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.228776324681442e-08,
+      "loss": 0.4353,
+      "step": 21613
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.210361770097264e-08,
+      "loss": 0.4105,
+      "step": 21614
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.191967757934072e-08,
+      "loss": 0.4009,
+      "step": 21615
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.173594288573228e-08,
+      "loss": 0.4761,
+      "step": 21616
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.155241362395094e-08,
+      "loss": 0.458,
+      "step": 21617
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.136908979779812e-08,
+      "loss": 0.446,
+      "step": 21618
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.118597141107076e-08,
+      "loss": 0.4192,
+      "step": 21619
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.100305846756252e-08,
+      "loss": 0.5471,
+      "step": 21620
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.082035097106144e-08,
+      "loss": 0.4033,
+      "step": 21621
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.063784892535232e-08,
+      "loss": 0.4529,
+      "step": 21622
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.045555233421431e-08,
+      "loss": 0.4844,
+      "step": 21623
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.027346120142332e-08,
+      "loss": 0.4441,
+      "step": 21624
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.009157553075186e-08,
+      "loss": 0.406,
+      "step": 21625
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.990989532596694e-08,
+      "loss": 0.4614,
+      "step": 21626
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.972842059083108e-08,
+      "loss": 0.4009,
+      "step": 21627
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.954715132910352e-08,
+      "loss": 0.561,
+      "step": 21628
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.9366087544539e-08,
+      "loss": 0.4473,
+      "step": 21629
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.918522924088679e-08,
+      "loss": 0.4343,
+      "step": 21630
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.900457642189386e-08,
+      "loss": 0.4421,
+      "step": 21631
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.882412909130166e-08,
+      "loss": 0.4246,
+      "step": 21632
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.864388725284833e-08,
+      "loss": 0.4551,
+      "step": 21633
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.846385091026531e-08,
+      "loss": 0.4634,
+      "step": 21634
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.828402006728408e-08,
+      "loss": 0.4302,
+      "step": 21635
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.810439472762721e-08,
+      "loss": 0.4722,
+      "step": 21636
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.792497489501616e-08,
+      "loss": 0.4719,
+      "step": 21637
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.774576057316686e-08,
+      "loss": 0.4211,
+      "step": 21638
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.756675176579298e-08,
+      "loss": 0.4678,
+      "step": 21639
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.738794847659937e-08,
+      "loss": 0.4244,
+      "step": 21640
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.72093507092908e-08,
+      "loss": 0.4236,
+      "step": 21641
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.70309584675677e-08,
+      "loss": 0.427,
+      "step": 21642
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.685277175512263e-08,
+      "loss": 0.4707,
+      "step": 21643
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.667479057564819e-08,
+      "loss": 0.4783,
+      "step": 21644
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.649701493283035e-08,
+      "loss": 0.4197,
+      "step": 21645
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.631944483035059e-08,
+      "loss": 0.4055,
+      "step": 21646
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.614208027188819e-08,
+      "loss": 0.4624,
+      "step": 21647
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.596492126111466e-08,
+      "loss": 0.4265,
+      "step": 21648
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.578796780170151e-08,
+      "loss": 0.4175,
+      "step": 21649
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.561121989731245e-08,
+      "loss": 0.4375,
+      "step": 21650
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.543467755161016e-08,
+      "loss": 0.3943,
+      "step": 21651
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.525834076824834e-08,
+      "loss": 0.4583,
+      "step": 21652
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.508220955088185e-08,
+      "loss": 0.4038,
+      "step": 21653
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.49062839031589e-08,
+      "loss": 0.4419,
+      "step": 21654
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.473056382872102e-08,
+      "loss": 0.4402,
+      "step": 21655
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.455504933121082e-08,
+      "loss": 0.4365,
+      "step": 21656
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.437974041426099e-08,
+      "loss": 0.4675,
+      "step": 21657
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.420463708150416e-08,
+      "loss": 0.4463,
+      "step": 21658
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.402973933656743e-08,
+      "loss": 0.4763,
+      "step": 21659
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.385504718307124e-08,
+      "loss": 0.4336,
+      "step": 21660
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.368056062463713e-08,
+      "loss": 0.4618,
+      "step": 21661
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.350627966487556e-08,
+      "loss": 0.4438,
+      "step": 21662
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.333220430739919e-08,
+      "loss": 0.5168,
+      "step": 21663
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.315833455581178e-08,
+      "loss": 0.4937,
+      "step": 21664
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.298467041371605e-08,
+      "loss": 0.4512,
+      "step": 21665
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.281121188470797e-08,
+      "loss": 0.4609,
+      "step": 21666
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.263795897238024e-08,
+      "loss": 0.4807,
+      "step": 21667
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.24649116803211e-08,
+      "loss": 0.4607,
+      "step": 21668
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.229207001211658e-08,
+      "loss": 0.5464,
+      "step": 21669
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.211943397134491e-08,
+      "loss": 0.4249,
+      "step": 21670
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.194700356158324e-08,
+      "loss": 0.5144,
+      "step": 21671
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.177477878640093e-08,
+      "loss": 0.4319,
+      "step": 21672
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.160275964936736e-08,
+      "loss": 0.4413,
+      "step": 21673
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.1430946154043e-08,
+      "loss": 0.4189,
+      "step": 21674
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.125933830398946e-08,
+      "loss": 0.4087,
+      "step": 21675
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.108793610275943e-08,
+      "loss": 0.4805,
+      "step": 21676
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.091673955390344e-08,
+      "loss": 0.4563,
+      "step": 21677
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.074574866096751e-08,
+      "loss": 0.4802,
+      "step": 21678
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.057496342749215e-08,
+      "loss": 0.4097,
+      "step": 21679
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.040438385701675e-08,
+      "loss": 0.4399,
+      "step": 21680
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.023400995307295e-08,
+      "loss": 0.4331,
+      "step": 21681
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.006384171919012e-08,
+      "loss": 0.4517,
+      "step": 21682
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.989387915889323e-08,
+      "loss": 0.4102,
+      "step": 21683
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.972412227570168e-08,
+      "loss": 0.3792,
+      "step": 21684
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.955457107313269e-08,
+      "loss": 0.4575,
+      "step": 21685
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.938522555469674e-08,
+      "loss": 0.4595,
+      "step": 21686
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.921608572390214e-08,
+      "loss": 0.4368,
+      "step": 21687
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.90471515842528e-08,
+      "loss": 0.3923,
+      "step": 21688
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.887842313924809e-08,
+      "loss": 0.4958,
+      "step": 21689
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.870990039237968e-08,
+      "loss": 0.4558,
+      "step": 21690
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.854158334714145e-08,
+      "loss": 0.4285,
+      "step": 21691
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.837347200701839e-08,
+      "loss": 0.4364,
+      "step": 21692
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.820556637549325e-08,
+      "loss": 0.4102,
+      "step": 21693
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.803786645604216e-08,
+      "loss": 0.4429,
+      "step": 21694
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.787037225214011e-08,
+      "loss": 0.4463,
+      "step": 21695
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.770308376725543e-08,
+      "loss": 0.4434,
+      "step": 21696
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.753600100485314e-08,
+      "loss": 0.4749,
+      "step": 21697
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.73691239683949e-08,
+      "loss": 0.4492,
+      "step": 21698
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.720245266133685e-08,
+      "loss": 0.4377,
+      "step": 21699
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.703598708712955e-08,
+      "loss": 0.4485,
+      "step": 21700
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.686972724922247e-08,
+      "loss": 0.458,
+      "step": 21701
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.670367315105952e-08,
+      "loss": 0.4072,
+      "step": 21702
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.653782479607907e-08,
+      "loss": 0.4814,
+      "step": 21703
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.637218218771724e-08,
+      "loss": 0.4829,
+      "step": 21704
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.620674532940352e-08,
+      "loss": 0.4775,
+      "step": 21705
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.604151422456629e-08,
+      "loss": 0.4341,
+      "step": 21706
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.587648887662612e-08,
+      "loss": 0.5098,
+      "step": 21707
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.571166928900252e-08,
+      "loss": 0.4731,
+      "step": 21708
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.55470554651083e-08,
+      "loss": 0.4504,
+      "step": 21709
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.53826474083541e-08,
+      "loss": 0.4115,
+      "step": 21710
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.521844512214381e-08,
+      "loss": 0.4609,
+      "step": 21711
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.50544486098792e-08,
+      "loss": 0.4304,
+      "step": 21712
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.489065787495751e-08,
+      "loss": 0.4419,
+      "step": 21713
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.472707292077052e-08,
+      "loss": 0.4492,
+      "step": 21714
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.45636937507077e-08,
+      "loss": 0.5251,
+      "step": 21715
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.440052036815081e-08,
+      "loss": 0.4272,
+      "step": 21716
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.423755277648159e-08,
+      "loss": 0.5188,
+      "step": 21717
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.40747909790751e-08,
+      "loss": 0.4187,
+      "step": 21718
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.391223497930198e-08,
+      "loss": 0.4753,
+      "step": 21719
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.374988478052846e-08,
+      "loss": 0.4526,
+      "step": 21720
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.35877403861196e-08,
+      "loss": 0.4333,
+      "step": 21721
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.342580179943158e-08,
+      "loss": 0.4817,
+      "step": 21722
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.326406902381953e-08,
+      "loss": 0.4695,
+      "step": 21723
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.3102542062633e-08,
+      "loss": 0.5229,
+      "step": 21724
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.294122091921817e-08,
+      "loss": 0.4978,
+      "step": 21725
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.278010559691461e-08,
+      "loss": 0.5078,
+      "step": 21726
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.261919609906186e-08,
+      "loss": 0.4141,
+      "step": 21727
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.245849242899171e-08,
+      "loss": 0.5371,
+      "step": 21728
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.22979945900326e-08,
+      "loss": 0.4414,
+      "step": 21729
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.213770258550855e-08,
+      "loss": 0.4158,
+      "step": 21730
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.197761641874134e-08,
+      "loss": 0.4272,
+      "step": 21731
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.181773609304387e-08,
+      "loss": 0.437,
+      "step": 21732
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.165806161173016e-08,
+      "loss": 0.3831,
+      "step": 21733
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.149859297810757e-08,
+      "loss": 0.446,
+      "step": 21734
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.133933019547567e-08,
+      "loss": 0.4478,
+      "step": 21735
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.118027326713738e-08,
+      "loss": 0.4265,
+      "step": 21736
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.102142219638452e-08,
+      "loss": 0.4294,
+      "step": 21737
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.086277698650778e-08,
+      "loss": 0.4797,
+      "step": 21738
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.070433764079454e-08,
+      "loss": 0.4844,
+      "step": 21739
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.054610416252438e-08,
+      "loss": 0.3772,
+      "step": 21740
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.038807655497581e-08,
+      "loss": 0.3678,
+      "step": 21741
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.023025482142287e-08,
+      "loss": 0.5056,
+      "step": 21742
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.007263896513182e-08,
+      "loss": 0.4207,
+      "step": 21743
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.991522898937008e-08,
+      "loss": 0.4277,
+      "step": 21744
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.975802489739613e-08,
+      "loss": 0.3877,
+      "step": 21745
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.960102669246737e-08,
+      "loss": 0.4543,
+      "step": 21746
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.944423437783342e-08,
+      "loss": 0.4507,
+      "step": 21747
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.928764795674502e-08,
+      "loss": 0.4634,
+      "step": 21748
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.913126743244402e-08,
+      "loss": 0.4458,
+      "step": 21749
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.897509280816782e-08,
+      "loss": 0.4478,
+      "step": 21750
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.8819124087152734e-08,
+      "loss": 0.4705,
+      "step": 21751
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.866336127263062e-08,
+      "loss": 0.4902,
+      "step": 21752
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.850780436782444e-08,
+      "loss": 0.4111,
+      "step": 21753
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.8352453375959406e-08,
+      "loss": 0.426,
+      "step": 21754
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.819730830025072e-08,
+      "loss": 0.4399,
+      "step": 21755
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.804236914391359e-08,
+      "loss": 0.4421,
+      "step": 21756
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.788763591015656e-08,
+      "loss": 0.4761,
+      "step": 21757
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.773310860218373e-08,
+      "loss": 0.4683,
+      "step": 21758
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.7578787223198096e-08,
+      "loss": 0.5125,
+      "step": 21759
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.742467177639266e-08,
+      "loss": 0.4487,
+      "step": 21760
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.7270762264961535e-08,
+      "loss": 0.3782,
+      "step": 21761
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.711705869209328e-08,
+      "loss": 0.4509,
+      "step": 21762
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6963561060969784e-08,
+      "loss": 0.4492,
+      "step": 21763
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.681026937477185e-08,
+      "loss": 0.4634,
+      "step": 21764
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.66571836366725e-08,
+      "loss": 0.4292,
+      "step": 21765
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6504303849844734e-08,
+      "loss": 0.4502,
+      "step": 21766
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6351630017453804e-08,
+      "loss": 0.4027,
+      "step": 21767
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6199162142662746e-08,
+      "loss": 0.4822,
+      "step": 21768
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.604690022862791e-08,
+      "loss": 0.4744,
+      "step": 21769
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5894844278505666e-08,
+      "loss": 0.4446,
+      "step": 21770
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.574299429544239e-08,
+      "loss": 0.4124,
+      "step": 21771
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5591350282586666e-08,
+      "loss": 0.4167,
+      "step": 21772
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.543991224307599e-08,
+      "loss": 0.4097,
+      "step": 21773
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.528868018005007e-08,
+      "loss": 0.4751,
+      "step": 21774
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.513765409663863e-08,
+      "loss": 0.4653,
+      "step": 21775
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4986833995972485e-08,
+      "loss": 0.4885,
+      "step": 21776
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4836219881172494e-08,
+      "loss": 0.4258,
+      "step": 21777
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4685811755361695e-08,
+      "loss": 0.5146,
+      "step": 21778
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4535609621652055e-08,
+      "loss": 0.4553,
+      "step": 21779
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.438561348315552e-08,
+      "loss": 0.4678,
+      "step": 21780
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.423582334298072e-08,
+      "loss": 0.4294,
+      "step": 21781
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.408623920422851e-08,
+      "loss": 0.4198,
+      "step": 21782
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.3936861069998625e-08,
+      "loss": 0.4111,
+      "step": 21783
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.378768894338304e-08,
+      "loss": 0.3815,
+      "step": 21784
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.363872282747262e-08,
+      "loss": 0.405,
+      "step": 21785
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.348996272535156e-08,
+      "loss": 0.4817,
+      "step": 21786
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.334140864010407e-08,
+      "loss": 0.4824,
+      "step": 21787
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.3193060574803226e-08,
+      "loss": 0.532,
+      "step": 21788
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.304491853252547e-08,
+      "loss": 0.4792,
+      "step": 21789
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2896982516336125e-08,
+      "loss": 0.4236,
+      "step": 21790
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.274925252930052e-08,
+      "loss": 0.426,
+      "step": 21791
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.260172857447954e-08,
+      "loss": 0.4253,
+      "step": 21792
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.245441065492629e-08,
+      "loss": 0.4353,
+      "step": 21793
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2307298773695004e-08,
+      "loss": 0.4739,
+      "step": 21794
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2160392933829907e-08,
+      "loss": 0.4707,
+      "step": 21795
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.201369313837634e-08,
+      "loss": 0.4519,
+      "step": 21796
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.186719939037077e-08,
+      "loss": 0.4817,
+      "step": 21797
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.1720911692848544e-08,
+      "loss": 0.4594,
+      "step": 21798
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.157483004883834e-08,
+      "loss": 0.4436,
+      "step": 21799
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.142895446136886e-08,
+      "loss": 0.4304,
+      "step": 21800
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.12832849334588e-08,
+      "loss": 0.3918,
+      "step": 21801
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.113782146812574e-08,
+      "loss": 0.4011,
+      "step": 21802
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.099256406838282e-08,
+      "loss": 0.4175,
+      "step": 21803
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.084751273723876e-08,
+      "loss": 0.4351,
+      "step": 21804
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.070266747769892e-08,
+      "loss": 0.3982,
+      "step": 21805
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.055802829276202e-08,
+      "loss": 0.4778,
+      "step": 21806
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.041359518542455e-08,
+      "loss": 0.4749,
+      "step": 21807
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.0269368158677445e-08,
+      "loss": 0.4436,
+      "step": 21808
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.0125347215509437e-08,
+      "loss": 0.478,
+      "step": 21809
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.998153235890257e-08,
+      "loss": 0.4324,
+      "step": 21810
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.983792359183448e-08,
+      "loss": 0.4814,
+      "step": 21811
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.9694520917281666e-08,
+      "loss": 0.4595,
+      "step": 21812
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.955132433821286e-08,
+      "loss": 0.4082,
+      "step": 21813
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.940833385759569e-08,
+      "loss": 0.4106,
+      "step": 21814
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.926554947839002e-08,
+      "loss": 0.4971,
+      "step": 21815
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.912297120355347e-08,
+      "loss": 0.4695,
+      "step": 21816
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8980599036041465e-08,
+      "loss": 0.447,
+      "step": 21817
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.883843297879942e-08,
+      "loss": 0.4111,
+      "step": 21818
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.869647303477498e-08,
+      "loss": 0.4058,
+      "step": 21819
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.855471920690691e-08,
+      "loss": 0.502,
+      "step": 21820
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.841317149813174e-08,
+      "loss": 0.4751,
+      "step": 21821
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8271829911380464e-08,
+      "loss": 0.4917,
+      "step": 21822
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.813069444958185e-08,
+      "loss": 0.4663,
+      "step": 21823
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7989765115659116e-08,
+      "loss": 0.4609,
+      "step": 21824
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7849041912529926e-08,
+      "loss": 0.4124,
+      "step": 21825
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.770852484310973e-08,
+      "loss": 0.4502,
+      "step": 21826
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.756821391030952e-08,
+      "loss": 0.5071,
+      "step": 21827
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.742810911703477e-08,
+      "loss": 0.4329,
+      "step": 21828
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.728821046618759e-08,
+      "loss": 0.4297,
+      "step": 21829
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.714851796066566e-08,
+      "loss": 0.5188,
+      "step": 21830
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.700903160336223e-08,
+      "loss": 0.4724,
+      "step": 21831
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.686975139716721e-08,
+      "loss": 0.47,
+      "step": 21832
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.673067734496384e-08,
+      "loss": 0.4883,
+      "step": 21833
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.659180944963537e-08,
+      "loss": 0.5469,
+      "step": 21834
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.645314771405507e-08,
+      "loss": 0.5518,
+      "step": 21835
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.6314692141096183e-08,
+      "loss": 0.4583,
+      "step": 21836
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.6176442733627535e-08,
+      "loss": 0.4529,
+      "step": 21837
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.603839949451128e-08,
+      "loss": 0.4636,
+      "step": 21838
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.590056242660623e-08,
+      "loss": 0.4568,
+      "step": 21839
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.5762931532769005e-08,
+      "loss": 0.4651,
+      "step": 21840
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.562550681584954e-08,
+      "loss": 0.4788,
+      "step": 21841
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.548828827869334e-08,
+      "loss": 0.4836,
+      "step": 21842
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.5351275924143675e-08,
+      "loss": 0.4497,
+      "step": 21843
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.521446975503718e-08,
+      "loss": 0.4575,
+      "step": 21844
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.507786977420936e-08,
+      "loss": 0.4941,
+      "step": 21845
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4941475984487946e-08,
+      "loss": 0.4431,
+      "step": 21846
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.480528838869958e-08,
+      "loss": 0.4272,
+      "step": 21847
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4669306989661986e-08,
+      "loss": 0.4668,
+      "step": 21848
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.453353179019515e-08,
+      "loss": 0.4458,
+      "step": 21849
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.439796279311015e-08,
+      "loss": 0.4536,
+      "step": 21850
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.426260000121363e-08,
+      "loss": 0.4336,
+      "step": 21851
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4127443417311124e-08,
+      "loss": 0.4038,
+      "step": 21852
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.399249304420039e-08,
+      "loss": 0.4202,
+      "step": 21853
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.385774888467698e-08,
+      "loss": 0.4285,
+      "step": 21854
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.37232109415342e-08,
+      "loss": 0.4556,
+      "step": 21855
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3588879217554283e-08,
+      "loss": 0.4094,
+      "step": 21856
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.345475371552388e-08,
+      "loss": 0.4585,
+      "step": 21857
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.332083443821855e-08,
+      "loss": 0.4181,
+      "step": 21858
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3187121388412745e-08,
+      "loss": 0.4399,
+      "step": 21859
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.305361456887536e-08,
+      "loss": 0.4624,
+      "step": 21860
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.292031398237306e-08,
+      "loss": 0.4502,
+      "step": 21861
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.278721963166588e-08,
+      "loss": 0.4319,
+      "step": 21862
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2654331519509376e-08,
+      "loss": 0.4399,
+      "step": 21863
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.252164964865913e-08,
+      "loss": 0.4121,
+      "step": 21864
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2389174021860715e-08,
+      "loss": 0.4153,
+      "step": 21865
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2256904641858607e-08,
+      "loss": 0.4126,
+      "step": 21866
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.212484151139285e-08,
+      "loss": 0.439,
+      "step": 21867
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.199298463319901e-08,
+      "loss": 0.4736,
+      "step": 21868
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.186133401000714e-08,
+      "loss": 0.4675,
+      "step": 21869
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.172988964454394e-08,
+      "loss": 0.4183,
+      "step": 21870
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.159865153953391e-08,
+      "loss": 0.4414,
+      "step": 21871
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.146761969769375e-08,
+      "loss": 0.4702,
+      "step": 21872
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1336794121736854e-08,
+      "loss": 0.4519,
+      "step": 21873
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.120617481437439e-08,
+      "loss": 0.4331,
+      "step": 21874
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.107576177831196e-08,
+      "loss": 0.4373,
+      "step": 21875
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.094555501624964e-08,
+      "loss": 0.4619,
+      "step": 21876
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.081555453088415e-08,
+      "loss": 0.4897,
+      "step": 21877
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0685760324908896e-08,
+      "loss": 0.4421,
+      "step": 21878
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.055617240101284e-08,
+      "loss": 0.4175,
+      "step": 21879
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.042679076187828e-08,
+      "loss": 0.3892,
+      "step": 21880
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.02976154101864e-08,
+      "loss": 0.4421,
+      "step": 21881
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.016864634861173e-08,
+      "loss": 0.5083,
+      "step": 21882
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0039883579826576e-08,
+      "loss": 0.4473,
+      "step": 21883
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.991132710649881e-08,
+      "loss": 0.3989,
+      "step": 21884
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.978297693128852e-08,
+      "loss": 0.467,
+      "step": 21885
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.96548330568558e-08,
+      "loss": 0.45,
+      "step": 21886
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.952689548585409e-08,
+      "loss": 0.4404,
+      "step": 21887
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.9399164220934595e-08,
+      "loss": 0.4575,
+      "step": 21888
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.9271639264741865e-08,
+      "loss": 0.459,
+      "step": 21889
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.914432061991713e-08,
+      "loss": 0.4482,
+      "step": 21890
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.901720828909827e-08,
+      "loss": 0.4338,
+      "step": 21891
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8890302274917635e-08,
+      "loss": 0.4282,
+      "step": 21892
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.876360258000311e-08,
+      "loss": 0.4629,
+      "step": 21893
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8637109206980384e-08,
+      "loss": 0.4712,
+      "step": 21894
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8510822158467354e-08,
+      "loss": 0.4622,
+      "step": 21895
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.838474143708304e-08,
+      "loss": 0.4218,
+      "step": 21896
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.825886704543647e-08,
+      "loss": 0.4226,
+      "step": 21897
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.8133198986134437e-08,
+      "loss": 0.4512,
+      "step": 21898
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.800773726178042e-08,
+      "loss": 0.3708,
+      "step": 21899
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7882481874974565e-08,
+      "loss": 0.3868,
+      "step": 21900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7757432828309235e-08,
+      "loss": 0.4309,
+      "step": 21901
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.763259012437459e-08,
+      "loss": 0.4043,
+      "step": 21902
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.750795376575856e-08,
+      "loss": 0.4648,
+      "step": 21903
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7383523755040176e-08,
+      "loss": 0.3773,
+      "step": 21904
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.725930009479739e-08,
+      "loss": 0.4763,
+      "step": 21905
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.713528278760481e-08,
+      "loss": 0.4927,
+      "step": 21906
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.701147183602816e-08,
+      "loss": 0.5078,
+      "step": 21907
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.688786724263427e-08,
+      "loss": 0.4602,
+      "step": 21908
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6764469009983314e-08,
+      "loss": 0.4192,
+      "step": 21909
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6641277140629925e-08,
+      "loss": 0.4509,
+      "step": 21910
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.651829163712539e-08,
+      "loss": 0.4636,
+      "step": 21911
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.639551250201878e-08,
+      "loss": 0.439,
+      "step": 21912
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6272939737852505e-08,
+      "loss": 0.4536,
+      "step": 21913
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6150573347164543e-08,
+      "loss": 0.4646,
+      "step": 21914
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.602841333248952e-08,
+      "loss": 0.4287,
+      "step": 21915
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.590645969635875e-08,
+      "loss": 0.4709,
+      "step": 21916
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5784712441296886e-08,
+      "loss": 0.4125,
+      "step": 21917
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.566317156982746e-08,
+      "loss": 0.4391,
+      "step": 21918
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5541837084465125e-08,
+      "loss": 0.4226,
+      "step": 21919
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.542070898772565e-08,
+      "loss": 0.4534,
+      "step": 21920
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.529978728211703e-08,
+      "loss": 0.4138,
+      "step": 21921
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5179071970142807e-08,
+      "loss": 0.4111,
+      "step": 21922
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.505856305430433e-08,
+      "loss": 0.4403,
+      "step": 21923
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.493826053709848e-08,
+      "loss": 0.4749,
+      "step": 21924
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.481816442101549e-08,
+      "loss": 0.4221,
+      "step": 21925
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.469827470854226e-08,
+      "loss": 0.4617,
+      "step": 21926
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.457859140216457e-08,
+      "loss": 0.4065,
+      "step": 21927
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.445911450435935e-08,
+      "loss": 0.4167,
+      "step": 21928
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4339844017601264e-08,
+      "loss": 0.5071,
+      "step": 21929
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.422077994436168e-08,
+      "loss": 0.4268,
+      "step": 21930
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.41019222871064e-08,
+      "loss": 0.4639,
+      "step": 21931
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.39832710482968e-08,
+      "loss": 0.4661,
+      "step": 21932
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.386482623039089e-08,
+      "loss": 0.449,
+      "step": 21933
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.374658783584117e-08,
+      "loss": 0.4294,
+      "step": 21934
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3628555867097904e-08,
+      "loss": 0.4426,
+      "step": 21935
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.351073032660579e-08,
+      "loss": 0.4692,
+      "step": 21936
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3393111216804e-08,
+      "loss": 0.4131,
+      "step": 21937
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.327569854012946e-08,
+      "loss": 0.4592,
+      "step": 21938
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.315849229901469e-08,
+      "loss": 0.4939,
+      "step": 21939
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.304149249588662e-08,
+      "loss": 0.3959,
+      "step": 21940
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.292469913316776e-08,
+      "loss": 0.4043,
+      "step": 21941
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.280811221327951e-08,
+      "loss": 0.3911,
+      "step": 21942
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.269173173863438e-08,
+      "loss": 0.4373,
+      "step": 21943
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.257555771164378e-08,
+      "loss": 0.5093,
+      "step": 21944
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.245959013471356e-08,
+      "loss": 0.4485,
+      "step": 21945
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.2343829010247355e-08,
+      "loss": 0.3967,
+      "step": 21946
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.222827434063991e-08,
+      "loss": 0.4189,
+      "step": 21947
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.21129261282882e-08,
+      "loss": 0.4009,
+      "step": 21948
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.19977843755781e-08,
+      "loss": 0.3809,
+      "step": 21949
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.188284908489658e-08,
+      "loss": 0.3735,
+      "step": 21950
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.176812025862397e-08,
+      "loss": 0.4426,
+      "step": 21951
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.165359789913502e-08,
+      "loss": 0.498,
+      "step": 21952
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1539282008803405e-08,
+      "loss": 0.5034,
+      "step": 21953
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.142517258999722e-08,
+      "loss": 0.4364,
+      "step": 21954
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.131126964507791e-08,
+      "loss": 0.4668,
+      "step": 21955
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1197573176406925e-08,
+      "loss": 0.408,
+      "step": 21956
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.108408318633793e-08,
+      "loss": 0.438,
+      "step": 21957
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0970799677221276e-08,
+      "loss": 0.4719,
+      "step": 21958
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.085772265140508e-08,
+      "loss": 0.499,
+      "step": 21959
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.074485211122968e-08,
+      "loss": 0.449,
+      "step": 21960
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.063218805903323e-08,
+      "loss": 0.4377,
+      "step": 21961
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.05197304971494e-08,
+      "loss": 0.4456,
+      "step": 21962
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.040747942790856e-08,
+      "loss": 0.3855,
+      "step": 21963
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.02954348536344e-08,
+      "loss": 0.4746,
+      "step": 21964
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.018359677664839e-08,
+      "loss": 0.397,
+      "step": 21965
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.007196519926536e-08,
+      "loss": 0.4514,
+      "step": 21966
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.996054012380012e-08,
+      "loss": 0.4451,
+      "step": 21967
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.98493215525586e-08,
+      "loss": 0.4829,
+      "step": 21968
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9738309487844507e-08,
+      "loss": 0.4001,
+      "step": 21969
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9627503931958234e-08,
+      "loss": 0.4688,
+      "step": 21970
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.951690488719461e-08,
+      "loss": 0.4316,
+      "step": 21971
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.940651235584402e-08,
+      "loss": 0.5054,
+      "step": 21972
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9296326340192414e-08,
+      "loss": 0.4534,
+      "step": 21973
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9186346842522416e-08,
+      "loss": 0.4873,
+      "step": 21974
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.90765738651122e-08,
+      "loss": 0.4463,
+      "step": 21975
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8967007410236614e-08,
+      "loss": 0.4785,
+      "step": 21976
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.885764748016273e-08,
+      "loss": 0.4949,
+      "step": 21977
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8748494077156518e-08,
+      "loss": 0.4324,
+      "step": 21978
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.86395472034795e-08,
+      "loss": 0.4434,
+      "step": 21979
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.853080686138876e-08,
+      "loss": 0.4607,
+      "step": 21980
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8422273053133608e-08,
+      "loss": 0.4138,
+      "step": 21981
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.831394578096558e-08,
+      "loss": 0.4688,
+      "step": 21982
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8205825047126212e-08,
+      "loss": 0.4744,
+      "step": 21983
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8097910853855936e-08,
+      "loss": 0.4319,
+      "step": 21984
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.799020320338963e-08,
+      "loss": 0.4927,
+      "step": 21985
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.7882702097958847e-08,
+      "loss": 0.4541,
+      "step": 21986
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.7775407539788468e-08,
+      "loss": 0.5107,
+      "step": 21987
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7668319531102273e-08,
+      "loss": 0.4827,
+      "step": 21988
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7561438074118485e-08,
+      "loss": 0.3645,
+      "step": 21989
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.745476317104867e-08,
+      "loss": 0.4741,
+      "step": 21990
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7348294824105505e-08,
+      "loss": 0.4561,
+      "step": 21991
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7242033035492776e-08,
+      "loss": 0.4604,
+      "step": 21992
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.713597780741095e-08,
+      "loss": 0.436,
+      "step": 21993
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.703012914205605e-08,
+      "loss": 0.4717,
+      "step": 21994
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.692448704162298e-08,
+      "loss": 0.4661,
+      "step": 21995
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6819051508297778e-08,
+      "loss": 0.4194,
+      "step": 21996
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6713822544265354e-08,
+      "loss": 0.4675,
+      "step": 21997
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6608800151703974e-08,
+      "loss": 0.3975,
+      "step": 21998
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6503984332790776e-08,
+      "loss": 0.4301,
+      "step": 21999
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6399375089695145e-08,
+      "loss": 0.4399,
+      "step": 22000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.629497242458423e-08,
+      "loss": 0.4225,
+      "step": 22001
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.619077633962075e-08,
+      "loss": 0.425,
+      "step": 22002
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6086786836961863e-08,
+      "loss": 0.4568,
+      "step": 22003
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5983003918763627e-08,
+      "loss": 0.4268,
+      "step": 22004
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.587942758717321e-08,
+      "loss": 0.5066,
+      "step": 22005
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5776057844337788e-08,
+      "loss": 0.4602,
+      "step": 22006
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5672894692397865e-08,
+      "loss": 0.4219,
+      "step": 22007
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5569938133489514e-08,
+      "loss": 0.4436,
+      "step": 22008
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.546718816974547e-08,
+      "loss": 0.4941,
+      "step": 22009
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5364644803294037e-08,
+      "loss": 0.3965,
+      "step": 22010
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5262308036260174e-08,
+      "loss": 0.48,
+      "step": 22011
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5160177870762193e-08,
+      "loss": 0.4548,
+      "step": 22012
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5058254308915063e-08,
+      "loss": 0.4688,
+      "step": 22013
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4956537352831545e-08,
+      "loss": 0.4653,
+      "step": 22014
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4855027004617728e-08,
+      "loss": 0.4199,
+      "step": 22015
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4753723266376373e-08,
+      "loss": 0.4329,
+      "step": 22016
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4652626140204694e-08,
+      "loss": 0.436,
+      "step": 22017
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4551735628197683e-08,
+      "loss": 0.4189,
+      "step": 22018
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4451051732444776e-08,
+      "loss": 0.4502,
+      "step": 22019
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4350574455030974e-08,
+      "loss": 0.4888,
+      "step": 22020
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4250303798037944e-08,
+      "loss": 0.4487,
+      "step": 22021
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4150239763541804e-08,
+      "loss": 0.4734,
+      "step": 22022
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4050382353616452e-08,
+      "loss": 0.4622,
+      "step": 22023
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3950731570329122e-08,
+      "loss": 0.4353,
+      "step": 22024
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3851287415743717e-08,
+      "loss": 0.4854,
+      "step": 22025
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3752049891920815e-08,
+      "loss": 0.4312,
+      "step": 22026
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3653019000915434e-08,
+      "loss": 0.5347,
+      "step": 22027
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.355419474477816e-08,
+      "loss": 0.4739,
+      "step": 22028
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.345557712555735e-08,
+      "loss": 0.4602,
+      "step": 22029
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3357166145294707e-08,
+      "loss": 0.4482,
+      "step": 22030
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3258961806028603e-08,
+      "loss": 0.4827,
+      "step": 22031
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.316096410979185e-08,
+      "loss": 0.5125,
+      "step": 22032
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.306317305861616e-08,
+      "loss": 0.3867,
+      "step": 22033
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.296558865452658e-08,
+      "loss": 0.4199,
+      "step": 22034
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.286821089954372e-08,
+      "loss": 0.3795,
+      "step": 22035
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2771039795685957e-08,
+      "loss": 0.4893,
+      "step": 22036
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2674075344963908e-08,
+      "loss": 0.4526,
+      "step": 22037
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2577317549387078e-08,
+      "loss": 0.4084,
+      "step": 22038
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.248076641095831e-08,
+      "loss": 0.3997,
+      "step": 22039
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2384421931680446e-08,
+      "loss": 0.4707,
+      "step": 22040
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2288284113545222e-08,
+      "loss": 0.4463,
+      "step": 22041
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2192352958546602e-08,
+      "loss": 0.4678,
+      "step": 22042
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2096628468670777e-08,
+      "loss": 0.4431,
+      "step": 22043
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.20011106459006e-08,
+      "loss": 0.4976,
+      "step": 22044
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1905799492213387e-08,
+      "loss": 0.4333,
+      "step": 22045
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1810695009584215e-08,
+      "loss": 0.45,
+      "step": 22046
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1715797199983733e-08,
+      "loss": 0.4672,
+      "step": 22047
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1621106065375928e-08,
+      "loss": 0.4155,
+      "step": 22048
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1526621607722565e-08,
+      "loss": 0.4885,
+      "step": 22049
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.143234382898096e-08,
+      "loss": 0.4199,
+      "step": 22050
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1338272731104004e-08,
+      "loss": 0.4912,
+      "step": 22051
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.124440831604013e-08,
+      "loss": 0.4751,
+      "step": 22052
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.115075058573335e-08,
+      "loss": 0.4617,
+      "step": 22053
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.10572995421221e-08,
+      "loss": 0.4358,
+      "step": 22054
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.096405518714484e-08,
+      "loss": 0.4312,
+      "step": 22055
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.087101752273002e-08,
+      "loss": 0.4304,
+      "step": 22056
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.077818655080721e-08,
+      "loss": 0.4731,
+      "step": 22057
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.06855622732971e-08,
+      "loss": 0.3831,
+      "step": 22058
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.059314469211926e-08,
+      "loss": 0.4646,
+      "step": 22059
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0500933809187718e-08,
+      "loss": 0.3909,
+      "step": 22060
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0408929626412054e-08,
+      "loss": 0.4497,
+      "step": 22061
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0317132145698526e-08,
+      "loss": 0.4001,
+      "step": 22062
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.022554136894894e-08,
+      "loss": 0.4497,
+      "step": 22063
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0134157298058454e-08,
+      "loss": 0.4126,
+      "step": 22064
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0042979934921102e-08,
+      "loss": 0.4543,
+      "step": 22065
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9952009281426487e-08,
+      "loss": 0.4133,
+      "step": 22066
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9861245339456438e-08,
+      "loss": 0.5127,
+      "step": 22067
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9770688110892778e-08,
+      "loss": 0.3958,
+      "step": 22068
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9680337597609567e-08,
+      "loss": 0.4141,
+      "step": 22069
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.959019380147975e-08,
+      "loss": 0.4392,
+      "step": 22070
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9500256724369614e-08,
+      "loss": 0.4896,
+      "step": 22071
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9410526368142114e-08,
+      "loss": 0.4519,
+      "step": 22072
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.932100273465687e-08,
+      "loss": 0.4574,
+      "step": 22073
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9231685825765732e-08,
+      "loss": 0.4048,
+      "step": 22074
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.914257564332056e-08,
+      "loss": 0.4382,
+      "step": 22075
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.905367218916654e-08,
+      "loss": 0.512,
+      "step": 22076
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.896497546514553e-08,
+      "loss": 0.4539,
+      "step": 22077
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8876485473092732e-08,
+      "loss": 0.4067,
+      "step": 22078
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8788202214844452e-08,
+      "loss": 0.4695,
+      "step": 22079
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8700125692225903e-08,
+      "loss": 0.4714,
+      "step": 22080
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8612255907063393e-08,
+      "loss": 0.3949,
+      "step": 22081
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8524592861175474e-08,
+      "loss": 0.4736,
+      "step": 22082
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8437136556379575e-08,
+      "loss": 0.4363,
+      "step": 22083
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8349886994485365e-08,
+      "loss": 0.4487,
+      "step": 22084
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.82628441773014e-08,
+      "loss": 0.5376,
+      "step": 22085
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8176008106629562e-08,
+      "loss": 0.532,
+      "step": 22086
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.808937878426953e-08,
+      "loss": 0.4751,
+      "step": 22087
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.800295621201431e-08,
+      "loss": 0.4089,
+      "step": 22088
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7916740391654698e-08,
+      "loss": 0.4365,
+      "step": 22089
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7830731324977034e-08,
+      "loss": 0.4844,
+      "step": 22090
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7744929013761015e-08,
+      "loss": 0.4924,
+      "step": 22091
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.765933345978521e-08,
+      "loss": 0.4558,
+      "step": 22092
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7573944664822652e-08,
+      "loss": 0.3918,
+      "step": 22093
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.748876263064081e-08,
+      "loss": 0.4431,
+      "step": 22094
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7403787359004942e-08,
+      "loss": 0.4077,
+      "step": 22095
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.731901885167475e-08,
+      "loss": 0.4863,
+      "step": 22096
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.723445711040661e-08,
+      "loss": 0.5042,
+      "step": 22097
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7150102136951342e-08,
+      "loss": 0.459,
+      "step": 22098
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7065953933055324e-08,
+      "loss": 0.467,
+      "step": 22099
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.698201250046383e-08,
+      "loss": 0.45,
+      "step": 22100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6898277840913246e-08,
+      "loss": 0.4531,
+      "step": 22101
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6814749956138852e-08,
+      "loss": 0.4116,
+      "step": 22102
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.673142884787038e-08,
+      "loss": 0.4697,
+      "step": 22103
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.664831451783422e-08,
+      "loss": 0.4087,
+      "step": 22104
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.656540696775122e-08,
+      "loss": 0.4141,
+      "step": 22105
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6482706199338893e-08,
+      "loss": 0.3977,
+      "step": 22106
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.640021221431143e-08,
+      "loss": 0.4548,
+      "step": 22107
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6317925014375235e-08,
+      "loss": 0.3813,
+      "step": 22108
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6235844601235617e-08,
+      "loss": 0.4333,
+      "step": 22109
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6153970976592325e-08,
+      "loss": 0.4045,
+      "step": 22110
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.607230414214178e-08,
+      "loss": 0.4351,
+      "step": 22111
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5990844099575963e-08,
+      "loss": 0.4744,
+      "step": 22112
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5909590850581304e-08,
+      "loss": 0.4612,
+      "step": 22113
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5828544396840894e-08,
+      "loss": 0.4009,
+      "step": 22114
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5747704740033398e-08,
+      "loss": 0.3789,
+      "step": 22115
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5667071881833028e-08,
+      "loss": 0.4824,
+      "step": 22116
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5586645823910673e-08,
+      "loss": 0.3911,
+      "step": 22117
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5506426567930555e-08,
+      "loss": 0.4631,
+      "step": 22118
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5426414115555786e-08,
+      "loss": 0.4375,
+      "step": 22119
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5346608468442826e-08,
+      "loss": 0.4077,
+      "step": 22120
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.52670096282459e-08,
+      "loss": 0.4541,
+      "step": 22121
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5187617596611472e-08,
+      "loss": 0.3999,
+      "step": 22122
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5108432375186e-08,
+      "loss": 0.4351,
+      "step": 22123
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5029453965608177e-08,
+      "loss": 0.4451,
+      "step": 22124
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4950682369514468e-08,
+      "loss": 0.5134,
+      "step": 22125
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4872117588536905e-08,
+      "loss": 0.4546,
+      "step": 22126
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4793759624301962e-08,
+      "loss": 0.3779,
+      "step": 22127
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4715608478432786e-08,
+      "loss": 0.4819,
+      "step": 22128
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4637664152548082e-08,
+      "loss": 0.4124,
+      "step": 22129
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4559926648263223e-08,
+      "loss": 0.4863,
+      "step": 22130
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4482395967186924e-08,
+      "loss": 0.4629,
+      "step": 22131
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4405072110925678e-08,
+      "loss": 0.4873,
+      "step": 22132
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4327955081080424e-08,
+      "loss": 0.4309,
+      "step": 22133
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4251044879249887e-08,
+      "loss": 0.4375,
+      "step": 22134
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4174341507026123e-08,
+      "loss": 0.4348,
+      "step": 22135
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4097844965997864e-08,
+      "loss": 0.4221,
+      "step": 22136
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4021555257749398e-08,
+      "loss": 0.5017,
+      "step": 22137
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.394547238386057e-08,
+      "loss": 0.4832,
+      "step": 22138
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3869596345909009e-08,
+      "loss": 0.4534,
+      "step": 22139
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3793927145464569e-08,
+      "loss": 0.4048,
+      "step": 22140
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3718464784094887e-08,
+      "loss": 0.4312,
+      "step": 22141
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3643209263363156e-08,
+      "loss": 0.4954,
+      "step": 22142
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3568160584828127e-08,
+      "loss": 0.4524,
+      "step": 22143
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3493318750044116e-08,
+      "loss": 0.4426,
+      "step": 22144
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.341868376056099e-08,
+      "loss": 0.4487,
+      "step": 22145
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3344255617925294e-08,
+      "loss": 0.4609,
+      "step": 22146
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3270034323679127e-08,
+      "loss": 0.4666,
+      "step": 22147
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3196019879357924e-08,
+      "loss": 0.4595,
+      "step": 22148
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3122212286496017e-08,
+      "loss": 0.426,
+      "step": 22149
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3048611546622181e-08,
+      "loss": 0.4158,
+      "step": 22150
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2975217661260753e-08,
+      "loss": 0.3837,
+      "step": 22151
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2902030631931629e-08,
+      "loss": 0.4609,
+      "step": 22152
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2829050460150261e-08,
+      "loss": 0.4717,
+      "step": 22153
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2756277147428775e-08,
+      "loss": 0.415,
+      "step": 22154
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2683710695274853e-08,
+      "loss": 0.4558,
+      "step": 22155
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2611351105191739e-08,
+      "loss": 0.4714,
+      "step": 22156
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.253919837867601e-08,
+      "loss": 0.4575,
+      "step": 22157
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2467252517225359e-08,
+      "loss": 0.4551,
+      "step": 22158
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2395513522327485e-08,
+      "loss": 0.4265,
+      "step": 22159
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2323981395468975e-08,
+      "loss": 0.4087,
+      "step": 22160
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.225265613813198e-08,
+      "loss": 0.449,
+      "step": 22161
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2181537751793093e-08,
+      "loss": 0.4517,
+      "step": 22162
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2110626237926693e-08,
+      "loss": 0.4443,
+      "step": 22163
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2039921597999381e-08,
+      "loss": 0.4492,
+      "step": 22164
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1969423833477767e-08,
+      "loss": 0.4182,
+      "step": 22165
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.189913294582068e-08,
+      "loss": 0.3711,
+      "step": 22166
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1829048936483622e-08,
+      "loss": 0.4329,
+      "step": 22167
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1759171806919878e-08,
+      "loss": 0.4473,
+      "step": 22168
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1689501558576066e-08,
+      "loss": 0.4414,
+      "step": 22169
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1620038192894368e-08,
+      "loss": 0.4558,
+      "step": 22170
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.155078171131363e-08,
+      "loss": 0.3972,
+      "step": 22171
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1481732115270483e-08,
+      "loss": 0.4368,
+      "step": 22172
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1412889406192673e-08,
+      "loss": 0.4385,
+      "step": 22173
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1344253585506836e-08,
+      "loss": 0.5872,
+      "step": 22174
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.127582465463406e-08,
+      "loss": 0.5022,
+      "step": 22175
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1207602614993207e-08,
+      "loss": 0.5105,
+      "step": 22176
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1139587467996481e-08,
+      "loss": 0.4036,
+      "step": 22177
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1071779215052759e-08,
+      "loss": 0.4346,
+      "step": 22178
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1004177857566468e-08,
+      "loss": 0.4622,
+      "step": 22179
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.09367833969376e-08,
+      "loss": 0.4421,
+      "step": 22180
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0869595834561708e-08,
+      "loss": 0.4644,
+      "step": 22181
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.080261517183101e-08,
+      "loss": 0.4124,
+      "step": 22182
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0735841410132175e-08,
+      "loss": 0.4309,
+      "step": 22183
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.066927455084965e-08,
+      "loss": 0.4761,
+      "step": 22184
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0602914595361224e-08,
+      "loss": 0.4832,
+      "step": 22185
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0536761545041351e-08,
+      "loss": 0.4111,
+      "step": 22186
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0470815401260048e-08,
+      "loss": 0.407,
+      "step": 22187
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0405076165383997e-08,
+      "loss": 0.4248,
+      "step": 22188
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0339543838774335e-08,
+      "loss": 0.3979,
+      "step": 22189
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0274218422787752e-08,
+      "loss": 0.4543,
+      "step": 22190
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.020909991877872e-08,
+      "loss": 0.4277,
+      "step": 22191
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.014418832809505e-08,
+      "loss": 0.386,
+      "step": 22192
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0079483652081223e-08,
+      "loss": 0.4243,
+      "step": 22193
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0014985892077278e-08,
+      "loss": 0.4507,
+      "step": 22194
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.950695049421034e-09,
+      "loss": 0.4375,
+      "step": 22195
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.886611125441425e-09,
+      "loss": 0.4678,
+      "step": 22196
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.822734121466282e-09,
+      "loss": 0.4958,
+      "step": 22197
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.759064038820098e-09,
+      "loss": 0.4551,
+      "step": 22198
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.69560087882071e-09,
+      "loss": 0.4744,
+      "step": 22199
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.63234464278262e-09,
+      "loss": 0.3801,
+      "step": 22200
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.569295332017003e-09,
+      "loss": 0.4473,
+      "step": 22201
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.50645294782726e-09,
+      "loss": 0.4111,
+      "step": 22202
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.443817491517904e-09,
+      "loss": 0.4697,
+      "step": 22203
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.381388964384564e-09,
+      "loss": 0.4614,
+      "step": 22204
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.319167367720649e-09,
+      "loss": 0.411,
+      "step": 22205
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.257152702815131e-09,
+      "loss": 0.4824,
+      "step": 22206
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.195344970952535e-09,
+      "loss": 0.437,
+      "step": 22207
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.13374417341295e-09,
+      "loss": 0.438,
+      "step": 22208
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.072350311472022e-09,
+      "loss": 0.4575,
+      "step": 22209
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.011163386400957e-09,
+      "loss": 0.4253,
+      "step": 22210
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.950183399468736e-09,
+      "loss": 0.4229,
+      "step": 22211
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.889410351936577e-09,
+      "loss": 0.3857,
+      "step": 22212
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.828844245064582e-09,
+      "loss": 0.4727,
+      "step": 22213
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.768485080106192e-09,
+      "loss": 0.4521,
+      "step": 22214
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.70833285831263e-09,
+      "loss": 0.3911,
+      "step": 22215
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.648387580928453e-09,
+      "loss": 0.4661,
+      "step": 22216
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.588649249196002e-09,
+      "loss": 0.4429,
+      "step": 22217
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.529117864354285e-09,
+      "loss": 0.4229,
+      "step": 22218
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.469793427633432e-09,
+      "loss": 0.4866,
+      "step": 22219
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.410675940263568e-09,
+      "loss": 0.4573,
+      "step": 22220
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.351765403469269e-09,
+      "loss": 0.4512,
+      "step": 22221
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.293061818469562e-09,
+      "loss": 0.417,
+      "step": 22222
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.234565186482358e-09,
+      "loss": 0.4177,
+      "step": 22223
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.176275508717802e-09,
+      "loss": 0.4045,
+      "step": 22224
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.118192786384927e-09,
+      "loss": 0.4934,
+      "step": 22225
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.060317020683884e-09,
+      "loss": 0.4753,
+      "step": 22226
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.002648212815934e-09,
+      "loss": 0.4448,
+      "step": 22227
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.945186363973456e-09,
+      "loss": 0.4656,
+      "step": 22228
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.88793147534883e-09,
+      "loss": 0.3789,
+      "step": 22229
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.830883548126667e-09,
+      "loss": 0.4746,
+      "step": 22230
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.77404258348935e-09,
+      "loss": 0.4382,
+      "step": 22231
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.717408582612607e-09,
+      "loss": 0.4624,
+      "step": 22232
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.660981546671053e-09,
+      "loss": 0.4368,
+      "step": 22233
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.604761476833756e-09,
+      "loss": 0.4414,
+      "step": 22234
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.548748374263115e-09,
+      "loss": 0.4915,
+      "step": 22235
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.492942240120426e-09,
+      "loss": 0.45,
+      "step": 22236
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.437343075562542e-09,
+      "loss": 0.4785,
+      "step": 22237
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.3819508817396526e-09,
+      "loss": 0.4419,
+      "step": 22238
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.3267656597997284e-09,
+      "loss": 0.4385,
+      "step": 22239
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.2717874108851896e-09,
+      "loss": 0.4568,
+      "step": 22240
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.217016136136234e-09,
+      "loss": 0.4841,
+      "step": 22241
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.162451836685291e-09,
+      "loss": 0.4993,
+      "step": 22242
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.108094513663677e-09,
+      "loss": 0.4253,
+      "step": 22243
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.053944168198268e-09,
+      "loss": 0.4468,
+      "step": 22244
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.000000801408169e-09,
+      "loss": 0.4602,
+      "step": 22245
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.9462644144124845e-09,
+      "loss": 0.4412,
+      "step": 22246
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.892735008323659e-09,
+      "loss": 0.4729,
+      "step": 22247
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.839412584250804e-09,
+      "loss": 0.4851,
+      "step": 22248
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.786297143298592e-09,
+      "loss": 0.4224,
+      "step": 22249
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.733388686566145e-09,
+      "loss": 0.4692,
+      "step": 22250
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.68068721514925e-09,
+      "loss": 0.4412,
+      "step": 22251
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.6281927301414806e-09,
+      "loss": 0.4514,
+      "step": 22252
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.575905232627522e-09,
+      "loss": 0.4255,
+      "step": 22253
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.523824723692063e-09,
+      "loss": 0.4441,
+      "step": 22254
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.4719512044131295e-09,
+      "loss": 0.4526,
+      "step": 22255
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.420284675865418e-09,
+      "loss": 0.4712,
+      "step": 22256
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.368825139119184e-09,
+      "loss": 0.5024,
+      "step": 22257
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.31757259524024e-09,
+      "loss": 0.4673,
+      "step": 22258
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.2665270452899606e-09,
+      "loss": 0.4666,
+      "step": 22259
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.215688490325278e-09,
+      "loss": 0.4294,
+      "step": 22260
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.1650569313997935e-09,
+      "loss": 0.4768,
+      "step": 22261
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.1146323695615575e-09,
+      "loss": 0.4824,
+      "step": 22262
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.0644148058564e-09,
+      "loss": 0.5002,
+      "step": 22263
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.014404241323491e-09,
+      "loss": 0.4348,
+      "step": 22264
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.9646006769975565e-09,
+      "loss": 0.4612,
+      "step": 22265
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.915004113912215e-09,
+      "loss": 0.4729,
+      "step": 22266
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.865614553093313e-09,
+      "loss": 0.4714,
+      "step": 22267
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.816431995564476e-09,
+      "loss": 0.4905,
+      "step": 22268
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.767456442344887e-09,
+      "loss": 0.4309,
+      "step": 22269
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.7186878944481825e-09,
+      "loss": 0.4265,
+      "step": 22270
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.6701263528835535e-09,
+      "loss": 0.4395,
+      "step": 22271
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.621771818659083e-09,
+      "loss": 0.4727,
+      "step": 22272
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.573624292773971e-09,
+      "loss": 0.4041,
+      "step": 22273
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.525683776227419e-09,
+      "loss": 0.4343,
+      "step": 22274
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.477950270011967e-09,
+      "loss": 0.4121,
+      "step": 22275
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.430423775114601e-09,
+      "loss": 0.4869,
+      "step": 22276
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.383104292522312e-09,
+      "loss": 0.4161,
+      "step": 22277
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.335991823213205e-09,
+      "loss": 0.5344,
+      "step": 22278
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.289086368163165e-09,
+      "loss": 0.4294,
+      "step": 22279
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.24238792834586e-09,
+      "loss": 0.4734,
+      "step": 22280
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.195896504726072e-09,
+      "loss": 0.5195,
+      "step": 22281
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.149612098267475e-09,
+      "loss": 0.4673,
+      "step": 22282
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.1035347099304136e-09,
+      "loss": 0.4373,
+      "step": 22283
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.0576643406663465e-09,
+      "loss": 0.4275,
+      "step": 22284
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.012000991427845e-09,
+      "loss": 0.5439,
+      "step": 22285
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.966544663158601e-09,
+      "loss": 0.4749,
+      "step": 22286
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.921295356802303e-09,
+      "loss": 0.4087,
+      "step": 22287
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.876253073293757e-09,
+      "loss": 0.4504,
+      "step": 22288
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.831417813568884e-09,
+      "loss": 0.4281,
+      "step": 22289
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.786789578553608e-09,
+      "loss": 0.4653,
+      "step": 22290
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.742368369173855e-09,
+      "loss": 0.3867,
+      "step": 22291
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.698154186348891e-09,
+      "loss": 0.4165,
+      "step": 22292
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.654147030994649e-09,
+      "loss": 0.4902,
+      "step": 22293
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.6103469040237326e-09,
+      "loss": 0.4668,
+      "step": 22294
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.566753806342084e-09,
+      "loss": 0.4229,
+      "step": 22295
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.5233677388523135e-09,
+      "loss": 0.4326,
+      "step": 22296
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.480188702453703e-09,
+      "loss": 0.4265,
+      "step": 22297
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.4372166980410914e-09,
+      "loss": 0.4634,
+      "step": 22298
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.394451726504878e-09,
+      "loss": 0.4639,
+      "step": 22299
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.3518937887288005e-09,
+      "loss": 0.4207,
+      "step": 22300
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.309542885595486e-09,
+      "loss": 0.4324,
+      "step": 22301
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.26739901798312e-09,
+      "loss": 0.4268,
+      "step": 22302
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.22546218676323e-09,
+      "loss": 0.4734,
+      "step": 22303
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.183732392804007e-09,
+      "loss": 0.4473,
+      "step": 22304
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.142209636971428e-09,
+      "loss": 0.4548,
+      "step": 22305
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1008939201248046e-09,
+      "loss": 0.4033,
+      "step": 22306
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.059785243120118e-09,
+      "loss": 0.4338,
+      "step": 22307
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.0188836068077994e-09,
+      "loss": 0.4797,
+      "step": 22308
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.97818901203606e-09,
+      "loss": 0.4407,
+      "step": 22309
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.937701459647558e-09,
+      "loss": 0.4824,
+      "step": 22310
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.897420950480513e-09,
+      "loss": 0.4155,
+      "step": 22311
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.857347485369811e-09,
+      "loss": 0.4673,
+      "step": 22312
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.8174810651459e-09,
+      "loss": 0.4253,
+      "step": 22313
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.777821690632566e-09,
+      "loss": 0.4539,
+      "step": 22314
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.738369362653593e-09,
+      "loss": 0.4524,
+      "step": 22315
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.699124082023886e-09,
+      "loss": 0.4316,
+      "step": 22316
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.660085849558348e-09,
+      "loss": 0.5664,
+      "step": 22317
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6212546660641113e-09,
+      "loss": 0.4736,
+      "step": 22318
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5826305323471975e-09,
+      "loss": 0.3733,
+      "step": 22319
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5442134492047477e-09,
+      "loss": 0.4412,
+      "step": 22320
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.506003417435011e-09,
+      "loss": 0.4373,
+      "step": 22321
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.4680004378295772e-09,
+      "loss": 0.426,
+      "step": 22322
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.430204511173374e-09,
+      "loss": 0.4167,
+      "step": 22323
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3926156382502184e-09,
+      "loss": 0.3855,
+      "step": 22324
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3552338198394875e-09,
+      "loss": 0.3986,
+      "step": 22325
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.318059056715006e-09,
+      "loss": 0.4458,
+      "step": 22326
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.2810913496461593e-09,
+      "loss": 0.4524,
+      "step": 22327
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.244330699399001e-09,
+      "loss": 0.4143,
+      "step": 22328
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.207777106736254e-09,
+      "loss": 0.469,
+      "step": 22329
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.17143057241287e-09,
+      "loss": 0.4358,
+      "step": 22330
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.1352910971826913e-09,
+      "loss": 0.4089,
+      "step": 22331
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.099358681794007e-09,
+      "loss": 0.468,
+      "step": 22332
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.063633326990667e-09,
+      "loss": 0.4521,
+      "step": 22333
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.0281150335143005e-09,
+      "loss": 0.4272,
+      "step": 22334
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.992803802099875e-09,
+      "loss": 0.4861,
+      "step": 22335
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.9576996334768072e-09,
+      "loss": 0.4568,
+      "step": 22336
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.9228025283745134e-09,
+      "loss": 0.421,
+      "step": 22337
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.8881124875157483e-09,
+      "loss": 0.4041,
+      "step": 22338
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.853629511617717e-09,
+      "loss": 0.4141,
+      "step": 22339
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.8193536013954025e-09,
+      "loss": 0.4836,
+      "step": 22340
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.7852847575582375e-09,
+      "loss": 0.4631,
+      "step": 22341
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.7514229808123237e-09,
+      "loss": 0.4016,
+      "step": 22342
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.7177682718593223e-09,
+      "loss": 0.4031,
+      "step": 22343
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.684320631395343e-09,
+      "loss": 0.4609,
+      "step": 22344
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.651080060114275e-09,
+      "loss": 0.4353,
+      "step": 22345
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.618046558704457e-09,
+      "loss": 0.3816,
+      "step": 22346
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5852201278486755e-09,
+      "loss": 0.3878,
+      "step": 22347
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.552600768228608e-09,
+      "loss": 0.4458,
+      "step": 22348
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.52018848051927e-09,
+      "loss": 0.5076,
+      "step": 22349
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.487983265391236e-09,
+      "loss": 0.446,
+      "step": 22350
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4559851235128608e-09,
+      "loss": 0.4575,
+      "step": 22351
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4241940555458367e-09,
+      "loss": 0.4626,
+      "step": 22352
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3926100621496364e-09,
+      "loss": 0.4255,
+      "step": 22353
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3612331439770708e-09,
+      "loss": 0.522,
+      "step": 22354
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3300633016798412e-09,
+      "loss": 0.4485,
+      "step": 22355
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.299100535901877e-09,
+      "loss": 0.4534,
+      "step": 22356
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2683448472848867e-09,
+      "loss": 0.4121,
+      "step": 22357
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.237796236467249e-09,
+      "loss": 0.431,
+      "step": 22358
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.20745470407957e-09,
+      "loss": 0.4475,
+      "step": 22359
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.177320250751347e-09,
+      "loss": 0.4819,
+      "step": 22360
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1473928771076347e-09,
+      "loss": 0.425,
+      "step": 22361
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1176725837668276e-09,
+      "loss": 0.4465,
+      "step": 22362
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0881593713450997e-09,
+      "loss": 0.469,
+      "step": 22363
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.058853240454184e-09,
+      "loss": 0.4102,
+      "step": 22364
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0297541917002617e-09,
+      "loss": 0.4395,
+      "step": 22365
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0008622256861843e-09,
+      "loss": 0.4756,
+      "step": 22366
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.972177343010362e-09,
+      "loss": 0.4082,
+      "step": 22367
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.943699544267874e-09,
+      "loss": 0.4092,
+      "step": 22368
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9154288300482493e-09,
+      "loss": 0.4954,
+      "step": 22369
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8873652009354647e-09,
+      "loss": 0.4985,
+      "step": 22370
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8595086575123876e-09,
+      "loss": 0.4093,
+      "step": 22371
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.831859200356334e-09,
+      "loss": 0.4382,
+      "step": 22372
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8044168300390686e-09,
+      "loss": 0.4482,
+      "step": 22373
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7771815471290254e-09,
+      "loss": 0.4683,
+      "step": 22374
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7501533521901981e-09,
+      "loss": 0.4823,
+      "step": 22375
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7233322457821388e-09,
+      "loss": 0.438,
+      "step": 22376
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6967182284621797e-09,
+      "loss": 0.4534,
+      "step": 22377
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.670311300779881e-09,
+      "loss": 0.4263,
+      "step": 22378
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6441114632825828e-09,
+      "loss": 0.4692,
+      "step": 22379
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6181187165120737e-09,
+      "loss": 0.4229,
+      "step": 22380
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5923330610090326e-09,
+      "loss": 0.498,
+      "step": 22381
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5667544973052563e-09,
+      "loss": 0.4431,
+      "step": 22382
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.541383025931431e-09,
+      "loss": 0.458,
+      "step": 22383
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.516218647413803e-09,
+      "loss": 0.4021,
+      "step": 22384
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4912613622719563e-09,
+      "loss": 0.4263,
+      "step": 22385
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4665111710243651e-09,
+      "loss": 0.4658,
+      "step": 22386
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4419680741828423e-09,
+      "loss": 0.5193,
+      "step": 22387
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4176320722558701e-09,
+      "loss": 0.5781,
+      "step": 22388
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3935031657474895e-09,
+      "loss": 0.4692,
+      "step": 22389
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.369581355157301e-09,
+      "loss": 0.5042,
+      "step": 22390
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3458666409815746e-09,
+      "loss": 0.4915,
+      "step": 22391
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3223590237099182e-09,
+      "loss": 0.4229,
+      "step": 22392
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2990585038308302e-09,
+      "loss": 0.4246,
+      "step": 22393
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2759650818272574e-09,
+      "loss": 0.4941,
+      "step": 22394
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2530787581754855e-09,
+      "loss": 0.4126,
+      "step": 22395
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2303995333506903e-09,
+      "loss": 0.4634,
+      "step": 22396
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.207927407823606e-09,
+      "loss": 0.4258,
+      "step": 22397
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1856623820571956e-09,
+      "loss": 0.4631,
+      "step": 22398
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1636044565155325e-09,
+      "loss": 0.3977,
+      "step": 22399
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1417536316526977e-09,
+      "loss": 0.5229,
+      "step": 22400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1201099079238831e-09,
+      "loss": 0.4856,
+      "step": 22401
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0986732857753978e-09,
+      "loss": 0.4365,
+      "step": 22402
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0774437656524417e-09,
+      "loss": 0.4548,
+      "step": 22403
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0564213479935526e-09,
+      "loss": 0.3921,
+      "step": 22404
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0356060332361584e-09,
+      "loss": 0.4241,
+      "step": 22405
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0149978218088053e-09,
+      "loss": 0.4333,
+      "step": 22406
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.945967141400393e-10,
+      "loss": 0.4258,
+      "step": 22407
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.744027106528553e-10,
+      "loss": 0.4404,
+      "step": 22408
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.544158117635872e-10,
+      "loss": 0.4341,
+      "step": 22409
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.34636017887458e-10,
+      "loss": 0.479,
+      "step": 22410
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.150633294341404e-10,
+      "loss": 0.4514,
+      "step": 22411
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.956977468088656e-10,
+      "loss": 0.4438,
+      "step": 22412
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.765392704135345e-10,
+      "loss": 0.4692,
+      "step": 22413
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.575879006433862e-10,
+      "loss": 0.4832,
+      "step": 22414
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.388436378925502e-10,
+      "loss": 0.428,
+      "step": 22415
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.203064825473838e-10,
+      "loss": 0.4609,
+      "step": 22416
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.019764349942449e-10,
+      "loss": 0.5039,
+      "step": 22417
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.838534956106092e-10,
+      "loss": 0.4119,
+      "step": 22418
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.659376647728422e-10,
+      "loss": 0.4346,
+      "step": 22419
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.482289428528688e-10,
+      "loss": 0.4282,
+      "step": 22420
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.307273302148421e-10,
+      "loss": 0.4497,
+      "step": 22421
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.134328272240254e-10,
+      "loss": 0.4207,
+      "step": 22422
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.963454342379106e-10,
+      "loss": 0.5059,
+      "step": 22423
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.794651516095485e-10,
+      "loss": 0.4905,
+      "step": 22424
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.627919796886595e-10,
+      "loss": 0.4597,
+      "step": 22425
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.463259188216331e-10,
+      "loss": 0.4722,
+      "step": 22426
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.300669693493078e-10,
+      "loss": 0.5425,
+      "step": 22427
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.140151316069709e-10,
+      "loss": 0.4675,
+      "step": 22428
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.981704059287996e-10,
+      "loss": 0.4021,
+      "step": 22429
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.825327926423096e-10,
+      "loss": 0.4268,
+      "step": 22430
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.671022920716863e-10,
+      "loss": 0.4261,
+      "step": 22431
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.518789045366734e-10,
+      "loss": 0.4585,
+      "step": 22432
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.368626303514646e-10,
+      "loss": 0.4668,
+      "step": 22433
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.220534698280322e-10,
+      "loss": 0.481,
+      "step": 22434
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.074514232727978e-10,
+      "loss": 0.4778,
+      "step": 22435
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.930564909877422e-10,
+      "loss": 0.4243,
+      "step": 22436
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.788686732726255e-10,
+      "loss": 0.3914,
+      "step": 22437
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.6488797041943645e-10,
+      "loss": 0.4377,
+      "step": 22438
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.511143827190534e-10,
+      "loss": 0.4497,
+      "step": 22439
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.3754791045569344e-10,
+      "loss": 0.4983,
+      "step": 22440
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.241885539124635e-10,
+      "loss": 0.4319,
+      "step": 22441
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.1103631336358863e-10,
+      "loss": 0.4663,
+      "step": 22442
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.980911890821837e-10,
+      "loss": 0.3953,
+      "step": 22443
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.8535318133692267e-10,
+      "loss": 0.4641,
+      "step": 22444
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.7282229039092843e-10,
+      "loss": 0.3887,
+      "step": 22445
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.6049851650510336e-10,
+      "loss": 0.4727,
+      "step": 22446
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.483818599325783e-10,
+      "loss": 0.4001,
+      "step": 22447
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.364723209264842e-10,
+      "loss": 0.458,
+      "step": 22448
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.247698997321802e-10,
+      "loss": 0.4434,
+      "step": 22449
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.1327459659280523e-10,
+      "loss": 0.4449,
+      "step": 22450
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.01986411745947e-10,
+      "loss": 0.4656,
+      "step": 22451
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.9090534542475236e-10,
+      "loss": 0.5374,
+      "step": 22452
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.8003139786014765e-10,
+      "loss": 0.3936,
+      "step": 22453
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.69364569276398e-10,
+      "loss": 0.4463,
+      "step": 22454
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.589048598955479e-10,
+      "loss": 0.5193,
+      "step": 22455
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.4865226993298074e-10,
+      "loss": 0.5022,
+      "step": 22456
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.3860679960074906e-10,
+      "loss": 0.4695,
+      "step": 22457
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.28768449108685e-10,
+      "loss": 0.4939,
+      "step": 22458
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.191372186599594e-10,
+      "loss": 0.4954,
+      "step": 22459
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0971310845219196e-10,
+      "loss": 0.4519,
+      "step": 22460
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0049611868300233e-10,
+      "loss": 0.4155,
+      "step": 22461
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.914862495422387e-10,
+      "loss": 0.4585,
+      "step": 22462
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.8268350121641853e-10,
+      "loss": 0.5022,
+      "step": 22463
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.740878738887286e-10,
+      "loss": 0.4514,
+      "step": 22464
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6569936773680463e-10,
+      "loss": 0.4233,
+      "step": 22465
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5751798293273113e-10,
+      "loss": 0.4065,
+      "step": 22466
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4954371964859272e-10,
+      "loss": 0.4663,
+      "step": 22467
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4177657804759215e-10,
+      "loss": 0.4431,
+      "step": 22468
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.34216558291822e-10,
+      "loss": 0.4744,
+      "step": 22469
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2686366053782372e-10,
+      "loss": 0.416,
+      "step": 22470
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1971788493658765e-10,
+      "loss": 0.4272,
+      "step": 22471
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1277923163799387e-10,
+      "loss": 0.4425,
+      "step": 22472
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0604770078415094e-10,
+      "loss": 0.4067,
+      "step": 22473
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.952329251494697e-11,
+      "loss": 0.4731,
+      "step": 22474
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.320600696582915e-11,
+      "loss": 0.4475,
+      "step": 22475
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.709584426780381e-11,
+      "loss": 0.4431,
+      "step": 22476
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.119280454632617e-11,
+      "loss": 0.4124,
+      "step": 22477
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.549688792574116e-11,
+      "loss": 0.4038,
+      "step": 22478
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.000809452151203e-11,
+      "loss": 0.4377,
+      "step": 22479
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.472642444910193e-11,
+      "loss": 0.4604,
+      "step": 22480
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.965187781731275e-11,
+      "loss": 0.4368,
+      "step": 22481
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.4784454731615644e-11,
+      "loss": 0.4631,
+      "step": 22482
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.012415529193071e-11,
+      "loss": 0.4397,
+      "step": 22483
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.567097959595757e-11,
+      "loss": 0.3743,
+      "step": 22484
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.1424927735844724e-11,
+      "loss": 0.46,
+      "step": 22485
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.738599979818958e-11,
+      "loss": 0.4478,
+      "step": 22486
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3554195867369074e-11,
+      "loss": 0.4434,
+      "step": 22487
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.992951602331928e-11,
+      "loss": 0.4104,
+      "step": 22488
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.6511960340425135e-11,
+      "loss": 0.4055,
+      "step": 22489
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.330152888974091e-11,
+      "loss": 0.4446,
+      "step": 22490
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0298221737879985e-11,
+      "loss": 0.4829,
+      "step": 22491
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.7502038947014854e-11,
+      "loss": 0.4502,
+      "step": 22492
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.491298057376689e-11,
+      "loss": 0.4578,
+      "step": 22493
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.2531046674757463e-11,
+      "loss": 0.4885,
+      "step": 22494
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0356237296615946e-11,
+      "loss": 0.4045,
+      "step": 22495
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.38855248486148e-12,
+      "loss": 0.4771,
+      "step": 22496
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.6279922816825385e-12,
+      "loss": 0.4509,
+      "step": 22497
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.074556721496038e-12,
+      "loss": 0.4568,
+      "step": 22498
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.7282458387188914e-12,
+      "loss": 0.4092,
+      "step": 22499
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.589059658886228e-12,
+      "loss": 0.4067,
+      "step": 22500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6569982075331781e-12,
+      "loss": 0.4697,
+      "step": 22501
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.320615035335322e-13,
+      "loss": 0.5095,
+      "step": 22502
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.142495602099672e-13,
+      "loss": 0.4558,
+      "step": 22503
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0356239088515908e-13,
+      "loss": 0.366,
+      "step": 22504
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.406,
+      "step": 22505
+    },
+    {
+      "epoch": 5.0,
+      "eval_cIoU": 0.8030166625976562,
+      "eval_gIoU": 0.8004721999168396,
+      "eval_runtime": 120.467,
+      "eval_samples_per_second": 31.635,
+      "eval_steps_per_second": 3.96,
+      "step": 22505
+    },
+    {
+      "epoch": 5.0,
+      "step": 22505,
+      "total_flos": 6.612604947001246e+19,
+      "train_loss": 0.801273219666463,
+      "train_runtime": 198092.5505,
+      "train_samples_per_second": 14.54,
+      "train_steps_per_second": 0.114
+    }
+  ],
+  "max_steps": 22505,
+  "num_train_epochs": 5,
+  "total_flos": 6.612604947001246e+19,
+  "trial_name": null,
+  "trial_params": null
+}