diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.4435573297848747,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2177866489243734e-08,
+      "loss": 2.8313,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.435573297848747e-08,
+      "loss": 2.7487,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.653359946773119e-08,
+      "loss": 2.8504,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.871146595697494e-08,
+      "loss": 2.902,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1088933244621866e-07,
+      "loss": 2.7682,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3306719893546238e-07,
+      "loss": 2.8621,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5524506542470611e-07,
+      "loss": 2.9124,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7742293191394987e-07,
+      "loss": 2.7847,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.996007984031936e-07,
+      "loss": 2.8189,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2177866489243732e-07,
+      "loss": 2.8245,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.439565313816811e-07,
+      "loss": 2.841,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6613439787092476e-07,
+      "loss": 2.7664,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8831226436016855e-07,
+      "loss": 2.888,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1049013084941223e-07,
+      "loss": 2.7292,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.32667997338656e-07,
+      "loss": 2.7483,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5484586382789975e-07,
+      "loss": 2.7987,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.770237303171435e-07,
+      "loss": 2.6385,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.992015968063872e-07,
+      "loss": 2.4894,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2137946329563095e-07,
+      "loss": 2.641,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4355732978487463e-07,
+      "loss": 2.4582,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6573519627411837e-07,
+      "loss": 2.4435,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.879130627633622e-07,
+      "loss": 2.3291,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.100909292526058e-07,
+      "loss": 2.2823,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.322687957418495e-07,
+      "loss": 2.2867,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.544466622310934e-07,
+      "loss": 2.2412,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.766245287203371e-07,
+      "loss": 2.0598,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.988023952095807e-07,
+      "loss": 1.9862,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.209802616988245e-07,
+      "loss": 1.9683,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.431581281880682e-07,
+      "loss": 2.0052,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.65335994677312e-07,
+      "loss": 1.8498,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.875138611665558e-07,
+      "loss": 1.8486,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.096917276557995e-07,
+      "loss": 1.988,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.318695941450432e-07,
+      "loss": 1.8465,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.54047460634287e-07,
+      "loss": 1.8648,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.762253271235307e-07,
+      "loss": 1.8335,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.984031936127744e-07,
+      "loss": 1.7241,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205810601020182e-07,
+      "loss": 1.7742,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.427589265912619e-07,
+      "loss": 1.8215,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.649367930805055e-07,
+      "loss": 1.7852,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.871146595697493e-07,
+      "loss": 1.7705,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.09292526058993e-07,
+      "loss": 1.6308,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.314703925482367e-07,
+      "loss": 1.7693,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.536482590374805e-07,
+      "loss": 1.6551,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.758261255267243e-07,
+      "loss": 1.5075,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98003992015968e-07,
+      "loss": 1.4972,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0201818585052116e-06,
+      "loss": 1.5363,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0423597249944554e-06,
+      "loss": 1.4753,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.064537591483699e-06,
+      "loss": 1.4665,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0867154579729429e-06,
+      "loss": 1.4726,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1088933244621867e-06,
+      "loss": 1.4492,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1310711909514303e-06,
+      "loss": 1.4465,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1532490574406742e-06,
+      "loss": 1.4425,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1754269239299178e-06,
+      "loss": 1.4439,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1976047904191614e-06,
+      "loss": 1.3326,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2197826569084053e-06,
+      "loss": 1.3006,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.241960523397649e-06,
+      "loss": 1.373,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2641383898868928e-06,
+      "loss": 1.3718,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2863162563761364e-06,
+      "loss": 1.2873,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3084941228653802e-06,
+      "loss": 1.2811,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.330671989354624e-06,
+      "loss": 1.2886,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3528498558438677e-06,
+      "loss": 1.3269,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3750277223331115e-06,
+      "loss": 1.25,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3972055888223552e-06,
+      "loss": 1.1905,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.419383455311599e-06,
+      "loss": 1.2987,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4415613218008426e-06,
+      "loss": 1.1388,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4637391882900865e-06,
+      "loss": 1.2263,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.48591705477933e-06,
+      "loss": 1.1782,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.508094921268574e-06,
+      "loss": 1.1724,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5302727877578176e-06,
+      "loss": 1.1999,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5524506542470614e-06,
+      "loss": 1.2571,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.574628520736305e-06,
+      "loss": 1.1427,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5968063872255489e-06,
+      "loss": 1.2122,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6189842537147925e-06,
+      "loss": 1.0237,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6411621202040363e-06,
+      "loss": 1.116,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6633399866932798e-06,
+      "loss": 1.2039,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6855178531825238e-06,
+      "loss": 1.1293,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7076957196717672e-06,
+      "loss": 1.1776,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.729873586161011e-06,
+      "loss": 1.061,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7520514526502551e-06,
+      "loss": 1.148,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7742293191394985e-06,
+      "loss": 1.0244,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7964071856287424e-06,
+      "loss": 1.0787,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.818585052117986e-06,
+      "loss": 1.0696,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8407629186072298e-06,
+      "loss": 1.0233,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8629407850964735e-06,
+      "loss": 1.1158,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8851186515857173e-06,
+      "loss": 1.061,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.907296518074961e-06,
+      "loss": 1.0074,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.929474384564205e-06,
+      "loss": 1.0662,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9516522510534486e-06,
+      "loss": 0.9835,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9738301175426923e-06,
+      "loss": 1.1022,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996007984031936e-06,
+      "loss": 1.0651,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.01818585052118e-06,
+      "loss": 1.0127,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.040363717010423e-06,
+      "loss": 1.0198,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.062541583499667e-06,
+      "loss": 1.0291,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.084719449988911e-06,
+      "loss": 1.0236,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1068973164781544e-06,
+      "loss": 1.0818,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.129075182967398e-06,
+      "loss": 0.9517,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.151253049456642e-06,
+      "loss": 1.0331,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1734309159458858e-06,
+      "loss": 0.981,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1956087824351294e-06,
+      "loss": 0.8392,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2177866489243734e-06,
+      "loss": 0.9658,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.239964515413617e-06,
+      "loss": 0.9179,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2621423819028607e-06,
+      "loss": 1.021,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2843202483921043e-06,
+      "loss": 0.9978,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3064981148813484e-06,
+      "loss": 0.9548,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.328675981370592e-06,
+      "loss": 0.965,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3508538478598356e-06,
+      "loss": 0.8066,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3730317143490797e-06,
+      "loss": 0.9906,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.395209580838323e-06,
+      "loss": 0.9835,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.417387447327567e-06,
+      "loss": 0.9315,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4395653138168106e-06,
+      "loss": 0.9492,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4617431803060546e-06,
+      "loss": 0.932,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.483921046795298e-06,
+      "loss": 0.9623,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.506098913284542e-06,
+      "loss": 0.9875,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5282767797737855e-06,
+      "loss": 0.8434,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5504546462630296e-06,
+      "loss": 0.8793,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5726325127522728e-06,
+      "loss": 0.9128,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.594810379241517e-06,
+      "loss": 0.9344,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6169882457307604e-06,
+      "loss": 0.9646,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6391661122200045e-06,
+      "loss": 0.9398,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.661343978709248e-06,
+      "loss": 0.8822,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6835218451984913e-06,
+      "loss": 0.8302,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7056997116877354e-06,
+      "loss": 0.9665,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7278775781769794e-06,
+      "loss": 0.6695,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.750055444666223e-06,
+      "loss": 0.9326,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7722333111554663e-06,
+      "loss": 0.8981,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7944111776447103e-06,
+      "loss": 0.7882,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.816589044133954e-06,
+      "loss": 0.8513,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.838766910623198e-06,
+      "loss": 0.8556,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.860944777112441e-06,
+      "loss": 0.8452,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8831226436016853e-06,
+      "loss": 0.7643,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.905300510090929e-06,
+      "loss": 0.8394,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.927478376580173e-06,
+      "loss": 0.8253,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.949656243069416e-06,
+      "loss": 0.7562,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.97183410955866e-06,
+      "loss": 0.8059,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994011976047904e-06,
+      "loss": 0.7796,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.016189842537148e-06,
+      "loss": 0.8027,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0383677090263915e-06,
+      "loss": 0.8394,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.060545575515635e-06,
+      "loss": 0.8452,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0827234420048788e-06,
+      "loss": 0.7442,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.104901308494123e-06,
+      "loss": 0.8402,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1270791749833664e-06,
+      "loss": 0.8145,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.14925704147261e-06,
+      "loss": 0.8235,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1714349079618537e-06,
+      "loss": 0.8113,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1936127744510977e-06,
+      "loss": 0.7299,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2157906409403414e-06,
+      "loss": 0.7867,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.237968507429585e-06,
+      "loss": 0.7841,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2601463739188286e-06,
+      "loss": 0.783,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2823242404080727e-06,
+      "loss": 0.8421,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3045021068973163e-06,
+      "loss": 0.7819,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3266799733865595e-06,
+      "loss": 0.8405,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3488578398758036e-06,
+      "loss": 0.8994,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3710357063650476e-06,
+      "loss": 0.7685,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3932135728542912e-06,
+      "loss": 0.8166,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4153914393435345e-06,
+      "loss": 0.7238,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4375693058327785e-06,
+      "loss": 0.7702,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.459747172322022e-06,
+      "loss": 0.7955,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.481925038811266e-06,
+      "loss": 0.6396,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5041029053005102e-06,
+      "loss": 0.707,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5262807717897534e-06,
+      "loss": 0.765,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.548458638278997e-06,
+      "loss": 0.8138,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.570636504768241e-06,
+      "loss": 0.7156,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5928143712574848e-06,
+      "loss": 0.6935,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6149922377467284e-06,
+      "loss": 0.7004,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.637170104235972e-06,
+      "loss": 0.6829,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.659347970725216e-06,
+      "loss": 0.7142,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6815258372144597e-06,
+      "loss": 0.6585,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7037037037037033e-06,
+      "loss": 0.7126,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.725881570192947e-06,
+      "loss": 0.7424,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.748059436682191e-06,
+      "loss": 0.6921,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7702373031714346e-06,
+      "loss": 0.7613,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7924151696606783e-06,
+      "loss": 0.6666,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.814593036149922e-06,
+      "loss": 0.5887,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.836770902639166e-06,
+      "loss": 0.6462,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.85894876912841e-06,
+      "loss": 0.7045,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.881126635617653e-06,
+      "loss": 0.708,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.903304502106897e-06,
+      "loss": 0.7253,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9254823685961404e-06,
+      "loss": 0.6908,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9476602350853845e-06,
+      "loss": 0.7164,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9698381015746286e-06,
+      "loss": 0.6964,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.992015968063872e-06,
+      "loss": 0.754,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.014193834553116e-06,
+      "loss": 0.6646,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.03637170104236e-06,
+      "loss": 0.7009,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.058549567531603e-06,
+      "loss": 0.7587,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.080727434020846e-06,
+      "loss": 0.7198,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.10290530051009e-06,
+      "loss": 0.7272,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.125083166999334e-06,
+      "loss": 0.7187,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1472610334885784e-06,
+      "loss": 0.7017,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.169438899977822e-06,
+      "loss": 0.6305,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.191616766467066e-06,
+      "loss": 0.693,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.213794632956309e-06,
+      "loss": 0.6175,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.235972499445553e-06,
+      "loss": 0.6278,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.258150365934796e-06,
+      "loss": 0.5923,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.28032823242404e-06,
+      "loss": 0.6659,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.302506098913284e-06,
+      "loss": 0.5381,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.324683965402528e-06,
+      "loss": 0.5334,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3468618318917715e-06,
+      "loss": 0.6919,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3690396983810156e-06,
+      "loss": 0.6283,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.391217564870259e-06,
+      "loss": 0.6371,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.413395431359503e-06,
+      "loss": 0.6531,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.435573297848747e-06,
+      "loss": 0.6319,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.45775116433799e-06,
+      "loss": 0.6143,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.479929030827234e-06,
+      "loss": 0.5262,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.502106897316478e-06,
+      "loss": 0.6664,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.524284763805721e-06,
+      "loss": 0.5916,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5464626302949654e-06,
+      "loss": 0.601,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.568640496784209e-06,
+      "loss": 0.5858,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.590818363273453e-06,
+      "loss": 0.5855,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.612996229762697e-06,
+      "loss": 0.6645,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.63517409625194e-06,
+      "loss": 0.6118,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.657351962741184e-06,
+      "loss": 0.6421,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.679529829230428e-06,
+      "loss": 0.6102,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.701707695719671e-06,
+      "loss": 0.624,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.723885562208915e-06,
+      "loss": 0.5834,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.746063428698159e-06,
+      "loss": 0.5613,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.768241295187403e-06,
+      "loss": 0.5491,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.790419161676646e-06,
+      "loss": 0.607,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.81259702816589e-06,
+      "loss": 0.6186,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.834774894655134e-06,
+      "loss": 0.6103,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.856952761144377e-06,
+      "loss": 0.5804,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.879130627633621e-06,
+      "loss": 0.5932,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.901308494122865e-06,
+      "loss": 0.6535,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923486360612109e-06,
+      "loss": 0.6241,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.945664227101353e-06,
+      "loss": 0.5673,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967842093590596e-06,
+      "loss": 0.6156,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99001996007984e-06,
+      "loss": 0.625,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.012197826569084e-06,
+      "loss": 0.5592,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.034375693058327e-06,
+      "loss": 0.6498,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.056553559547571e-06,
+      "loss": 0.5579,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.078731426036815e-06,
+      "loss": 0.527,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.100909292526059e-06,
+      "loss": 0.6242,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.123087159015302e-06,
+      "loss": 0.5667,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1452650255045455e-06,
+      "loss": 0.6074,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1674428919937896e-06,
+      "loss": 0.574,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.189620758483034e-06,
+      "loss": 0.6063,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.211798624972277e-06,
+      "loss": 0.6181,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.233976491461521e-06,
+      "loss": 0.5447,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.256154357950765e-06,
+      "loss": 0.5969,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.278332224440009e-06,
+      "loss": 0.5981,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.300510090929252e-06,
+      "loss": 0.6974,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.322687957418496e-06,
+      "loss": 0.6201,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3448658239077394e-06,
+      "loss": 0.6023,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.367043690396983e-06,
+      "loss": 0.5582,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.389221556886227e-06,
+      "loss": 0.5824,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.411399423375471e-06,
+      "loss": 0.5813,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.433577289864715e-06,
+      "loss": 0.5636,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.455755156353959e-06,
+      "loss": 0.5541,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.477933022843202e-06,
+      "loss": 0.5868,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.500110889332446e-06,
+      "loss": 0.6396,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.522288755821689e-06,
+      "loss": 0.5673,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5444666223109325e-06,
+      "loss": 0.5009,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.566644488800177e-06,
+      "loss": 0.6358,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.588822355289421e-06,
+      "loss": 0.5676,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.611000221778665e-06,
+      "loss": 0.5779,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.633178088267908e-06,
+      "loss": 0.5558,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.655355954757152e-06,
+      "loss": 0.5312,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.677533821246396e-06,
+      "loss": 0.5788,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.69971168773564e-06,
+      "loss": 0.5527,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.721889554224882e-06,
+      "loss": 0.6031,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.7440674207141265e-06,
+      "loss": 0.5611,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.7662452872033705e-06,
+      "loss": 0.5636,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.7884231536926146e-06,
+      "loss": 0.5614,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.810601020181858e-06,
+      "loss": 0.5675,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.832778886671102e-06,
+      "loss": 0.5268,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.854956753160346e-06,
+      "loss": 0.5917,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.87713461964959e-06,
+      "loss": 0.5171,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.899312486138832e-06,
+      "loss": 0.5676,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.921490352628076e-06,
+      "loss": 0.584,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.94366821911732e-06,
+      "loss": 0.5108,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9658460856065644e-06,
+      "loss": 0.5572,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988023952095808e-06,
+      "loss": 0.5271,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.010201818585052e-06,
+      "loss": 0.5457,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.032379685074296e-06,
+      "loss": 0.5845,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.05455755156354e-06,
+      "loss": 0.4928,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.076735418052783e-06,
+      "loss": 0.5059,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.098913284542026e-06,
+      "loss": 0.5587,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.12109115103127e-06,
+      "loss": 0.5984,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1432690175205135e-06,
+      "loss": 0.5167,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1654468840097575e-06,
+      "loss": 0.5177,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1876247504990016e-06,
+      "loss": 0.5412,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.209802616988246e-06,
+      "loss": 0.5236,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.23198048347749e-06,
+      "loss": 0.5421,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.254158349966733e-06,
+      "loss": 0.6238,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.276336216455976e-06,
+      "loss": 0.6297,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.29851408294522e-06,
+      "loss": 0.5002,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.320691949434463e-06,
+      "loss": 0.6187,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.342869815923707e-06,
+      "loss": 0.5069,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.3650476824129514e-06,
+      "loss": 0.5452,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.3872255489021955e-06,
+      "loss": 0.4812,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.409403415391439e-06,
+      "loss": 0.5877,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.431581281880683e-06,
+      "loss": 0.5552,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.453759148369926e-06,
+      "loss": 0.508,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.47593701485917e-06,
+      "loss": 0.4953,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.498114881348413e-06,
+      "loss": 0.5056,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.520292747837657e-06,
+      "loss": 0.5024,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.542470614326901e-06,
+      "loss": 0.5926,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.564648480816145e-06,
+      "loss": 0.5195,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5868263473053886e-06,
+      "loss": 0.5041,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.609004213794633e-06,
+      "loss": 0.5387,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.631182080283877e-06,
+      "loss": 0.5171,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.653359946773119e-06,
+      "loss": 0.5112,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.675537813262363e-06,
+      "loss": 0.5271,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.697715679751607e-06,
+      "loss": 0.5658,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.719893546240851e-06,
+      "loss": 0.5382,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.742071412730095e-06,
+      "loss": 0.4593,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.7642492792193384e-06,
+      "loss": 0.5251,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.7864271457085825e-06,
+      "loss": 0.5381,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.8086050121978265e-06,
+      "loss": 0.5517,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.830782878687069e-06,
+      "loss": 0.5353,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.852960745176313e-06,
+      "loss": 0.5464,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.875138611665557e-06,
+      "loss": 0.5274,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.897316478154801e-06,
+      "loss": 0.5787,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.919494344644044e-06,
+      "loss": 0.5178,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.941672211133288e-06,
+      "loss": 0.4982,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.963850077622532e-06,
+      "loss": 0.5167,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.986027944111776e-06,
+      "loss": 0.5424,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.0082058106010205e-06,
+      "loss": 0.5339,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.030383677090263e-06,
+      "loss": 0.489,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.052561543579507e-06,
+      "loss": 0.4601,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.074739410068751e-06,
+      "loss": 0.5233,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.096917276557994e-06,
+      "loss": 0.5209,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.119095143047238e-06,
+      "loss": 0.4784,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.141273009536482e-06,
+      "loss": 0.5041,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.163450876025726e-06,
+      "loss": 0.5089,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.1856287425149695e-06,
+      "loss": 0.5949,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.207806609004213e-06,
+      "loss": 0.5288,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.229984475493457e-06,
+      "loss": 0.5446,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.252162341982701e-06,
+      "loss": 0.4889,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.274340208471944e-06,
+      "loss": 0.5784,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.296518074961188e-06,
+      "loss": 0.5475,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.318695941450432e-06,
+      "loss": 0.5565,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.340873807939676e-06,
+      "loss": 0.4659,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.363051674428919e-06,
+      "loss": 0.5566,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.3852295409181634e-06,
+      "loss": 0.5255,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.407407407407407e-06,
+      "loss": 0.491,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.42958527389665e-06,
+      "loss": 0.5255,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.451763140385894e-06,
+      "loss": 0.4794,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.473941006875138e-06,
+      "loss": 0.4766,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.496118873364382e-06,
+      "loss": 0.5221,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.518296739853626e-06,
+      "loss": 0.5087,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.540474606342869e-06,
+      "loss": 0.5149,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.562652472832113e-06,
+      "loss": 0.4021,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.5848303393213565e-06,
+      "loss": 0.5769,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.6070082058106e-06,
+      "loss": 0.4924,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.629186072299844e-06,
+      "loss": 0.5353,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.651363938789087e-06,
+      "loss": 0.5227,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.673541805278332e-06,
+      "loss": 0.4971,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.695719671767575e-06,
+      "loss": 0.5455,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.71789753825682e-06,
+      "loss": 0.4853,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.740075404746063e-06,
+      "loss": 0.4695,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.762253271235306e-06,
+      "loss": 0.49,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.78443113772455e-06,
+      "loss": 0.5604,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.806609004213794e-06,
+      "loss": 0.5101,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.828786870703038e-06,
+      "loss": 0.5147,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.850964737192281e-06,
+      "loss": 0.4793,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.873142603681526e-06,
+      "loss": 0.5228,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.895320470170769e-06,
+      "loss": 0.5124,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.917498336660012e-06,
+      "loss": 0.5098,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.939676203149257e-06,
+      "loss": 0.5497,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.9618540696385e-06,
+      "loss": 0.4812,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.984031936127744e-06,
+      "loss": 0.5357,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.006209802616987e-06,
+      "loss": 0.5543,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.028387669106232e-06,
+      "loss": 0.5094,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.050565535595475e-06,
+      "loss": 0.5662,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.07274340208472e-06,
+      "loss": 0.5144,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.094921268573963e-06,
+      "loss": 0.4295,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.117099135063206e-06,
+      "loss": 0.5113,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.13927700155245e-06,
+      "loss": 0.4658,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.161454868041693e-06,
+      "loss": 0.5513,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.183632734530937e-06,
+      "loss": 0.4416,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.20581060102018e-06,
+      "loss": 0.5313,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.227988467509426e-06,
+      "loss": 0.5215,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.250166333998669e-06,
+      "loss": 0.5032,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.272344200487912e-06,
+      "loss": 0.5262,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.294522066977157e-06,
+      "loss": 0.4861,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.3166999334664e-06,
+      "loss": 0.4971,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.338877799955643e-06,
+      "loss": 0.5647,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.361055666444886e-06,
+      "loss": 0.5539,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.383233532934131e-06,
+      "loss": 0.4273,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.405411399423375e-06,
+      "loss": 0.5022,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.427589265912618e-06,
+      "loss": 0.4747,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.449767132401863e-06,
+      "loss": 0.4979,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.471944998891106e-06,
+      "loss": 0.4913,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.49412286538035e-06,
+      "loss": 0.5294,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.516300731869592e-06,
+      "loss": 0.5181,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.538478598358837e-06,
+      "loss": 0.5288,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.56065646484808e-06,
+      "loss": 0.4473,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.582834331337325e-06,
+      "loss": 0.4666,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.605012197826568e-06,
+      "loss": 0.5298,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.627190064315812e-06,
+      "loss": 0.5063,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.649367930805057e-06,
+      "loss": 0.5236,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.6715457972943e-06,
+      "loss": 0.4632,
+      "step": 391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.693723663783543e-06,
+      "loss": 0.5094,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.715901530272786e-06,
+      "loss": 0.4806,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.738079396762031e-06,
+      "loss": 0.4754,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.760257263251274e-06,
+      "loss": 0.5047,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.782435129740518e-06,
+      "loss": 0.5149,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.804612996229762e-06,
+      "loss": 0.5037,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.826790862719006e-06,
+      "loss": 0.4599,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.84896872920825e-06,
+      "loss": 0.513,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.871146595697494e-06,
+      "loss": 0.505,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.893324462186737e-06,
+      "loss": 0.4368,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.91550232867598e-06,
+      "loss": 0.4828,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.937680195165223e-06,
+      "loss": 0.4948,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.959858061654468e-06,
+      "loss": 0.4626,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.982035928143711e-06,
+      "loss": 0.5123,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.004213794632956e-06,
+      "loss": 0.5066,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.0263916611222e-06,
+      "loss": 0.4849,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.048569527611443e-06,
+      "loss": 0.5121,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.070747394100688e-06,
+      "loss": 0.5033,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.092925260589931e-06,
+      "loss": 0.4852,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.115103127079174e-06,
+      "loss": 0.49,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.137280993568417e-06,
+      "loss": 0.4818,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.159458860057662e-06,
+      "loss": 0.4528,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.181636726546905e-06,
+      "loss": 0.489,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.203814593036149e-06,
+      "loss": 0.5218,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.225992459525393e-06,
+      "loss": 0.5351,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.248170326014637e-06,
+      "loss": 0.4759,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.27034819250388e-06,
+      "loss": 0.4914,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.292526058993123e-06,
+      "loss": 0.4874,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.314703925482368e-06,
+      "loss": 0.5033,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.336881791971611e-06,
+      "loss": 0.5058,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.359059658460856e-06,
+      "loss": 0.4284,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.3812375249501e-06,
+      "loss": 0.4551,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.403415391439343e-06,
+      "loss": 0.5031,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.425593257928587e-06,
+      "loss": 0.4763,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.44777112441783e-06,
+      "loss": 0.4639,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.469948990907074e-06,
+      "loss": 0.4794,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.492126857396319e-06,
+      "loss": 0.4558,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.514304723885562e-06,
+      "loss": 0.4524,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.536482590374807e-06,
+      "loss": 0.4792,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.558660456864048e-06,
+      "loss": 0.4515,
+      "step": 431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.580838323353292e-06,
+      "loss": 0.4816,
+      "step": 432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.603016189842536e-06,
+      "loss": 0.4815,
+      "step": 433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.62519405633178e-06,
+      "loss": 0.485,
+      "step": 434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.647371922821023e-06,
+      "loss": 0.4594,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.669549789310268e-06,
+      "loss": 0.4608,
+      "step": 436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.691727655799511e-06,
+      "loss": 0.4858,
+      "step": 437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.713905522288754e-06,
+      "loss": 0.4971,
+      "step": 438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.736083388777999e-06,
+      "loss": 0.4833,
+      "step": 439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758261255267242e-06,
+      "loss": 0.5102,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780439121756487e-06,
+      "loss": 0.4823,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.80261698824573e-06,
+      "loss": 0.5385,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.824794854734974e-06,
+      "loss": 0.4378,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.846972721224218e-06,
+      "loss": 0.4991,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869150587713462e-06,
+      "loss": 0.6081,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.891328454202707e-06,
+      "loss": 0.4613,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.913506320691948e-06,
+      "loss": 0.5294,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935684187181191e-06,
+      "loss": 0.4406,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957862053670434e-06,
+      "loss": 0.5072,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.98003992015968e-06,
+      "loss": 0.4784,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0002217786648923e-05,
+      "loss": 0.4736,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0024395653138167e-05,
+      "loss": 0.4316,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.004657351962741e-05,
+      "loss": 0.5066,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0068751386116654e-05,
+      "loss": 0.5111,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0090929252605899e-05,
+      "loss": 0.4739,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0113107119095142e-05,
+      "loss": 0.5182,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0135284985584387e-05,
+      "loss": 0.4594,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.015746285207363e-05,
+      "loss": 0.4351,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0179640718562873e-05,
+      "loss": 0.4826,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0201818585052118e-05,
+      "loss": 0.4659,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0223996451541361e-05,
+      "loss": 0.4524,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0246174318030605e-05,
+      "loss": 0.4398,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.026835218451985e-05,
+      "loss": 0.4414,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0290530051009091e-05,
+      "loss": 0.4399,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0312707917498334e-05,
+      "loss": 0.4737,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0334885783987579e-05,
+      "loss": 0.4813,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0357063650476822e-05,
+      "loss": 0.4913,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0379241516966067e-05,
+      "loss": 0.4592,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.040141938345531e-05,
+      "loss": 0.5093,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0423597249944554e-05,
+      "loss": 0.486,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0445775116433799e-05,
+      "loss": 0.4548,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0467952982923042e-05,
+      "loss": 0.4568,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0490130849412285e-05,
+      "loss": 0.5081,
+      "step": 473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.051230871590153e-05,
+      "loss": 0.5028,
+      "step": 474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0534486582390773e-05,
+      "loss": 0.5197,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0556664448880018e-05,
+      "loss": 0.4384,
+      "step": 476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0578842315369261e-05,
+      "loss": 0.4704,
+      "step": 477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0601020181858504e-05,
+      "loss": 0.5181,
+      "step": 478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.062319804834775e-05,
+      "loss": 0.4882,
+      "step": 479
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0645375914836992e-05,
+      "loss": 0.4615,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0667553781326234e-05,
+      "loss": 0.4893,
+      "step": 481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0689731647815479e-05,
+      "loss": 0.4803,
+      "step": 482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0711909514304722e-05,
+      "loss": 0.4333,
+      "step": 483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0734087380793965e-05,
+      "loss": 0.5059,
+      "step": 484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.075626524728321e-05,
+      "loss": 0.4658,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0778443113772453e-05,
+      "loss": 0.4807,
+      "step": 486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0800620980261698e-05,
+      "loss": 0.4712,
+      "step": 487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0822798846750942e-05,
+      "loss": 0.5223,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0844976713240185e-05,
+      "loss": 0.4904,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.086715457972943e-05,
+      "loss": 0.4164,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0889332446218673e-05,
+      "loss": 0.4574,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0911510312707918e-05,
+      "loss": 0.4742,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0933688179197161e-05,
+      "loss": 0.454,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0955866045686404e-05,
+      "loss": 0.4399,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0978043912175649e-05,
+      "loss": 0.465,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1000221778664892e-05,
+      "loss": 0.4415,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1022399645154135e-05,
+      "loss": 0.4393,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1044577511643379e-05,
+      "loss": 0.518,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1066755378132622e-05,
+      "loss": 0.4782,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1088933244621865e-05,
+      "loss": 0.4457,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.111111111111111e-05,
+      "loss": 0.4348,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1133288977600353e-05,
+      "loss": 0.4724,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1155466844089598e-05,
+      "loss": 0.4442,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1177644710578841e-05,
+      "loss": 0.5095,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1199822577068084e-05,
+      "loss": 0.5246,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.122200044355733e-05,
+      "loss": 0.5194,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1244178310046573e-05,
+      "loss": 0.4453,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1266356176535816e-05,
+      "loss": 0.5117,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.128853404302506e-05,
+      "loss": 0.4832,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1310711909514304e-05,
+      "loss": 0.4727,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1332889776003549e-05,
+      "loss": 0.4378,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1355067642492792e-05,
+      "loss": 0.4381,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1377245508982035e-05,
+      "loss": 0.5033,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.139942337547128e-05,
+      "loss": 0.4309,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1421601241960522e-05,
+      "loss": 0.4446,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1443779108449765e-05,
+      "loss": 0.4559,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.146595697493901e-05,
+      "loss": 0.4639,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1488134841428253e-05,
+      "loss": 0.4874,
+      "step": 518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1510312707917496e-05,
+      "loss": 0.4599,
+      "step": 519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1532490574406741e-05,
+      "loss": 0.475,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1554668440895984e-05,
+      "loss": 0.4641,
+      "step": 521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1576846307385229e-05,
+      "loss": 0.4506,
+      "step": 522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1599024173874472e-05,
+      "loss": 0.5239,
+      "step": 523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1621202040363716e-05,
+      "loss": 0.4947,
+      "step": 524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.164337990685296e-05,
+      "loss": 0.4583,
+      "step": 525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1665557773342204e-05,
+      "loss": 0.4701,
+      "step": 526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1687735639831449e-05,
+      "loss": 0.4927,
+      "step": 527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1709913506320692e-05,
+      "loss": 0.4697,
+      "step": 528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1732091372809935e-05,
+      "loss": 0.5041,
+      "step": 529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.175426923929918e-05,
+      "loss": 0.4655,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1776447105788423e-05,
+      "loss": 0.4818,
+      "step": 531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1798624972277665e-05,
+      "loss": 0.5016,
+      "step": 532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.182080283876691e-05,
+      "loss": 0.4247,
+      "step": 533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1842980705256153e-05,
+      "loss": 0.4317,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1865158571745396e-05,
+      "loss": 0.5049,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.188733643823464e-05,
+      "loss": 0.3929,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1909514304723884e-05,
+      "loss": 0.4507,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1931692171213129e-05,
+      "loss": 0.4426,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1953870037702372e-05,
+      "loss": 0.5181,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1976047904191615e-05,
+      "loss": 0.3991,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.199822577068086e-05,
+      "loss": 0.4324,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2020403637170103e-05,
+      "loss": 0.4559,
+      "step": 542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2042581503659347e-05,
+      "loss": 0.4659,
+      "step": 543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2064759370148591e-05,
+      "loss": 0.426,
+      "step": 544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2086937236637835e-05,
+      "loss": 0.4244,
+      "step": 545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.210911510312708e-05,
+      "loss": 0.4945,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2131292969616323e-05,
+      "loss": 0.4781,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2153470836105566e-05,
+      "loss": 0.4728,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.217564870259481e-05,
+      "loss": 0.4421,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2197826569084052e-05,
+      "loss": 0.4534,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2220004435573296e-05,
+      "loss": 0.43,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.224218230206254e-05,
+      "loss": 0.4791,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2264360168551784e-05,
+      "loss": 0.458,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2286538035041027e-05,
+      "loss": 0.4749,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2308715901530272e-05,
+      "loss": 0.421,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2330893768019515e-05,
+      "loss": 0.4324,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.235307163450876e-05,
+      "loss": 0.5113,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2375249500998003e-05,
+      "loss": 0.4913,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2397427367487246e-05,
+      "loss": 0.4863,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2419605233976491e-05,
+      "loss": 0.411,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2441783100465734e-05,
+      "loss": 0.4254,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.246396096695498e-05,
+      "loss": 0.4491,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2486138833444223e-05,
+      "loss": 0.4543,
+      "step": 563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2508316699933466e-05,
+      "loss": 0.4878,
+      "step": 564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.253049456642271e-05,
+      "loss": 0.4763,
+      "step": 565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2552672432911952e-05,
+      "loss": 0.4452,
+      "step": 566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2574850299401195e-05,
+      "loss": 0.4323,
+      "step": 567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.259702816589044e-05,
+      "loss": 0.4554,
+      "step": 568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2619206032379683e-05,
+      "loss": 0.4707,
+      "step": 569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2641383898868927e-05,
+      "loss": 0.4194,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2663561765358172e-05,
+      "loss": 0.4538,
+      "step": 571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2685739631847415e-05,
+      "loss": 0.4397,
+      "step": 572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.270791749833666e-05,
+      "loss": 0.4239,
+      "step": 573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2730095364825903e-05,
+      "loss": 0.4659,
+      "step": 574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2752273231315146e-05,
+      "loss": 0.5404,
+      "step": 575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2774451097804391e-05,
+      "loss": 0.4315,
+      "step": 576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2796628964293634e-05,
+      "loss": 0.4617,
+      "step": 577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2818806830782877e-05,
+      "loss": 0.453,
+      "step": 578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2840984697272122e-05,
+      "loss": 0.4569,
+      "step": 579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2863162563761365e-05,
+      "loss": 0.4766,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.288534043025061e-05,
+      "loss": 0.4783,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2907518296739852e-05,
+      "loss": 0.4415,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2929696163229095e-05,
+      "loss": 0.461,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.295187402971834e-05,
+      "loss": 0.416,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2974051896207583e-05,
+      "loss": 0.466,
+      "step": 585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2996229762696826e-05,
+      "loss": 0.5056,
+      "step": 586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3018407629186071e-05,
+      "loss": 0.451,
+      "step": 587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3040585495675315e-05,
+      "loss": 0.4465,
+      "step": 588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3062763362164558e-05,
+      "loss": 0.4326,
+      "step": 589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3084941228653803e-05,
+      "loss": 0.441,
+      "step": 590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3107119095143046e-05,
+      "loss": 0.451,
+      "step": 591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.312929696163229e-05,
+      "loss": 0.4113,
+      "step": 592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3151474828121534e-05,
+      "loss": 0.4652,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3173652694610777e-05,
+      "loss": 0.483,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3195830561100022e-05,
+      "loss": 0.4298,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3218008427589265e-05,
+      "loss": 0.4662,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.324018629407851e-05,
+      "loss": 0.4736,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3262364160567753e-05,
+      "loss": 0.4567,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3284542027056995e-05,
+      "loss": 0.4735,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3306719893546238e-05,
+      "loss": 0.4606,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3328897760035483e-05,
+      "loss": 0.4645,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3351075626524726e-05,
+      "loss": 0.4931,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3373253493013971e-05,
+      "loss": 0.3991,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3395431359503214e-05,
+      "loss": 0.4722,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3417609225992457e-05,
+      "loss": 0.432,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3439787092481702e-05,
+      "loss": 0.45,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3461964958970946e-05,
+      "loss": 0.4481,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.348414282546019e-05,
+      "loss": 0.4507,
+      "step": 608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3506320691949434e-05,
+      "loss": 0.4629,
+      "step": 609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3528498558438677e-05,
+      "loss": 0.5211,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3550676424927922e-05,
+      "loss": 0.459,
+      "step": 611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3572854291417165e-05,
+      "loss": 0.5222,
+      "step": 612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3595032157906408e-05,
+      "loss": 0.3938,
+      "step": 613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3617210024395653e-05,
+      "loss": 0.4786,
+      "step": 614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3639387890884896e-05,
+      "loss": 0.4883,
+      "step": 615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3661565757374138e-05,
+      "loss": 0.4312,
+      "step": 616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3683743623863383e-05,
+      "loss": 0.4504,
+      "step": 617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3705921490352626e-05,
+      "loss": 0.378,
+      "step": 618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.372809935684187e-05,
+      "loss": 0.5099,
+      "step": 619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3750277223331114e-05,
+      "loss": 0.4661,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3772455089820357e-05,
+      "loss": 0.4389,
+      "step": 621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3794632956309602e-05,
+      "loss": 0.4964,
+      "step": 622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3816810822798845e-05,
+      "loss": 0.414,
+      "step": 623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3838988689288089e-05,
+      "loss": 0.4809,
+      "step": 624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3861166555777333e-05,
+      "loss": 0.5048,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3883344422266577e-05,
+      "loss": 0.439,
+      "step": 626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3905522288755822e-05,
+      "loss": 0.4196,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3927700155245065e-05,
+      "loss": 0.4269,
+      "step": 628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3949878021734308e-05,
+      "loss": 0.4719,
+      "step": 629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3972055888223553e-05,
+      "loss": 0.4278,
+      "step": 630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3994233754712796e-05,
+      "loss": 0.4804,
+      "step": 631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4016411621202041e-05,
+      "loss": 0.4654,
+      "step": 632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4038589487691282e-05,
+      "loss": 0.4286,
+      "step": 633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4060767354180526e-05,
+      "loss": 0.4956,
+      "step": 634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4082945220669769e-05,
+      "loss": 0.4778,
+      "step": 635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4105123087159014e-05,
+      "loss": 0.4402,
+      "step": 636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4127300953648257e-05,
+      "loss": 0.3898,
+      "step": 637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4149478820137502e-05,
+      "loss": 0.4426,
+      "step": 638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4171656686626745e-05,
+      "loss": 0.4953,
+      "step": 639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4193834553115988e-05,
+      "loss": 0.4492,
+      "step": 640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4216012419605233e-05,
+      "loss": 0.449,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4238190286094476e-05,
+      "loss": 0.4479,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4260368152583721e-05,
+      "loss": 0.4801,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4282546019072964e-05,
+      "loss": 0.4505,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4304723885562208e-05,
+      "loss": 0.4903,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4326901752051453e-05,
+      "loss": 0.4761,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4349079618540696e-05,
+      "loss": 0.4442,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4371257485029939e-05,
+      "loss": 0.4555,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4393435351519184e-05,
+      "loss": 0.4458,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4415613218008425e-05,
+      "loss": 0.417,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4437791084497669e-05,
+      "loss": 0.4468,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4459968950986914e-05,
+      "loss": 0.4537,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4482146817476157e-05,
+      "loss": 0.4453,
+      "step": 653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4504324683965402e-05,
+      "loss": 0.4569,
+      "step": 654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4526502550454645e-05,
+      "loss": 0.4398,
+      "step": 655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4548680416943888e-05,
+      "loss": 0.4445,
+      "step": 656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4570858283433133e-05,
+      "loss": 0.4228,
+      "step": 657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4593036149922376e-05,
+      "loss": 0.4657,
+      "step": 658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.461521401641162e-05,
+      "loss": 0.4258,
+      "step": 659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4637391882900864e-05,
+      "loss": 0.4581,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4659569749390107e-05,
+      "loss": 0.4366,
+      "step": 661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4681747615879352e-05,
+      "loss": 0.4509,
+      "step": 662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4703925482368596e-05,
+      "loss": 0.4358,
+      "step": 663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4726103348857839e-05,
+      "loss": 0.4371,
+      "step": 664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4748281215347084e-05,
+      "loss": 0.472,
+      "step": 665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4770459081836327e-05,
+      "loss": 0.4551,
+      "step": 666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4792636948325568e-05,
+      "loss": 0.4181,
+      "step": 667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4814814814814813e-05,
+      "loss": 0.4645,
+      "step": 668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4836992681304056e-05,
+      "loss": 0.4201,
+      "step": 669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.48591705477933e-05,
+      "loss": 0.4935,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4881348414282545e-05,
+      "loss": 0.4109,
+      "step": 671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4903526280771788e-05,
+      "loss": 0.4841,
+      "step": 672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4925704147261033e-05,
+      "loss": 0.4515,
+      "step": 673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4947882013750276e-05,
+      "loss": 0.4903,
+      "step": 674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4970059880239519e-05,
+      "loss": 0.4994,
+      "step": 675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4992237746728764e-05,
+      "loss": 0.4577,
+      "step": 676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5014415613218007e-05,
+      "loss": 0.4667,
+      "step": 677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5036593479707252e-05,
+      "loss": 0.4598,
+      "step": 678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5058771346196495e-05,
+      "loss": 0.4386,
+      "step": 679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5080949212685739e-05,
+      "loss": 0.4256,
+      "step": 680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5103127079174983e-05,
+      "loss": 0.4317,
+      "step": 681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5125304945664227e-05,
+      "loss": 0.4281,
+      "step": 682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.514748281215347e-05,
+      "loss": 0.3935,
+      "step": 683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5169660678642713e-05,
+      "loss": 0.4106,
+      "step": 684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5191838545131956e-05,
+      "loss": 0.4401,
+      "step": 685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.52140164116212e-05,
+      "loss": 0.4625,
+      "step": 686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5236194278110444e-05,
+      "loss": 0.4046,
+      "step": 687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5258372144599688e-05,
+      "loss": 0.4095,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5280550011088932e-05,
+      "loss": 0.437,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5302727877578174e-05,
+      "loss": 0.4487,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.532490574406742e-05,
+      "loss": 0.3962,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5347083610556664e-05,
+      "loss": 0.4003,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.536926147704591e-05,
+      "loss": 0.4392,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.539143934353515e-05,
+      "loss": 0.4925,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5413617210024395e-05,
+      "loss": 0.4384,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.543579507651364e-05,
+      "loss": 0.3979,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.545797294300288e-05,
+      "loss": 0.3915,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5480150809492126e-05,
+      "loss": 0.4216,
+      "step": 698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.550232867598137e-05,
+      "loss": 0.4463,
+      "step": 699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5524506542470613e-05,
+      "loss": 0.4298,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5546684408959854e-05,
+      "loss": 0.4879,
+      "step": 701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.55688622754491e-05,
+      "loss": 0.4608,
+      "step": 702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5591040141938344e-05,
+      "loss": 0.4598,
+      "step": 703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.561321800842759e-05,
+      "loss": 0.4653,
+      "step": 704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.563539587491683e-05,
+      "loss": 0.4106,
+      "step": 705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5657573741406075e-05,
+      "loss": 0.4275,
+      "step": 706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.567975160789532e-05,
+      "loss": 0.4485,
+      "step": 707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5701929474384562e-05,
+      "loss": 0.4599,
+      "step": 708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5724107340873807e-05,
+      "loss": 0.4447,
+      "step": 709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.574628520736305e-05,
+      "loss": 0.4651,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5768463073852293e-05,
+      "loss": 0.4348,
+      "step": 711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5790640940341538e-05,
+      "loss": 0.4015,
+      "step": 712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5812818806830783e-05,
+      "loss": 0.4289,
+      "step": 713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5834996673320024e-05,
+      "loss": 0.4228,
+      "step": 714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.585717453980927e-05,
+      "loss": 0.4359,
+      "step": 715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5879352406298514e-05,
+      "loss": 0.4792,
+      "step": 716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5901530272787756e-05,
+      "loss": 0.4448,
+      "step": 717
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5923708139277e-05,
+      "loss": 0.4365,
+      "step": 718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5945886005766242e-05,
+      "loss": 0.4524,
+      "step": 719
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5968063872255487e-05,
+      "loss": 0.3983,
+      "step": 720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5990241738744732e-05,
+      "loss": 0.4251,
+      "step": 721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6012419605233973e-05,
+      "loss": 0.4072,
+      "step": 722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.603459747172322e-05,
+      "loss": 0.4632,
+      "step": 723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6056775338212463e-05,
+      "loss": 0.4476,
+      "step": 724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6078953204701705e-05,
+      "loss": 0.4334,
+      "step": 725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.610113107119095e-05,
+      "loss": 0.4518,
+      "step": 726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6123308937680195e-05,
+      "loss": 0.4372,
+      "step": 727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.614548680416944e-05,
+      "loss": 0.4191,
+      "step": 728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.616766467065868e-05,
+      "loss": 0.4718,
+      "step": 729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6189842537147926e-05,
+      "loss": 0.4798,
+      "step": 730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.621202040363717e-05,
+      "loss": 0.524,
+      "step": 731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6234198270126412e-05,
+      "loss": 0.4581,
+      "step": 732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6256376136615657e-05,
+      "loss": 0.4299,
+      "step": 733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.62785540031049e-05,
+      "loss": 0.4086,
+      "step": 734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6300731869594144e-05,
+      "loss": 0.4065,
+      "step": 735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6322909736083385e-05,
+      "loss": 0.4083,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.634508760257263e-05,
+      "loss": 0.463,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6367265469061875e-05,
+      "loss": 0.4282,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.638944333555112e-05,
+      "loss": 0.4702,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.641162120204036e-05,
+      "loss": 0.4364,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6433799068529606e-05,
+      "loss": 0.3807,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.645597693501885e-05,
+      "loss": 0.422,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6478154801508093e-05,
+      "loss": 0.4158,
+      "step": 743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6500332667997337e-05,
+      "loss": 0.4588,
+      "step": 744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6522510534486582e-05,
+      "loss": 0.4484,
+      "step": 745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6544688400975824e-05,
+      "loss": 0.4315,
+      "step": 746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.656686626746507e-05,
+      "loss": 0.4086,
+      "step": 747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6589044133954314e-05,
+      "loss": 0.4361,
+      "step": 748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6611222000443555e-05,
+      "loss": 0.469,
+      "step": 749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.66333998669328e-05,
+      "loss": 0.4309,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.665557773342204e-05,
+      "loss": 0.4217,
+      "step": 751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6677755599911287e-05,
+      "loss": 0.4495,
+      "step": 752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.669993346640053e-05,
+      "loss": 0.4405,
+      "step": 753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6722111332889773e-05,
+      "loss": 0.4729,
+      "step": 754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6744289199379018e-05,
+      "loss": 0.422,
+      "step": 755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6766467065868263e-05,
+      "loss": 0.4317,
+      "step": 756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6788644932357504e-05,
+      "loss": 0.4727,
+      "step": 757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.681082279884675e-05,
+      "loss": 0.4338,
+      "step": 758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6833000665335994e-05,
+      "loss": 0.4781,
+      "step": 759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6855178531825236e-05,
+      "loss": 0.494,
+      "step": 760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.687735639831448e-05,
+      "loss": 0.411,
+      "step": 761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6899534264803725e-05,
+      "loss": 0.4725,
+      "step": 762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.692171213129297e-05,
+      "loss": 0.5179,
+      "step": 763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6943889997782212e-05,
+      "loss": 0.4271,
+      "step": 764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6966067864271457e-05,
+      "loss": 0.3896,
+      "step": 765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.69882457307607e-05,
+      "loss": 0.4828,
+      "step": 766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7010423597249943e-05,
+      "loss": 0.4111,
+      "step": 767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7032601463739185e-05,
+      "loss": 0.4506,
+      "step": 768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.705477933022843e-05,
+      "loss": 0.4659,
+      "step": 769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7076957196717674e-05,
+      "loss": 0.4303,
+      "step": 770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7099135063206916e-05,
+      "loss": 0.4842,
+      "step": 771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.712131292969616e-05,
+      "loss": 0.4079,
+      "step": 772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7143490796185406e-05,
+      "loss": 0.4256,
+      "step": 773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.716566866267465e-05,
+      "loss": 0.4516,
+      "step": 774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7187846529163892e-05,
+      "loss": 0.3898,
+      "step": 775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7210024395653137e-05,
+      "loss": 0.4766,
+      "step": 776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7232202262142382e-05,
+      "loss": 0.4822,
+      "step": 777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7254380128631623e-05,
+      "loss": 0.4652,
+      "step": 778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.727655799512087e-05,
+      "loss": 0.471,
+      "step": 779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7298735861610113e-05,
+      "loss": 0.4128,
+      "step": 780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7320913728099355e-05,
+      "loss": 0.4279,
+      "step": 781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.73430915945886e-05,
+      "loss": 0.4594,
+      "step": 782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7365269461077845e-05,
+      "loss": 0.43,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7387447327567086e-05,
+      "loss": 0.4506,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.740962519405633e-05,
+      "loss": 0.4468,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7431803060545572e-05,
+      "loss": 0.4679,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7453980927034817e-05,
+      "loss": 0.432,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7476158793524062e-05,
+      "loss": 0.4524,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7498336660013304e-05,
+      "loss": 0.4229,
+      "step": 789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.752051452650255e-05,
+      "loss": 0.431,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7542692392991794e-05,
+      "loss": 0.455,
+      "step": 791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7564870259481035e-05,
+      "loss": 0.44,
+      "step": 792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.758704812597028e-05,
+      "loss": 0.4411,
+      "step": 793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7609225992459525e-05,
+      "loss": 0.4379,
+      "step": 794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7631403858948766e-05,
+      "loss": 0.4489,
+      "step": 795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.765358172543801e-05,
+      "loss": 0.4573,
+      "step": 796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7675759591927256e-05,
+      "loss": 0.4849,
+      "step": 797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.76979374584165e-05,
+      "loss": 0.3936,
+      "step": 798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7720115324905743e-05,
+      "loss": 0.4317,
+      "step": 799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7742293191394987e-05,
+      "loss": 0.4125,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7764471057884232e-05,
+      "loss": 0.4456,
+      "step": 801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7786648924373474e-05,
+      "loss": 0.4836,
+      "step": 802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7808826790862715e-05,
+      "loss": 0.4553,
+      "step": 803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.783100465735196e-05,
+      "loss": 0.4133,
+      "step": 804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7853182523841205e-05,
+      "loss": 0.4515,
+      "step": 805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7875360390330447e-05,
+      "loss": 0.4809,
+      "step": 806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.789753825681969e-05,
+      "loss": 0.4593,
+      "step": 807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7919716123308936e-05,
+      "loss": 0.3937,
+      "step": 808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.794189398979818e-05,
+      "loss": 0.4261,
+      "step": 809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7964071856287423e-05,
+      "loss": 0.4134,
+      "step": 810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7986249722776668e-05,
+      "loss": 0.4627,
+      "step": 811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8008427589265913e-05,
+      "loss": 0.4026,
+      "step": 812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8030605455755154e-05,
+      "loss": 0.4235,
+      "step": 813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.80527833222444e-05,
+      "loss": 0.4327,
+      "step": 814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8074961188733644e-05,
+      "loss": 0.4373,
+      "step": 815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8097139055222886e-05,
+      "loss": 0.3969,
+      "step": 816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.811931692171213e-05,
+      "loss": 0.4447,
+      "step": 817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8141494788201375e-05,
+      "loss": 0.4466,
+      "step": 818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8163672654690617e-05,
+      "loss": 0.435,
+      "step": 819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8185850521179862e-05,
+      "loss": 0.4219,
+      "step": 820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8208028387669103e-05,
+      "loss": 0.4486,
+      "step": 821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8230206254158348e-05,
+      "loss": 0.4115,
+      "step": 822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8252384120647593e-05,
+      "loss": 0.442,
+      "step": 823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8274561987136835e-05,
+      "loss": 0.4195,
+      "step": 824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.829673985362608e-05,
+      "loss": 0.3851,
+      "step": 825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8318917720115324e-05,
+      "loss": 0.4582,
+      "step": 826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8341095586604566e-05,
+      "loss": 0.4596,
+      "step": 827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.836327345309381e-05,
+      "loss": 0.4646,
+      "step": 828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8385451319583056e-05,
+      "loss": 0.4143,
+      "step": 829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8407629186072297e-05,
+      "loss": 0.4365,
+      "step": 830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8429807052561542e-05,
+      "loss": 0.4208,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8451984919050787e-05,
+      "loss": 0.4473,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8474162785540032e-05,
+      "loss": 0.4749,
+      "step": 833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8496340652029273e-05,
+      "loss": 0.4687,
+      "step": 834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8518518518518515e-05,
+      "loss": 0.4469,
+      "step": 835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.854069638500776e-05,
+      "loss": 0.4878,
+      "step": 836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8562874251497005e-05,
+      "loss": 0.3666,
+      "step": 837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8585052117986246e-05,
+      "loss": 0.4719,
+      "step": 838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860722998447549e-05,
+      "loss": 0.4489,
+      "step": 839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629407850964736e-05,
+      "loss": 0.4494,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651585717453978e-05,
+      "loss": 0.4397,
+      "step": 841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673763583943222e-05,
+      "loss": 0.3865,
+      "step": 842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695941450432467e-05,
+      "loss": 0.4634,
+      "step": 843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718119316921712e-05,
+      "loss": 0.4133,
+      "step": 844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740297183410954e-05,
+      "loss": 0.3615,
+      "step": 845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87624750499002e-05,
+      "loss": 0.4129,
+      "step": 846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8784652916389444e-05,
+      "loss": 0.4255,
+      "step": 847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8806830782878685e-05,
+      "loss": 0.4315,
+      "step": 848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.882900864936793e-05,
+      "loss": 0.3932,
+      "step": 849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8851186515857175e-05,
+      "loss": 0.431,
+      "step": 850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8873364382346416e-05,
+      "loss": 0.4423,
+      "step": 851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.889554224883566e-05,
+      "loss": 0.4723,
+      "step": 852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8917720115324906e-05,
+      "loss": 0.4266,
+      "step": 853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8939897981814148e-05,
+      "loss": 0.4684,
+      "step": 854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8962075848303393e-05,
+      "loss": 0.4279,
+      "step": 855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8984253714792637e-05,
+      "loss": 0.407,
+      "step": 856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9006431581281882e-05,
+      "loss": 0.3883,
+      "step": 857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9028609447771124e-05,
+      "loss": 0.4516,
+      "step": 858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.905078731426037e-05,
+      "loss": 0.4704,
+      "step": 859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9072965180749614e-05,
+      "loss": 0.4629,
+      "step": 860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9095143047238852e-05,
+      "loss": 0.4257,
+      "step": 861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9117320913728097e-05,
+      "loss": 0.441,
+      "step": 862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9139498780217338e-05,
+      "loss": 0.4787,
+      "step": 863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9161676646706583e-05,
+      "loss": 0.4539,
+      "step": 864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9183854513195828e-05,
+      "loss": 0.4402,
+      "step": 865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9206032379685073e-05,
+      "loss": 0.4252,
+      "step": 866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9228210246174314e-05,
+      "loss": 0.4245,
+      "step": 867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.925038811266356e-05,
+      "loss": 0.4217,
+      "step": 868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9272565979152804e-05,
+      "loss": 0.4373,
+      "step": 869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9294743845642046e-05,
+      "loss": 0.4393,
+      "step": 870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.931692171213129e-05,
+      "loss": 0.4136,
+      "step": 871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9339099578620535e-05,
+      "loss": 0.4171,
+      "step": 872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9361277445109777e-05,
+      "loss": 0.4318,
+      "step": 873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9383455311599022e-05,
+      "loss": 0.4611,
+      "step": 874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9405633178088267e-05,
+      "loss": 0.4191,
+      "step": 875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.942781104457751e-05,
+      "loss": 0.4147,
+      "step": 876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9449988911066753e-05,
+      "loss": 0.4015,
+      "step": 877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9472166777555998e-05,
+      "loss": 0.4035,
+      "step": 878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9494344644045243e-05,
+      "loss": 0.3993,
+      "step": 879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9516522510534485e-05,
+      "loss": 0.4198,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.953870037702373e-05,
+      "loss": 0.4172,
+      "step": 881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9560878243512974e-05,
+      "loss": 0.3826,
+      "step": 882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9583056110002216e-05,
+      "loss": 0.4433,
+      "step": 883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.960523397649146e-05,
+      "loss": 0.4684,
+      "step": 884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9627411842980706e-05,
+      "loss": 0.4117,
+      "step": 885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9649589709469947e-05,
+      "loss": 0.4342,
+      "step": 886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9671767575959192e-05,
+      "loss": 0.3756,
+      "step": 887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9693945442448437e-05,
+      "loss": 0.3728,
+      "step": 888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.971612330893768e-05,
+      "loss": 0.4317,
+      "step": 889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9738301175426923e-05,
+      "loss": 0.402,
+      "step": 890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9760479041916168e-05,
+      "loss": 0.45,
+      "step": 891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9782656908405413e-05,
+      "loss": 0.4745,
+      "step": 892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9804834774894655e-05,
+      "loss": 0.4186,
+      "step": 893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9827012641383896e-05,
+      "loss": 0.4129,
+      "step": 894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9849190507873138e-05,
+      "loss": 0.4245,
+      "step": 895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9871368374362383e-05,
+      "loss": 0.4184,
+      "step": 896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9893546240851627e-05,
+      "loss": 0.4647,
+      "step": 897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.991572410734087e-05,
+      "loss": 0.4556,
+      "step": 898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9937901973830114e-05,
+      "loss": 0.4256,
+      "step": 899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.996007984031936e-05,
+      "loss": 0.4104,
+      "step": 900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9982257706808604e-05,
+      "loss": 0.4697,
+      "step": 901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0004435573297845e-05,
+      "loss": 0.4044,
+      "step": 902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.002661343978709e-05,
+      "loss": 0.4541,
+      "step": 903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0048791306276335e-05,
+      "loss": 0.4916,
+      "step": 904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0070969172765577e-05,
+      "loss": 0.4209,
+      "step": 905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.009314703925482e-05,
+      "loss": 0.3875,
+      "step": 906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0115324905744066e-05,
+      "loss": 0.4426,
+      "step": 907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0137502772233308e-05,
+      "loss": 0.4404,
+      "step": 908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0159680638722553e-05,
+      "loss": 0.4126,
+      "step": 909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0181858505211798e-05,
+      "loss": 0.4081,
+      "step": 910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.020403637170104e-05,
+      "loss": 0.4643,
+      "step": 911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0226214238190284e-05,
+      "loss": 0.4073,
+      "step": 912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.024839210467953e-05,
+      "loss": 0.4001,
+      "step": 913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0270569971168774e-05,
+      "loss": 0.3984,
+      "step": 914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0292747837658015e-05,
+      "loss": 0.4122,
+      "step": 915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.031492570414726e-05,
+      "loss": 0.4228,
+      "step": 916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0337103570636505e-05,
+      "loss": 0.4326,
+      "step": 917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0359281437125747e-05,
+      "loss": 0.415,
+      "step": 918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.038145930361499e-05,
+      "loss": 0.4292,
+      "step": 919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0403637170104236e-05,
+      "loss": 0.3872,
+      "step": 920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0425815036593478e-05,
+      "loss": 0.3755,
+      "step": 921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0447992903082723e-05,
+      "loss": 0.434,
+      "step": 922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0470170769571968e-05,
+      "loss": 0.3831,
+      "step": 923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049234863606121e-05,
+      "loss": 0.4461,
+      "step": 924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0514526502550454e-05,
+      "loss": 0.3982,
+      "step": 925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.05367043690397e-05,
+      "loss": 0.4791,
+      "step": 926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0558882235528944e-05,
+      "loss": 0.4439,
+      "step": 927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0581060102018182e-05,
+      "loss": 0.4312,
+      "step": 928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0603237968507427e-05,
+      "loss": 0.4018,
+      "step": 929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.062541583499667e-05,
+      "loss": 0.3895,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0647593701485913e-05,
+      "loss": 0.4323,
+      "step": 931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0669771567975158e-05,
+      "loss": 0.3949,
+      "step": 932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.06919494344644e-05,
+      "loss": 0.3965,
+      "step": 933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0714127300953645e-05,
+      "loss": 0.3972,
+      "step": 934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.073630516744289e-05,
+      "loss": 0.4307,
+      "step": 935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0758483033932134e-05,
+      "loss": 0.4566,
+      "step": 936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0780660900421376e-05,
+      "loss": 0.4387,
+      "step": 937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.080283876691062e-05,
+      "loss": 0.4068,
+      "step": 938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0825016633399866e-05,
+      "loss": 0.4361,
+      "step": 939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0847194499889107e-05,
+      "loss": 0.4437,
+      "step": 940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0869372366378352e-05,
+      "loss": 0.4674,
+      "step": 941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0891550232867597e-05,
+      "loss": 0.4412,
+      "step": 942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.091372809935684e-05,
+      "loss": 0.4099,
+      "step": 943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0935905965846084e-05,
+      "loss": 0.44,
+      "step": 944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.095808383233533e-05,
+      "loss": 0.3845,
+      "step": 945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.098026169882457e-05,
+      "loss": 0.396,
+      "step": 946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1002439565313815e-05,
+      "loss": 0.4443,
+      "step": 947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.102461743180306e-05,
+      "loss": 0.4489,
+      "step": 948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1046795298292305e-05,
+      "loss": 0.4462,
+      "step": 949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1068973164781546e-05,
+      "loss": 0.4408,
+      "step": 950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.109115103127079e-05,
+      "loss": 0.4144,
+      "step": 951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1113328897760036e-05,
+      "loss": 0.4572,
+      "step": 952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1135506764249277e-05,
+      "loss": 0.3811,
+      "step": 953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1157684630738522e-05,
+      "loss": 0.4192,
+      "step": 954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1179862497227767e-05,
+      "loss": 0.3762,
+      "step": 955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.120204036371701e-05,
+      "loss": 0.4066,
+      "step": 956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1224218230206254e-05,
+      "loss": 0.4439,
+      "step": 957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.12463960966955e-05,
+      "loss": 0.4076,
+      "step": 958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.126857396318474e-05,
+      "loss": 0.4054,
+      "step": 959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1290751829673985e-05,
+      "loss": 0.3984,
+      "step": 960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.131292969616323e-05,
+      "loss": 0.4403,
+      "step": 961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1335107562652468e-05,
+      "loss": 0.416,
+      "step": 962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1357285429141713e-05,
+      "loss": 0.4354,
+      "step": 963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1379463295630958e-05,
+      "loss": 0.4267,
+      "step": 964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.14016411621202e-05,
+      "loss": 0.3818,
+      "step": 965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1423819028609444e-05,
+      "loss": 0.4344,
+      "step": 966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.144599689509869e-05,
+      "loss": 0.4272,
+      "step": 967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.146817476158793e-05,
+      "loss": 0.428,
+      "step": 968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1490352628077176e-05,
+      "loss": 0.441,
+      "step": 969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.151253049456642e-05,
+      "loss": 0.436,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1534708361055665e-05,
+      "loss": 0.4264,
+      "step": 971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1556886227544907e-05,
+      "loss": 0.4293,
+      "step": 972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1579064094034152e-05,
+      "loss": 0.4302,
+      "step": 973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1601241960523397e-05,
+      "loss": 0.3866,
+      "step": 974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1623419827012638e-05,
+      "loss": 0.4559,
+      "step": 975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1645597693501883e-05,
+      "loss": 0.4201,
+      "step": 976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1667775559991128e-05,
+      "loss": 0.3988,
+      "step": 977
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.168995342648037e-05,
+      "loss": 0.3348,
+      "step": 978
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1712131292969614e-05,
+      "loss": 0.41,
+      "step": 979
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.173430915945886e-05,
+      "loss": 0.4524,
+      "step": 980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.17564870259481e-05,
+      "loss": 0.4466,
+      "step": 981
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1778664892437346e-05,
+      "loss": 0.4004,
+      "step": 982
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.180084275892659e-05,
+      "loss": 0.3761,
+      "step": 983
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1823020625415835e-05,
+      "loss": 0.4517,
+      "step": 984
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1845198491905077e-05,
+      "loss": 0.4541,
+      "step": 985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1867376358394322e-05,
+      "loss": 0.4271,
+      "step": 986
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1889554224883567e-05,
+      "loss": 0.4685,
+      "step": 987
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1911732091372808e-05,
+      "loss": 0.4501,
+      "step": 988
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1933909957862053e-05,
+      "loss": 0.3677,
+      "step": 989
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1956087824351298e-05,
+      "loss": 0.4961,
+      "step": 990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.197826569084054e-05,
+      "loss": 0.4045,
+      "step": 991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2000443557329784e-05,
+      "loss": 0.4181,
+      "step": 992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.202262142381903e-05,
+      "loss": 0.426,
+      "step": 993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.204479929030827e-05,
+      "loss": 0.4573,
+      "step": 994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2066977156797516e-05,
+      "loss": 0.46,
+      "step": 995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2089155023286757e-05,
+      "loss": 0.4472,
+      "step": 996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2111332889776e-05,
+      "loss": 0.4372,
+      "step": 997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2133510756265244e-05,
+      "loss": 0.3789,
+      "step": 998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.215568862275449e-05,
+      "loss": 0.4137,
+      "step": 999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.217786648924373e-05,
+      "loss": 0.4339,
+      "step": 1000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2200044355732975e-05,
+      "loss": 0.451,
+      "step": 1001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.222222222222222e-05,
+      "loss": 0.4704,
+      "step": 1002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.224440008871146e-05,
+      "loss": 0.369,
+      "step": 1003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2266577955200706e-05,
+      "loss": 0.4449,
+      "step": 1004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.228875582168995e-05,
+      "loss": 0.4209,
+      "step": 1005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2310933688179196e-05,
+      "loss": 0.438,
+      "step": 1006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2333111554668438e-05,
+      "loss": 0.3908,
+      "step": 1007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2355289421157683e-05,
+      "loss": 0.4306,
+      "step": 1008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2377467287646927e-05,
+      "loss": 0.4288,
+      "step": 1009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.239964515413617e-05,
+      "loss": 0.3716,
+      "step": 1010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2421823020625414e-05,
+      "loss": 0.433,
+      "step": 1011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.244400088711466e-05,
+      "loss": 0.4448,
+      "step": 1012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.24661787536039e-05,
+      "loss": 0.4625,
+      "step": 1013
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2488356620093145e-05,
+      "loss": 0.4469,
+      "step": 1014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.251053448658239e-05,
+      "loss": 0.3896,
+      "step": 1015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.253271235307163e-05,
+      "loss": 0.4373,
+      "step": 1016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2554890219560876e-05,
+      "loss": 0.4602,
+      "step": 1017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.257706808605012e-05,
+      "loss": 0.4267,
+      "step": 1018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2599245952539366e-05,
+      "loss": 0.3907,
+      "step": 1019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2621423819028608e-05,
+      "loss": 0.4274,
+      "step": 1020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2643601685517853e-05,
+      "loss": 0.3625,
+      "step": 1021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2665779552007098e-05,
+      "loss": 0.4186,
+      "step": 1022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.268795741849634e-05,
+      "loss": 0.4334,
+      "step": 1023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2710135284985584e-05,
+      "loss": 0.3744,
+      "step": 1024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.273231315147483e-05,
+      "loss": 0.3859,
+      "step": 1025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.275449101796407e-05,
+      "loss": 0.4563,
+      "step": 1026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2776668884453315e-05,
+      "loss": 0.4445,
+      "step": 1027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.279884675094256e-05,
+      "loss": 0.3902,
+      "step": 1028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2821024617431798e-05,
+      "loss": 0.3732,
+      "step": 1029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2843202483921043e-05,
+      "loss": 0.4189,
+      "step": 1030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2865380350410288e-05,
+      "loss": 0.4377,
+      "step": 1031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.288755821689953e-05,
+      "loss": 0.4206,
+      "step": 1032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2909736083388774e-05,
+      "loss": 0.4272,
+      "step": 1033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.293191394987802e-05,
+      "loss": 0.4609,
+      "step": 1034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.295409181636726e-05,
+      "loss": 0.4102,
+      "step": 1035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2976269682856506e-05,
+      "loss": 0.3726,
+      "step": 1036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.299844754934575e-05,
+      "loss": 0.4252,
+      "step": 1037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3020625415834992e-05,
+      "loss": 0.4476,
+      "step": 1038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3042803282324237e-05,
+      "loss": 0.4067,
+      "step": 1039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3064981148813482e-05,
+      "loss": 0.4762,
+      "step": 1040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3087159015302727e-05,
+      "loss": 0.4588,
+      "step": 1041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.310933688179197e-05,
+      "loss": 0.4022,
+      "step": 1042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3131514748281213e-05,
+      "loss": 0.4013,
+      "step": 1043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3153692614770458e-05,
+      "loss": 0.4338,
+      "step": 1044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.31758704812597e-05,
+      "loss": 0.4188,
+      "step": 1045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3198048347748945e-05,
+      "loss": 0.4207,
+      "step": 1046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.322022621423819e-05,
+      "loss": 0.4023,
+      "step": 1047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.324240408072743e-05,
+      "loss": 0.4433,
+      "step": 1048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3264581947216676e-05,
+      "loss": 0.4716,
+      "step": 1049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.328675981370592e-05,
+      "loss": 0.383,
+      "step": 1050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3308937680195162e-05,
+      "loss": 0.4482,
+      "step": 1051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3331115546684407e-05,
+      "loss": 0.4436,
+      "step": 1052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3353293413173652e-05,
+      "loss": 0.3965,
+      "step": 1053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3375471279662897e-05,
+      "loss": 0.3831,
+      "step": 1054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.339764914615214e-05,
+      "loss": 0.4,
+      "step": 1055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3419827012641383e-05,
+      "loss": 0.4021,
+      "step": 1056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.344200487913063e-05,
+      "loss": 0.3849,
+      "step": 1057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.346418274561987e-05,
+      "loss": 0.4579,
+      "step": 1058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3486360612109115e-05,
+      "loss": 0.439,
+      "step": 1059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.350853847859836e-05,
+      "loss": 0.3973,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.35307163450876e-05,
+      "loss": 0.4195,
+      "step": 1061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3552894211576846e-05,
+      "loss": 0.4397,
+      "step": 1062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3575072078066088e-05,
+      "loss": 0.3843,
+      "step": 1063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.359724994455533e-05,
+      "loss": 0.4138,
+      "step": 1064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3619427811044574e-05,
+      "loss": 0.4054,
+      "step": 1065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.364160567753382e-05,
+      "loss": 0.438,
+      "step": 1066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.366378354402306e-05,
+      "loss": 0.4777,
+      "step": 1067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3685961410512305e-05,
+      "loss": 0.4119,
+      "step": 1068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.370813927700155e-05,
+      "loss": 0.4231,
+      "step": 1069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3730317143490792e-05,
+      "loss": 0.4193,
+      "step": 1070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3752495009980037e-05,
+      "loss": 0.4296,
+      "step": 1071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.377467287646928e-05,
+      "loss": 0.4197,
+      "step": 1072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3796850742958523e-05,
+      "loss": 0.3309,
+      "step": 1073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3819028609447768e-05,
+      "loss": 0.4165,
+      "step": 1074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3841206475937013e-05,
+      "loss": 0.419,
+      "step": 1075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3863384342426258e-05,
+      "loss": 0.4838,
+      "step": 1076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.38855622089155e-05,
+      "loss": 0.4479,
+      "step": 1077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3907740075404744e-05,
+      "loss": 0.4237,
+      "step": 1078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.392991794189399e-05,
+      "loss": 0.4227,
+      "step": 1079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.395209580838323e-05,
+      "loss": 0.428,
+      "step": 1080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3974273674872475e-05,
+      "loss": 0.4246,
+      "step": 1081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.399645154136172e-05,
+      "loss": 0.4109,
+      "step": 1082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4018629407850962e-05,
+      "loss": 0.4381,
+      "step": 1083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4040807274340207e-05,
+      "loss": 0.381,
+      "step": 1084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.406298514082945e-05,
+      "loss": 0.4344,
+      "step": 1085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4085163007318693e-05,
+      "loss": 0.4091,
+      "step": 1086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4107340873807938e-05,
+      "loss": 0.4476,
+      "step": 1087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4129518740297183e-05,
+      "loss": 0.4578,
+      "step": 1088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4151696606786428e-05,
+      "loss": 0.4174,
+      "step": 1089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.417387447327567e-05,
+      "loss": 0.4311,
+      "step": 1090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4196052339764914e-05,
+      "loss": 0.3986,
+      "step": 1091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.421823020625416e-05,
+      "loss": 0.4552,
+      "step": 1092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.42404080727434e-05,
+      "loss": 0.445,
+      "step": 1093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4262585939232646e-05,
+      "loss": 0.4078,
+      "step": 1094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.428476380572189e-05,
+      "loss": 0.4011,
+      "step": 1095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4306941672211132e-05,
+      "loss": 0.3947,
+      "step": 1096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4329119538700373e-05,
+      "loss": 0.4194,
+      "step": 1097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.435129740518962e-05,
+      "loss": 0.4515,
+      "step": 1098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.437347527167886e-05,
+      "loss": 0.4091,
+      "step": 1099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4395653138168105e-05,
+      "loss": 0.442,
+      "step": 1100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.441783100465735e-05,
+      "loss": 0.4133,
+      "step": 1101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.444000887114659e-05,
+      "loss": 0.4451,
+      "step": 1102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4462186737635836e-05,
+      "loss": 0.4327,
+      "step": 1103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.448436460412508e-05,
+      "loss": 0.4319,
+      "step": 1104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4506542470614323e-05,
+      "loss": 0.4005,
+      "step": 1105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4528720337103567e-05,
+      "loss": 0.3787,
+      "step": 1106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4550898203592812e-05,
+      "loss": 0.4164,
+      "step": 1107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4573076070082054e-05,
+      "loss": 0.3999,
+      "step": 1108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.45952539365713e-05,
+      "loss": 0.3959,
+      "step": 1109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4617431803060544e-05,
+      "loss": 0.4056,
+      "step": 1110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.463960966954979e-05,
+      "loss": 0.4672,
+      "step": 1111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.466178753603903e-05,
+      "loss": 0.4244,
+      "step": 1112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4683965402528275e-05,
+      "loss": 0.4129,
+      "step": 1113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.470614326901752e-05,
+      "loss": 0.4149,
+      "step": 1114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.472832113550676e-05,
+      "loss": 0.3969,
+      "step": 1115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4750499001996006e-05,
+      "loss": 0.4003,
+      "step": 1116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.477267686848525e-05,
+      "loss": 0.3814,
+      "step": 1117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4794854734974493e-05,
+      "loss": 0.439,
+      "step": 1118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4817032601463738e-05,
+      "loss": 0.4174,
+      "step": 1119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4839210467952982e-05,
+      "loss": 0.4286,
+      "step": 1120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4861388334442224e-05,
+      "loss": 0.43,
+      "step": 1121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.488356620093147e-05,
+      "loss": 0.4494,
+      "step": 1122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4905744067420714e-05,
+      "loss": 0.4451,
+      "step": 1123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.492792193390996e-05,
+      "loss": 0.3908,
+      "step": 1124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.49500998003992e-05,
+      "loss": 0.4125,
+      "step": 1125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4972277666888445e-05,
+      "loss": 0.418,
+      "step": 1126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.499445553337769e-05,
+      "loss": 0.4409,
+      "step": 1127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.501663339986693e-05,
+      "loss": 0.3947,
+      "step": 1128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5038811266356176e-05,
+      "loss": 0.3525,
+      "step": 1129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.506098913284542e-05,
+      "loss": 0.4109,
+      "step": 1130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.508316699933466e-05,
+      "loss": 0.3955,
+      "step": 1131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5105344865823904e-05,
+      "loss": 0.3694,
+      "step": 1132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.512752273231315e-05,
+      "loss": 0.4811,
+      "step": 1133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.514970059880239e-05,
+      "loss": 0.4157,
+      "step": 1134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5171878465291636e-05,
+      "loss": 0.4347,
+      "step": 1135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.519405633178088e-05,
+      "loss": 0.4138,
+      "step": 1136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5216234198270122e-05,
+      "loss": 0.4169,
+      "step": 1137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5238412064759367e-05,
+      "loss": 0.4408,
+      "step": 1138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5260589931248612e-05,
+      "loss": 0.4644,
+      "step": 1139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5282767797737853e-05,
+      "loss": 0.4488,
+      "step": 1140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5304945664227098e-05,
+      "loss": 0.4553,
+      "step": 1141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5327123530716343e-05,
+      "loss": 0.4262,
+      "step": 1142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5349301397205585e-05,
+      "loss": 0.397,
+      "step": 1143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.537147926369483e-05,
+      "loss": 0.3971,
+      "step": 1144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5393657130184074e-05,
+      "loss": 0.3942,
+      "step": 1145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.541583499667332e-05,
+      "loss": 0.3894,
+      "step": 1146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.543801286316256e-05,
+      "loss": 0.3772,
+      "step": 1147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5460190729651806e-05,
+      "loss": 0.4163,
+      "step": 1148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.548236859614105e-05,
+      "loss": 0.3856,
+      "step": 1149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5504546462630292e-05,
+      "loss": 0.4117,
+      "step": 1150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5526724329119537e-05,
+      "loss": 0.4069,
+      "step": 1151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5548902195608782e-05,
+      "loss": 0.3932,
+      "step": 1152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5571080062098023e-05,
+      "loss": 0.3906,
+      "step": 1153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.559325792858727e-05,
+      "loss": 0.3876,
+      "step": 1154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5615435795076513e-05,
+      "loss": 0.4202,
+      "step": 1155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5637613661565755e-05,
+      "loss": 0.4416,
+      "step": 1156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5659791528055e-05,
+      "loss": 0.4515,
+      "step": 1157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5681969394544245e-05,
+      "loss": 0.4634,
+      "step": 1158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.570414726103349e-05,
+      "loss": 0.4411,
+      "step": 1159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.572632512752273e-05,
+      "loss": 0.4234,
+      "step": 1160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5748502994011976e-05,
+      "loss": 0.4215,
+      "step": 1161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.577068086050122e-05,
+      "loss": 0.4206,
+      "step": 1162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5792858726990462e-05,
+      "loss": 0.4067,
+      "step": 1163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5815036593479704e-05,
+      "loss": 0.3543,
+      "step": 1164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5837214459968945e-05,
+      "loss": 0.4071,
+      "step": 1165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.585939232645819e-05,
+      "loss": 0.465,
+      "step": 1166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5881570192947435e-05,
+      "loss": 0.3837,
+      "step": 1167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.590374805943668e-05,
+      "loss": 0.3677,
+      "step": 1168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.592592592592592e-05,
+      "loss": 0.4073,
+      "step": 1169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5948103792415166e-05,
+      "loss": 0.3766,
+      "step": 1170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.597028165890441e-05,
+      "loss": 0.3963,
+      "step": 1171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5992459525393653e-05,
+      "loss": 0.4273,
+      "step": 1172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6014637391882898e-05,
+      "loss": 0.4524,
+      "step": 1173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6036815258372143e-05,
+      "loss": 0.4073,
+      "step": 1174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6058993124861384e-05,
+      "loss": 0.4227,
+      "step": 1175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.608117099135063e-05,
+      "loss": 0.4284,
+      "step": 1176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6103348857839874e-05,
+      "loss": 0.4366,
+      "step": 1177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6125526724329115e-05,
+      "loss": 0.4182,
+      "step": 1178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.614770459081836e-05,
+      "loss": 0.4293,
+      "step": 1179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6169882457307605e-05,
+      "loss": 0.423,
+      "step": 1180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.619206032379685e-05,
+      "loss": 0.4272,
+      "step": 1181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.621423819028609e-05,
+      "loss": 0.4344,
+      "step": 1182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6236416056775337e-05,
+      "loss": 0.4391,
+      "step": 1183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.625859392326458e-05,
+      "loss": 0.4129,
+      "step": 1184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6280771789753823e-05,
+      "loss": 0.4511,
+      "step": 1185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6302949656243068e-05,
+      "loss": 0.3833,
+      "step": 1186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6325127522732313e-05,
+      "loss": 0.4079,
+      "step": 1187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6347305389221554e-05,
+      "loss": 0.4261,
+      "step": 1188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.63694832557108e-05,
+      "loss": 0.4475,
+      "step": 1189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6391661122200044e-05,
+      "loss": 0.3984,
+      "step": 1190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6413838988689286e-05,
+      "loss": 0.4111,
+      "step": 1191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.643601685517853e-05,
+      "loss": 0.4357,
+      "step": 1192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6458194721667775e-05,
+      "loss": 0.3529,
+      "step": 1193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.648037258815702e-05,
+      "loss": 0.4267,
+      "step": 1194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6502550454646262e-05,
+      "loss": 0.3748,
+      "step": 1195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6524728321135507e-05,
+      "loss": 0.412,
+      "step": 1196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.654690618762475e-05,
+      "loss": 0.4031,
+      "step": 1197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.656908405411399e-05,
+      "loss": 0.3452,
+      "step": 1198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6591261920603235e-05,
+      "loss": 0.4022,
+      "step": 1199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6613439787092476e-05,
+      "loss": 0.4162,
+      "step": 1200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.663561765358172e-05,
+      "loss": 0.4047,
+      "step": 1201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6657795520070966e-05,
+      "loss": 0.3991,
+      "step": 1202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.667997338656021e-05,
+      "loss": 0.4391,
+      "step": 1203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6702151253049452e-05,
+      "loss": 0.4376,
+      "step": 1204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6724329119538697e-05,
+      "loss": 0.3859,
+      "step": 1205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6746506986027942e-05,
+      "loss": 0.4031,
+      "step": 1206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6768684852517184e-05,
+      "loss": 0.3979,
+      "step": 1207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.679086271900643e-05,
+      "loss": 0.3992,
+      "step": 1208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6813040585495673e-05,
+      "loss": 0.4921,
+      "step": 1209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6835218451984915e-05,
+      "loss": 0.4015,
+      "step": 1210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.685739631847416e-05,
+      "loss": 0.391,
+      "step": 1211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6879574184963405e-05,
+      "loss": 0.3989,
+      "step": 1212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6901752051452646e-05,
+      "loss": 0.4301,
+      "step": 1213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.692392991794189e-05,
+      "loss": 0.3496,
+      "step": 1214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6946107784431136e-05,
+      "loss": 0.4593,
+      "step": 1215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.696828565092038e-05,
+      "loss": 0.4449,
+      "step": 1216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6990463517409622e-05,
+      "loss": 0.4586,
+      "step": 1217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7012641383898867e-05,
+      "loss": 0.3986,
+      "step": 1218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7034819250388112e-05,
+      "loss": 0.3992,
+      "step": 1219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7056997116877354e-05,
+      "loss": 0.4566,
+      "step": 1220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.70791749833666e-05,
+      "loss": 0.397,
+      "step": 1221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7101352849855844e-05,
+      "loss": 0.3654,
+      "step": 1222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7123530716345085e-05,
+      "loss": 0.3789,
+      "step": 1223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.714570858283433e-05,
+      "loss": 0.3727,
+      "step": 1224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7167886449323575e-05,
+      "loss": 0.4127,
+      "step": 1225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7190064315812816e-05,
+      "loss": 0.3794,
+      "step": 1226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.721224218230206e-05,
+      "loss": 0.3972,
+      "step": 1227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7234420048791306e-05,
+      "loss": 0.3696,
+      "step": 1228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.725659791528055e-05,
+      "loss": 0.401,
+      "step": 1229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7278775781769793e-05,
+      "loss": 0.4149,
+      "step": 1230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7300953648259037e-05,
+      "loss": 0.4782,
+      "step": 1231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7323131514748276e-05,
+      "loss": 0.3952,
+      "step": 1232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.734530938123752e-05,
+      "loss": 0.3995,
+      "step": 1233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7367487247726765e-05,
+      "loss": 0.4211,
+      "step": 1234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7389665114216007e-05,
+      "loss": 0.4338,
+      "step": 1235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7411842980705252e-05,
+      "loss": 0.4174,
+      "step": 1236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7434020847194497e-05,
+      "loss": 0.4083,
+      "step": 1237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.745619871368374e-05,
+      "loss": 0.4252,
+      "step": 1238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7478376580172983e-05,
+      "loss": 0.4334,
+      "step": 1239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7500554446662228e-05,
+      "loss": 0.397,
+      "step": 1240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7522732313151473e-05,
+      "loss": 0.4376,
+      "step": 1241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7544910179640714e-05,
+      "loss": 0.4131,
+      "step": 1242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.756708804612996e-05,
+      "loss": 0.389,
+      "step": 1243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7589265912619204e-05,
+      "loss": 0.3969,
+      "step": 1244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7611443779108446e-05,
+      "loss": 0.443,
+      "step": 1245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.763362164559769e-05,
+      "loss": 0.4385,
+      "step": 1246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7655799512086936e-05,
+      "loss": 0.4044,
+      "step": 1247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7677977378576177e-05,
+      "loss": 0.4441,
+      "step": 1248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7700155245065422e-05,
+      "loss": 0.3804,
+      "step": 1249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7722333111554667e-05,
+      "loss": 0.4116,
+      "step": 1250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7744510978043912e-05,
+      "loss": 0.3786,
+      "step": 1251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7766688844533153e-05,
+      "loss": 0.4411,
+      "step": 1252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7788866711022398e-05,
+      "loss": 0.3821,
+      "step": 1253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7811044577511643e-05,
+      "loss": 0.3907,
+      "step": 1254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7833222444000885e-05,
+      "loss": 0.3761,
+      "step": 1255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.785540031049013e-05,
+      "loss": 0.407,
+      "step": 1256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7877578176979374e-05,
+      "loss": 0.3985,
+      "step": 1257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7899756043468616e-05,
+      "loss": 0.4008,
+      "step": 1258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.792193390995786e-05,
+      "loss": 0.4176,
+      "step": 1259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7944111776447106e-05,
+      "loss": 0.4059,
+      "step": 1260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7966289642936347e-05,
+      "loss": 0.3959,
+      "step": 1261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7988467509425592e-05,
+      "loss": 0.4448,
+      "step": 1262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8010645375914837e-05,
+      "loss": 0.4142,
+      "step": 1263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8032823242404082e-05,
+      "loss": 0.4256,
+      "step": 1264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.805500110889332e-05,
+      "loss": 0.3794,
+      "step": 1265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8077178975382565e-05,
+      "loss": 0.3846,
+      "step": 1266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8099356841871806e-05,
+      "loss": 0.4257,
+      "step": 1267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.812153470836105e-05,
+      "loss": 0.3745,
+      "step": 1268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8143712574850296e-05,
+      "loss": 0.4515,
+      "step": 1269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8165890441339538e-05,
+      "loss": 0.4216,
+      "step": 1270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8188068307828783e-05,
+      "loss": 0.453,
+      "step": 1271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8210246174318028e-05,
+      "loss": 0.4292,
+      "step": 1272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8232424040807272e-05,
+      "loss": 0.4394,
+      "step": 1273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8254601907296514e-05,
+      "loss": 0.396,
+      "step": 1274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.827677977378576e-05,
+      "loss": 0.4122,
+      "step": 1275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8298957640275004e-05,
+      "loss": 0.4243,
+      "step": 1276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8321135506764245e-05,
+      "loss": 0.3983,
+      "step": 1277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.834331337325349e-05,
+      "loss": 0.3854,
+      "step": 1278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8365491239742735e-05,
+      "loss": 0.4247,
+      "step": 1279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8387669106231977e-05,
+      "loss": 0.4198,
+      "step": 1280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.840984697272122e-05,
+      "loss": 0.3772,
+      "step": 1281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8432024839210466e-05,
+      "loss": 0.4204,
+      "step": 1282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8454202705699708e-05,
+      "loss": 0.4102,
+      "step": 1283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8476380572188953e-05,
+      "loss": 0.3934,
+      "step": 1284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8498558438678198e-05,
+      "loss": 0.4385,
+      "step": 1285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8520736305167443e-05,
+      "loss": 0.3713,
+      "step": 1286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8542914171656684e-05,
+      "loss": 0.4148,
+      "step": 1287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.856509203814593e-05,
+      "loss": 0.387,
+      "step": 1288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8587269904635174e-05,
+      "loss": 0.4115,
+      "step": 1289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8609447771124415e-05,
+      "loss": 0.42,
+      "step": 1290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.863162563761366e-05,
+      "loss": 0.365,
+      "step": 1291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8653803504102905e-05,
+      "loss": 0.4202,
+      "step": 1292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8675981370592147e-05,
+      "loss": 0.3927,
+      "step": 1293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.869815923708139e-05,
+      "loss": 0.4535,
+      "step": 1294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8720337103570636e-05,
+      "loss": 0.3939,
+      "step": 1295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8742514970059878e-05,
+      "loss": 0.3745,
+      "step": 1296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8764692836549123e-05,
+      "loss": 0.4351,
+      "step": 1297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8786870703038368e-05,
+      "loss": 0.3861,
+      "step": 1298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8809048569527606e-05,
+      "loss": 0.4136,
+      "step": 1299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.883122643601685e-05,
+      "loss": 0.4225,
+      "step": 1300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8853404302506096e-05,
+      "loss": 0.3896,
+      "step": 1301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8875582168995337e-05,
+      "loss": 0.4233,
+      "step": 1302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8897760035484582e-05,
+      "loss": 0.3697,
+      "step": 1303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8919937901973827e-05,
+      "loss": 0.4055,
+      "step": 1304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.894211576846307e-05,
+      "loss": 0.4319,
+      "step": 1305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8964293634952313e-05,
+      "loss": 0.3987,
+      "step": 1306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.898647150144156e-05,
+      "loss": 0.4213,
+      "step": 1307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9008649367930803e-05,
+      "loss": 0.4416,
+      "step": 1308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9030827234420045e-05,
+      "loss": 0.3828,
+      "step": 1309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.905300510090929e-05,
+      "loss": 0.4052,
+      "step": 1310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9075182967398535e-05,
+      "loss": 0.455,
+      "step": 1311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9097360833887776e-05,
+      "loss": 0.3537,
+      "step": 1312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.911953870037702e-05,
+      "loss": 0.3815,
+      "step": 1313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9141716566866266e-05,
+      "loss": 0.4346,
+      "step": 1314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9163894433355507e-05,
+      "loss": 0.405,
+      "step": 1315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9186072299844752e-05,
+      "loss": 0.4075,
+      "step": 1316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9208250166333997e-05,
+      "loss": 0.4172,
+      "step": 1317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.923042803282324e-05,
+      "loss": 0.402,
+      "step": 1318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9252605899312484e-05,
+      "loss": 0.4017,
+      "step": 1319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.927478376580173e-05,
+      "loss": 0.4351,
+      "step": 1320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9296961632290973e-05,
+      "loss": 0.3701,
+      "step": 1321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9319139498780215e-05,
+      "loss": 0.4055,
+      "step": 1322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.934131736526946e-05,
+      "loss": 0.4054,
+      "step": 1323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9363495231758705e-05,
+      "loss": 0.3904,
+      "step": 1324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9385673098247946e-05,
+      "loss": 0.4217,
+      "step": 1325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.940785096473719e-05,
+      "loss": 0.3834,
+      "step": 1326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9430028831226436e-05,
+      "loss": 0.3656,
+      "step": 1327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9452206697715678e-05,
+      "loss": 0.4471,
+      "step": 1328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9474384564204922e-05,
+      "loss": 0.4095,
+      "step": 1329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9496562430694167e-05,
+      "loss": 0.4246,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.951874029718341e-05,
+      "loss": 0.41,
+      "step": 1331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9540918163672654e-05,
+      "loss": 0.4266,
+      "step": 1332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9563096030161895e-05,
+      "loss": 0.3966,
+      "step": 1333
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9585273896651137e-05,
+      "loss": 0.3996,
+      "step": 1334
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.960745176314038e-05,
+      "loss": 0.4733,
+      "step": 1335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9629629629629627e-05,
+      "loss": 0.4119,
+      "step": 1336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9651807496118868e-05,
+      "loss": 0.4078,
+      "step": 1337
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9673985362608113e-05,
+      "loss": 0.3887,
+      "step": 1338
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9696163229097358e-05,
+      "loss": 0.4258,
+      "step": 1339
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.97183410955866e-05,
+      "loss": 0.4484,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9740518962075844e-05,
+      "loss": 0.4036,
+      "step": 1341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.976269682856509e-05,
+      "loss": 0.3935,
+      "step": 1342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9784874695054334e-05,
+      "loss": 0.4223,
+      "step": 1343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9807052561543576e-05,
+      "loss": 0.3986,
+      "step": 1344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982923042803282e-05,
+      "loss": 0.3949,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9851408294522065e-05,
+      "loss": 0.4392,
+      "step": 1346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9873586161011307e-05,
+      "loss": 0.4308,
+      "step": 1347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9895764027500552e-05,
+      "loss": 0.4467,
+      "step": 1348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9917941893989797e-05,
+      "loss": 0.4158,
+      "step": 1349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9940119760479038e-05,
+      "loss": 0.4652,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9962297626968283e-05,
+      "loss": 0.3738,
+      "step": 1351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9984475493457528e-05,
+      "loss": 0.416,
+      "step": 1352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.000665335994677e-05,
+      "loss": 0.4057,
+      "step": 1353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0028831226436014e-05,
+      "loss": 0.4095,
+      "step": 1354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.005100909292526e-05,
+      "loss": 0.4376,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0073186959414504e-05,
+      "loss": 0.4278,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0095364825903746e-05,
+      "loss": 0.4099,
+      "step": 1357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.011754269239299e-05,
+      "loss": 0.36,
+      "step": 1358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0139720558882235e-05,
+      "loss": 0.4081,
+      "step": 1359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0161898425371477e-05,
+      "loss": 0.4475,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0184076291860722e-05,
+      "loss": 0.4121,
+      "step": 1361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0206254158349967e-05,
+      "loss": 0.3929,
+      "step": 1362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.022843202483921e-05,
+      "loss": 0.4144,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0250609891328453e-05,
+      "loss": 0.4106,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0272787757817698e-05,
+      "loss": 0.4143,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.029496562430694e-05,
+      "loss": 0.4406,
+      "step": 1366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.031714349079618e-05,
+      "loss": 0.3782,
+      "step": 1367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0339321357285426e-05,
+      "loss": 0.4306,
+      "step": 1368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0361499223774668e-05,
+      "loss": 0.3524,
+      "step": 1369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0383677090263912e-05,
+      "loss": 0.3927,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0405854956753157e-05,
+      "loss": 0.372,
+      "step": 1371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.04280328232424e-05,
+      "loss": 0.4022,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0450210689731644e-05,
+      "loss": 0.4186,
+      "step": 1373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.047238855622089e-05,
+      "loss": 0.4138,
+      "step": 1374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.049456642271013e-05,
+      "loss": 0.4046,
+      "step": 1375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.0516744289199375e-05,
+      "loss": 0.3757,
+      "step": 1376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.053892215568862e-05,
+      "loss": 0.3929,
+      "step": 1377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.0561100022177865e-05,
+      "loss": 0.4064,
+      "step": 1378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.058327788866711e-05,
+      "loss": 0.4547,
+      "step": 1379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.060545575515635e-05,
+      "loss": 0.4352,
+      "step": 1380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.062763362164559e-05,
+      "loss": 0.4239,
+      "step": 1381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.064981148813484e-05,
+      "loss": 0.4385,
+      "step": 1382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.067198935462408e-05,
+      "loss": 0.3763,
+      "step": 1383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.069416722111333e-05,
+      "loss": 0.4148,
+      "step": 1384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.071634508760257e-05,
+      "loss": 0.4368,
+      "step": 1385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.073852295409182e-05,
+      "loss": 0.3785,
+      "step": 1386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.0760700820581055e-05,
+      "loss": 0.3315,
+      "step": 1387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.07828786870703e-05,
+      "loss": 0.3958,
+      "step": 1388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.0805056553559545e-05,
+      "loss": 0.382,
+      "step": 1389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.082723442004879e-05,
+      "loss": 0.449,
+      "step": 1390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.0849412286538035e-05,
+      "loss": 0.4323,
+      "step": 1391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.087159015302728e-05,
+      "loss": 0.4274,
+      "step": 1392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.089376801951652e-05,
+      "loss": 0.4075,
+      "step": 1393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.091594588600576e-05,
+      "loss": 0.4184,
+      "step": 1394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.093812375249501e-05,
+      "loss": 0.3826,
+      "step": 1395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.096030161898425e-05,
+      "loss": 0.3944,
+      "step": 1396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.09824794854735e-05,
+      "loss": 0.4488,
+      "step": 1397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.100465735196274e-05,
+      "loss": 0.3769,
+      "step": 1398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.102683521845199e-05,
+      "loss": 0.4551,
+      "step": 1399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1049013084941226e-05,
+      "loss": 0.3843,
+      "step": 1400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.107119095143047e-05,
+      "loss": 0.4295,
+      "step": 1401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.109336881791971e-05,
+      "loss": 0.4178,
+      "step": 1402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1115546684408953e-05,
+      "loss": 0.3963,
+      "step": 1403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.11377245508982e-05,
+      "loss": 0.4192,
+      "step": 1404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.115990241738744e-05,
+      "loss": 0.4099,
+      "step": 1405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.118208028387669e-05,
+      "loss": 0.4048,
+      "step": 1406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.120425815036593e-05,
+      "loss": 0.3979,
+      "step": 1407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.122643601685518e-05,
+      "loss": 0.3844,
+      "step": 1408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1248613883344416e-05,
+      "loss": 0.4614,
+      "step": 1409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.127079174983366e-05,
+      "loss": 0.3734,
+      "step": 1410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1292969616322906e-05,
+      "loss": 0.4079,
+      "step": 1411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.131514748281215e-05,
+      "loss": 0.4395,
+      "step": 1412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1337325349301396e-05,
+      "loss": 0.4157,
+      "step": 1413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.135950321579064e-05,
+      "loss": 0.4211,
+      "step": 1414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.138168108227988e-05,
+      "loss": 0.4067,
+      "step": 1415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1403858948769124e-05,
+      "loss": 0.4284,
+      "step": 1416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.142603681525837e-05,
+      "loss": 0.4393,
+      "step": 1417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.144821468174761e-05,
+      "loss": 0.4437,
+      "step": 1418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.147039254823686e-05,
+      "loss": 0.4286,
+      "step": 1419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.14925704147261e-05,
+      "loss": 0.3696,
+      "step": 1420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.151474828121535e-05,
+      "loss": 0.4318,
+      "step": 1421
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1536926147704586e-05,
+      "loss": 0.3992,
+      "step": 1422
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.155910401419383e-05,
+      "loss": 0.4072,
+      "step": 1423
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1581281880683076e-05,
+      "loss": 0.4604,
+      "step": 1424
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.160345974717232e-05,
+      "loss": 0.3946,
+      "step": 1425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1625637613661566e-05,
+      "loss": 0.4163,
+      "step": 1426
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.164781548015081e-05,
+      "loss": 0.3793,
+      "step": 1427
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.166999334664005e-05,
+      "loss": 0.397,
+      "step": 1428
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1692171213129294e-05,
+      "loss": 0.3912,
+      "step": 1429
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.171434907961854e-05,
+      "loss": 0.3872,
+      "step": 1430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1736526946107784e-05,
+      "loss": 0.4228,
+      "step": 1431
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.175870481259703e-05,
+      "loss": 0.3939,
+      "step": 1432
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.178088267908627e-05,
+      "loss": 0.3997,
+      "step": 1433
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.180306054557551e-05,
+      "loss": 0.3786,
+      "step": 1434
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1825238412064756e-05,
+      "loss": 0.3826,
+      "step": 1435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1847416278554e-05,
+      "loss": 0.3781,
+      "step": 1436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.186959414504324e-05,
+      "loss": 0.3705,
+      "step": 1437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1891772011532484e-05,
+      "loss": 0.3839,
+      "step": 1438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.191394987802173e-05,
+      "loss": 0.3811,
+      "step": 1439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1936127744510974e-05,
+      "loss": 0.4113,
+      "step": 1440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.195830561100022e-05,
+      "loss": 0.4345,
+      "step": 1441
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1980483477489464e-05,
+      "loss": 0.3784,
+      "step": 1442
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.200266134397871e-05,
+      "loss": 0.4088,
+      "step": 1443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.202483921046795e-05,
+      "loss": 0.3921,
+      "step": 1444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.204701707695719e-05,
+      "loss": 0.3936,
+      "step": 1445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.206919494344644e-05,
+      "loss": 0.4358,
+      "step": 1446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.209137280993568e-05,
+      "loss": 0.4541,
+      "step": 1447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2113550676424926e-05,
+      "loss": 0.396,
+      "step": 1448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.213572854291417e-05,
+      "loss": 0.4653,
+      "step": 1449
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.215790640940341e-05,
+      "loss": 0.4138,
+      "step": 1450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2180084275892654e-05,
+      "loss": 0.4516,
+      "step": 1451
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.22022621423819e-05,
+      "loss": 0.4048,
+      "step": 1452
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2224440008871144e-05,
+      "loss": 0.4016,
+      "step": 1453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.224661787536039e-05,
+      "loss": 0.3418,
+      "step": 1454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2268795741849634e-05,
+      "loss": 0.4083,
+      "step": 1455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.229097360833888e-05,
+      "loss": 0.3779,
+      "step": 1456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.231315147482812e-05,
+      "loss": 0.3866,
+      "step": 1457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.233532934131736e-05,
+      "loss": 0.3903,
+      "step": 1458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.235750720780661e-05,
+      "loss": 0.3938,
+      "step": 1459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.237968507429585e-05,
+      "loss": 0.377,
+      "step": 1460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2401862940785097e-05,
+      "loss": 0.4068,
+      "step": 1461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.242404080727434e-05,
+      "loss": 0.4174,
+      "step": 1462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.244621867376358e-05,
+      "loss": 0.394,
+      "step": 1463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2468396540252825e-05,
+      "loss": 0.4125,
+      "step": 1464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.249057440674207e-05,
+      "loss": 0.424,
+      "step": 1465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.2512752273231314e-05,
+      "loss": 0.423,
+      "step": 1466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.253493013972056e-05,
+      "loss": 0.411,
+      "step": 1467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.25571080062098e-05,
+      "loss": 0.4145,
+      "step": 1468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.257928587269904e-05,
+      "loss": 0.3737,
+      "step": 1469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.260146373918829e-05,
+      "loss": 0.4175,
+      "step": 1470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.262364160567753e-05,
+      "loss": 0.4329,
+      "step": 1471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.264581947216677e-05,
+      "loss": 0.3566,
+      "step": 1472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.2667997338656015e-05,
+      "loss": 0.4005,
+      "step": 1473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.269017520514526e-05,
+      "loss": 0.3896,
+      "step": 1474
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.2712353071634505e-05,
+      "loss": 0.4803,
+      "step": 1475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.273453093812375e-05,
+      "loss": 0.3704,
+      "step": 1476
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.2756708804612995e-05,
+      "loss": 0.3964,
+      "step": 1477
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.277888667110224e-05,
+      "loss": 0.3878,
+      "step": 1478
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.280106453759148e-05,
+      "loss": 0.4076,
+      "step": 1479
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.282324240408072e-05,
+      "loss": 0.4211,
+      "step": 1480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.284542027056997e-05,
+      "loss": 0.3926,
+      "step": 1481
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.286759813705921e-05,
+      "loss": 0.387,
+      "step": 1482
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.288977600354846e-05,
+      "loss": 0.4153,
+      "step": 1483
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.29119538700377e-05,
+      "loss": 0.3917,
+      "step": 1484
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.293413173652694e-05,
+      "loss": 0.3921,
+      "step": 1485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.2956309603016185e-05,
+      "loss": 0.4333,
+      "step": 1486
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.297848746950543e-05,
+      "loss": 0.3943,
+      "step": 1487
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3000665335994675e-05,
+      "loss": 0.3701,
+      "step": 1488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.302284320248392e-05,
+      "loss": 0.4057,
+      "step": 1489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3045021068973165e-05,
+      "loss": 0.3848,
+      "step": 1490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.306719893546241e-05,
+      "loss": 0.3436,
+      "step": 1491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.308937680195165e-05,
+      "loss": 0.4104,
+      "step": 1492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.311155466844089e-05,
+      "loss": 0.381,
+      "step": 1493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.313373253493014e-05,
+      "loss": 0.3759,
+      "step": 1494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.315591040141938e-05,
+      "loss": 0.4178,
+      "step": 1495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.317808826790863e-05,
+      "loss": 0.4116,
+      "step": 1496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.320026613439787e-05,
+      "loss": 0.3871,
+      "step": 1497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.322244400088711e-05,
+      "loss": 0.3749,
+      "step": 1498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3244621867376355e-05,
+      "loss": 0.4124,
+      "step": 1499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.32667997338656e-05,
+      "loss": 0.3911,
+      "step": 1500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3288977600354845e-05,
+      "loss": 0.3984,
+      "step": 1501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.331115546684408e-05,
+      "loss": 0.3879,
+      "step": 1502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.333333333333333e-05,
+      "loss": 0.3936,
+      "step": 1503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.335551119982257e-05,
+      "loss": 0.4059,
+      "step": 1504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.337768906631182e-05,
+      "loss": 0.4139,
+      "step": 1505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.339986693280106e-05,
+      "loss": 0.3883,
+      "step": 1506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.34220447992903e-05,
+      "loss": 0.4262,
+      "step": 1507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3444222665779546e-05,
+      "loss": 0.4224,
+      "step": 1508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.346640053226879e-05,
+      "loss": 0.402,
+      "step": 1509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3488578398758036e-05,
+      "loss": 0.4106,
+      "step": 1510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.351075626524728e-05,
+      "loss": 0.4065,
+      "step": 1511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3532934131736525e-05,
+      "loss": 0.4014,
+      "step": 1512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.355511199822577e-05,
+      "loss": 0.3755,
+      "step": 1513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.357728986471501e-05,
+      "loss": 0.4354,
+      "step": 1514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3599467731204253e-05,
+      "loss": 0.4315,
+      "step": 1515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.36216455976935e-05,
+      "loss": 0.4024,
+      "step": 1516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.364382346418274e-05,
+      "loss": 0.4207,
+      "step": 1517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.366600133067199e-05,
+      "loss": 0.3886,
+      "step": 1518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.368817919716123e-05,
+      "loss": 0.4318,
+      "step": 1519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.371035706365047e-05,
+      "loss": 0.3709,
+      "step": 1520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3732534930139716e-05,
+      "loss": 0.4216,
+      "step": 1521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.375471279662896e-05,
+      "loss": 0.4355,
+      "step": 1522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3776890663118206e-05,
+      "loss": 0.4387,
+      "step": 1523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.379906852960745e-05,
+      "loss": 0.4,
+      "step": 1524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3821246396096696e-05,
+      "loss": 0.4094,
+      "step": 1525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.384342426258594e-05,
+      "loss": 0.4197,
+      "step": 1526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.386560212907518e-05,
+      "loss": 0.346,
+      "step": 1527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3887779995564424e-05,
+      "loss": 0.365,
+      "step": 1528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.390995786205367e-05,
+      "loss": 0.4241,
+      "step": 1529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.393213572854291e-05,
+      "loss": 0.4048,
+      "step": 1530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.395431359503216e-05,
+      "loss": 0.4231,
+      "step": 1531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.39764914615214e-05,
+      "loss": 0.3805,
+      "step": 1532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.399866932801064e-05,
+      "loss": 0.3909,
+      "step": 1533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4020847194499886e-05,
+      "loss": 0.4023,
+      "step": 1534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.404302506098913e-05,
+      "loss": 0.3792,
+      "step": 1535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.406520292747837e-05,
+      "loss": 0.4297,
+      "step": 1536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4087380793967614e-05,
+      "loss": 0.4119,
+      "step": 1537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.410955866045686e-05,
+      "loss": 0.4307,
+      "step": 1538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4131736526946104e-05,
+      "loss": 0.4095,
+      "step": 1539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.415391439343535e-05,
+      "loss": 0.4105,
+      "step": 1540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4176092259924594e-05,
+      "loss": 0.4036,
+      "step": 1541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.419827012641383e-05,
+      "loss": 0.4122,
+      "step": 1542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.422044799290308e-05,
+      "loss": 0.4126,
+      "step": 1543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.424262585939232e-05,
+      "loss": 0.3794,
+      "step": 1544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4264803725881566e-05,
+      "loss": 0.4177,
+      "step": 1545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.428698159237081e-05,
+      "loss": 0.3942,
+      "step": 1546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4309159458860056e-05,
+      "loss": 0.3977,
+      "step": 1547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.43313373253493e-05,
+      "loss": 0.3995,
+      "step": 1548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.435351519183854e-05,
+      "loss": 0.3946,
+      "step": 1549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4375693058327784e-05,
+      "loss": 0.4228,
+      "step": 1550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.439787092481703e-05,
+      "loss": 0.3972,
+      "step": 1551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4420048791306274e-05,
+      "loss": 0.42,
+      "step": 1552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.444222665779552e-05,
+      "loss": 0.429,
+      "step": 1553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4464404524284764e-05,
+      "loss": 0.3921,
+      "step": 1554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4486582390774e-05,
+      "loss": 0.3957,
+      "step": 1555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.450876025726325e-05,
+      "loss": 0.4033,
+      "step": 1556
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.453093812375249e-05,
+      "loss": 0.456,
+      "step": 1557
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.455311599024174e-05,
+      "loss": 0.4417,
+      "step": 1558
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.457529385673098e-05,
+      "loss": 0.4464,
+      "step": 1559
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4597471723220226e-05,
+      "loss": 0.4002,
+      "step": 1560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.461964958970947e-05,
+      "loss": 0.3866,
+      "step": 1561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.464182745619871e-05,
+      "loss": 0.4145,
+      "step": 1562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4664005322687954e-05,
+      "loss": 0.3569,
+      "step": 1563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.46861831891772e-05,
+      "loss": 0.3738,
+      "step": 1564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4708361055666444e-05,
+      "loss": 0.4861,
+      "step": 1565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.473053892215569e-05,
+      "loss": 0.3784,
+      "step": 1566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4752716788644934e-05,
+      "loss": 0.3784,
+      "step": 1567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.477489465513417e-05,
+      "loss": 0.3777,
+      "step": 1568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.479707252162342e-05,
+      "loss": 0.4004,
+      "step": 1569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.481925038811266e-05,
+      "loss": 0.3949,
+      "step": 1570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.48414282546019e-05,
+      "loss": 0.4137,
+      "step": 1571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4863606121091145e-05,
+      "loss": 0.4452,
+      "step": 1572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.488578398758039e-05,
+      "loss": 0.4036,
+      "step": 1573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4907961854069635e-05,
+      "loss": 0.3885,
+      "step": 1574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.493013972055888e-05,
+      "loss": 0.3865,
+      "step": 1575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4952317587048124e-05,
+      "loss": 0.344,
+      "step": 1576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.497449545353736e-05,
+      "loss": 0.3668,
+      "step": 1577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.499667332002661e-05,
+      "loss": 0.3638,
+      "step": 1578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.501885118651585e-05,
+      "loss": 0.4133,
+      "step": 1579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.50410290530051e-05,
+      "loss": 0.3528,
+      "step": 1580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.506320691949434e-05,
+      "loss": 0.4158,
+      "step": 1581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.508538478598359e-05,
+      "loss": 0.3566,
+      "step": 1582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.510756265247283e-05,
+      "loss": 0.3995,
+      "step": 1583
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.512974051896207e-05,
+      "loss": 0.3766,
+      "step": 1584
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5151918385451315e-05,
+      "loss": 0.4224,
+      "step": 1585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.517409625194056e-05,
+      "loss": 0.3885,
+      "step": 1586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5196274118429805e-05,
+      "loss": 0.4002,
+      "step": 1587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.521845198491905e-05,
+      "loss": 0.3917,
+      "step": 1588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5240629851408295e-05,
+      "loss": 0.3665,
+      "step": 1589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.526280771789753e-05,
+      "loss": 0.4202,
+      "step": 1590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.528498558438678e-05,
+      "loss": 0.3953,
+      "step": 1591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.530716345087602e-05,
+      "loss": 0.4072,
+      "step": 1592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.532934131736527e-05,
+      "loss": 0.3905,
+      "step": 1593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.535151918385451e-05,
+      "loss": 0.3915,
+      "step": 1594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.537369705034376e-05,
+      "loss": 0.4004,
+      "step": 1595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5395874916833e-05,
+      "loss": 0.4005,
+      "step": 1596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.541805278332224e-05,
+      "loss": 0.4209,
+      "step": 1597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5440230649811485e-05,
+      "loss": 0.4497,
+      "step": 1598
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.546240851630073e-05,
+      "loss": 0.3828,
+      "step": 1599
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5484586382789975e-05,
+      "loss": 0.4123,
+      "step": 1600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.550676424927922e-05,
+      "loss": 0.3812,
+      "step": 1601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5528942115768465e-05,
+      "loss": 0.4221,
+      "step": 1602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.55511199822577e-05,
+      "loss": 0.3923,
+      "step": 1603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.557329784874695e-05,
+      "loss": 0.3849,
+      "step": 1604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.559547571523619e-05,
+      "loss": 0.3604,
+      "step": 1605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.561765358172543e-05,
+      "loss": 0.3861,
+      "step": 1606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5639831448214676e-05,
+      "loss": 0.4047,
+      "step": 1607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.566200931470392e-05,
+      "loss": 0.3722,
+      "step": 1608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5684187181193165e-05,
+      "loss": 0.3743,
+      "step": 1609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.570636504768241e-05,
+      "loss": 0.3761,
+      "step": 1610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5728542914171655e-05,
+      "loss": 0.3904,
+      "step": 1611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5750720780660893e-05,
+      "loss": 0.3986,
+      "step": 1612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.577289864715014e-05,
+      "loss": 0.3615,
+      "step": 1613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.579507651363938e-05,
+      "loss": 0.3974,
+      "step": 1614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.581725438012863e-05,
+      "loss": 0.3916,
+      "step": 1615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.583943224661787e-05,
+      "loss": 0.3679,
+      "step": 1616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.586161011310712e-05,
+      "loss": 0.4208,
+      "step": 1617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.588378797959636e-05,
+      "loss": 0.4058,
+      "step": 1618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.59059658460856e-05,
+      "loss": 0.4108,
+      "step": 1619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5928143712574846e-05,
+      "loss": 0.3864,
+      "step": 1620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.595032157906409e-05,
+      "loss": 0.3677,
+      "step": 1621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5972499445553336e-05,
+      "loss": 0.4138,
+      "step": 1622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.599467731204258e-05,
+      "loss": 0.4275,
+      "step": 1623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6016855178531825e-05,
+      "loss": 0.4405,
+      "step": 1624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6039033045021064e-05,
+      "loss": 0.4163,
+      "step": 1625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.606121091151031e-05,
+      "loss": 0.3817,
+      "step": 1626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.608338877799955e-05,
+      "loss": 0.3927,
+      "step": 1627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.61055666444888e-05,
+      "loss": 0.3645,
+      "step": 1628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.612774451097804e-05,
+      "loss": 0.3879,
+      "step": 1629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.614992237746729e-05,
+      "loss": 0.3873,
+      "step": 1630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.617210024395653e-05,
+      "loss": 0.4152,
+      "step": 1631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.619427811044577e-05,
+      "loss": 0.4167,
+      "step": 1632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6216455976935016e-05,
+      "loss": 0.3797,
+      "step": 1633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.623863384342426e-05,
+      "loss": 0.4023,
+      "step": 1634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6260811709913506e-05,
+      "loss": 0.4179,
+      "step": 1635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.628298957640275e-05,
+      "loss": 0.4015,
+      "step": 1636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.630516744289199e-05,
+      "loss": 0.4163,
+      "step": 1637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6327345309381234e-05,
+      "loss": 0.3886,
+      "step": 1638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.634952317587048e-05,
+      "loss": 0.4364,
+      "step": 1639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6371701042359723e-05,
+      "loss": 0.3695,
+      "step": 1640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.639387890884896e-05,
+      "loss": 0.41,
+      "step": 1641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6416056775338206e-05,
+      "loss": 0.3797,
+      "step": 1642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.643823464182745e-05,
+      "loss": 0.4073,
+      "step": 1643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6460412508316696e-05,
+      "loss": 0.3807,
+      "step": 1644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.648259037480594e-05,
+      "loss": 0.3734,
+      "step": 1645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6504768241295186e-05,
+      "loss": 0.4448,
+      "step": 1646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6526946107784424e-05,
+      "loss": 0.3885,
+      "step": 1647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.654912397427367e-05,
+      "loss": 0.3868,
+      "step": 1648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6571301840762914e-05,
+      "loss": 0.3675,
+      "step": 1649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.659347970725216e-05,
+      "loss": 0.4024,
+      "step": 1650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6615657573741404e-05,
+      "loss": 0.3682,
+      "step": 1651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.663783544023065e-05,
+      "loss": 0.4276,
+      "step": 1652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6660013306719894e-05,
+      "loss": 0.4286,
+      "step": 1653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.668219117320913e-05,
+      "loss": 0.3968,
+      "step": 1654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.670436903969838e-05,
+      "loss": 0.3958,
+      "step": 1655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.672654690618762e-05,
+      "loss": 0.402,
+      "step": 1656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6748724772676866e-05,
+      "loss": 0.398,
+      "step": 1657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.677090263916611e-05,
+      "loss": 0.4102,
+      "step": 1658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6793080505655356e-05,
+      "loss": 0.3396,
+      "step": 1659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6815258372144594e-05,
+      "loss": 0.3935,
+      "step": 1660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.683743623863384e-05,
+      "loss": 0.4107,
+      "step": 1661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6859614105123084e-05,
+      "loss": 0.4494,
+      "step": 1662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.688179197161233e-05,
+      "loss": 0.3792,
+      "step": 1663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6903969838101574e-05,
+      "loss": 0.4047,
+      "step": 1664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.692614770459082e-05,
+      "loss": 0.4056,
+      "step": 1665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6948325571080064e-05,
+      "loss": 0.4762,
+      "step": 1666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.69705034375693e-05,
+      "loss": 0.385,
+      "step": 1667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.699268130405855e-05,
+      "loss": 0.4077,
+      "step": 1668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.701485917054779e-05,
+      "loss": 0.3948,
+      "step": 1669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.703703703703703e-05,
+      "loss": 0.3521,
+      "step": 1670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7059214903526275e-05,
+      "loss": 0.3615,
+      "step": 1671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.708139277001552e-05,
+      "loss": 0.378,
+      "step": 1672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7103570636504764e-05,
+      "loss": 0.3637,
+      "step": 1673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.712574850299401e-05,
+      "loss": 0.4156,
+      "step": 1674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7147926369483254e-05,
+      "loss": 0.4525,
+      "step": 1675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.717010423597249e-05,
+      "loss": 0.4095,
+      "step": 1676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.719228210246174e-05,
+      "loss": 0.4015,
+      "step": 1677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.721445996895098e-05,
+      "loss": 0.3398,
+      "step": 1678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.723663783544023e-05,
+      "loss": 0.3915,
+      "step": 1679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.725881570192947e-05,
+      "loss": 0.3957,
+      "step": 1680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.728099356841872e-05,
+      "loss": 0.4162,
+      "step": 1681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7303171434907955e-05,
+      "loss": 0.3922,
+      "step": 1682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.73253493013972e-05,
+      "loss": 0.4399,
+      "step": 1683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7347527167886445e-05,
+      "loss": 0.4163,
+      "step": 1684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.736970503437569e-05,
+      "loss": 0.3594,
+      "step": 1685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7391882900864935e-05,
+      "loss": 0.3785,
+      "step": 1686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.741406076735418e-05,
+      "loss": 0.3938,
+      "step": 1687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7436238633843424e-05,
+      "loss": 0.4085,
+      "step": 1688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.745841650033266e-05,
+      "loss": 0.3981,
+      "step": 1689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.748059436682191e-05,
+      "loss": 0.413,
+      "step": 1690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7502772233311146e-05,
+      "loss": 0.3545,
+      "step": 1691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.75249500998004e-05,
+      "loss": 0.4125,
+      "step": 1692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7547127966289635e-05,
+      "loss": 0.4054,
+      "step": 1693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.756930583277889e-05,
+      "loss": 0.3931,
+      "step": 1694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7591483699268125e-05,
+      "loss": 0.4098,
+      "step": 1695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.761366156575737e-05,
+      "loss": 0.4095,
+      "step": 1696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7635839432246615e-05,
+      "loss": 0.4181,
+      "step": 1697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.765801729873586e-05,
+      "loss": 0.3923,
+      "step": 1698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.76801951652251e-05,
+      "loss": 0.4071,
+      "step": 1699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.770237303171435e-05,
+      "loss": 0.3629,
+      "step": 1700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.772455089820359e-05,
+      "loss": 0.3726,
+      "step": 1701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.774672876469283e-05,
+      "loss": 0.4071,
+      "step": 1702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.776890663118208e-05,
+      "loss": 0.4588,
+      "step": 1703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.779108449767132e-05,
+      "loss": 0.4002,
+      "step": 1704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.781326236416056e-05,
+      "loss": 0.4005,
+      "step": 1705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.783544023064981e-05,
+      "loss": 0.3989,
+      "step": 1706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.785761809713905e-05,
+      "loss": 0.4259,
+      "step": 1707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7879795963628295e-05,
+      "loss": 0.4222,
+      "step": 1708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.790197383011754e-05,
+      "loss": 0.3845,
+      "step": 1709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7924151696606785e-05,
+      "loss": 0.3776,
+      "step": 1710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.794632956309602e-05,
+      "loss": 0.3977,
+      "step": 1711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7968507429585275e-05,
+      "loss": 0.4083,
+      "step": 1712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.799068529607451e-05,
+      "loss": 0.4797,
+      "step": 1713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8012863162563765e-05,
+      "loss": 0.4347,
+      "step": 1714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8035041029053e-05,
+      "loss": 0.3962,
+      "step": 1715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.805721889554225e-05,
+      "loss": 0.3903,
+      "step": 1716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8079396762031486e-05,
+      "loss": 0.4064,
+      "step": 1717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.810157462852074e-05,
+      "loss": 0.3882,
+      "step": 1718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8123752495009976e-05,
+      "loss": 0.451,
+      "step": 1719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.814593036149923e-05,
+      "loss": 0.393,
+      "step": 1720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8168108227988465e-05,
+      "loss": 0.3728,
+      "step": 1721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8190286094477704e-05,
+      "loss": 0.4148,
+      "step": 1722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8212463960966955e-05,
+      "loss": 0.4238,
+      "step": 1723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.823464182745619e-05,
+      "loss": 0.3761,
+      "step": 1724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.825681969394544e-05,
+      "loss": 0.4618,
+      "step": 1725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8278997560434676e-05,
+      "loss": 0.423,
+      "step": 1726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.830117542692393e-05,
+      "loss": 0.3558,
+      "step": 1727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8323353293413166e-05,
+      "loss": 0.4218,
+      "step": 1728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.834553115990242e-05,
+      "loss": 0.4041,
+      "step": 1729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8367709026391656e-05,
+      "loss": 0.3781,
+      "step": 1730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.83898868928809e-05,
+      "loss": 0.415,
+      "step": 1731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8412064759370146e-05,
+      "loss": 0.403,
+      "step": 1732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.843424262585939e-05,
+      "loss": 0.3884,
+      "step": 1733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.845642049234863e-05,
+      "loss": 0.3753,
+      "step": 1734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.847859835883788e-05,
+      "loss": 0.4121,
+      "step": 1735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.850077622532712e-05,
+      "loss": 0.4087,
+      "step": 1736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8522954091816363e-05,
+      "loss": 0.3931,
+      "step": 1737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.854513195830561e-05,
+      "loss": 0.3754,
+      "step": 1738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.856730982479485e-05,
+      "loss": 0.4269,
+      "step": 1739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.858948769128409e-05,
+      "loss": 0.4311,
+      "step": 1740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.861166555777334e-05,
+      "loss": 0.3552,
+      "step": 1741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.863384342426258e-05,
+      "loss": 0.4274,
+      "step": 1742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8656021290751826e-05,
+      "loss": 0.3696,
+      "step": 1743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.867819915724107e-05,
+      "loss": 0.3827,
+      "step": 1744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8700377023730316e-05,
+      "loss": 0.3762,
+      "step": 1745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8722554890219554e-05,
+      "loss": 0.4189,
+      "step": 1746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8744732756708806e-05,
+      "loss": 0.403,
+      "step": 1747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8766910623198044e-05,
+      "loss": 0.3867,
+      "step": 1748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8789088489687295e-05,
+      "loss": 0.3899,
+      "step": 1749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8811266356176534e-05,
+      "loss": 0.3912,
+      "step": 1750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.883344422266578e-05,
+      "loss": 0.3556,
+      "step": 1751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.885562208915502e-05,
+      "loss": 0.3751,
+      "step": 1752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.887779995564427e-05,
+      "loss": 0.4112,
+      "step": 1753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8899977822133506e-05,
+      "loss": 0.3888,
+      "step": 1754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8922155688622745e-05,
+      "loss": 0.4123,
+      "step": 1755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8944333555111996e-05,
+      "loss": 0.4162,
+      "step": 1756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8966511421601234e-05,
+      "loss": 0.3882,
+      "step": 1757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8988689288090486e-05,
+      "loss": 0.4364,
+      "step": 1758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9010867154579724e-05,
+      "loss": 0.399,
+      "step": 1759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.903304502106897e-05,
+      "loss": 0.3847,
+      "step": 1760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.905522288755821e-05,
+      "loss": 0.3674,
+      "step": 1761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.907740075404746e-05,
+      "loss": 0.4205,
+      "step": 1762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.90995786205367e-05,
+      "loss": 0.3451,
+      "step": 1763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.912175648702595e-05,
+      "loss": 0.3971,
+      "step": 1764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.914393435351519e-05,
+      "loss": 0.4382,
+      "step": 1765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.916611222000443e-05,
+      "loss": 0.4342,
+      "step": 1766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9188290086493677e-05,
+      "loss": 0.417,
+      "step": 1767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.921046795298292e-05,
+      "loss": 0.373,
+      "step": 1768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.923264581947216e-05,
+      "loss": 0.4237,
+      "step": 1769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.925482368596141e-05,
+      "loss": 0.3947,
+      "step": 1770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.927700155245065e-05,
+      "loss": 0.3925,
+      "step": 1771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9299179418939894e-05,
+      "loss": 0.3933,
+      "step": 1772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.932135728542914e-05,
+      "loss": 0.3384,
+      "step": 1773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9343535151918384e-05,
+      "loss": 0.3536,
+      "step": 1774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.936571301840762e-05,
+      "loss": 0.3886,
+      "step": 1775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9387890884896874e-05,
+      "loss": 0.427,
+      "step": 1776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.941006875138611e-05,
+      "loss": 0.3928,
+      "step": 1777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.943224661787536e-05,
+      "loss": 0.4535,
+      "step": 1778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.94544244843646e-05,
+      "loss": 0.4283,
+      "step": 1779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.947660235085385e-05,
+      "loss": 0.3633,
+      "step": 1780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9498780217343085e-05,
+      "loss": 0.4453,
+      "step": 1781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9520958083832336e-05,
+      "loss": 0.4312,
+      "step": 1782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9543135950321575e-05,
+      "loss": 0.3782,
+      "step": 1783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9565313816810826e-05,
+      "loss": 0.3898,
+      "step": 1784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9587491683300064e-05,
+      "loss": 0.3369,
+      "step": 1785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.960966954978931e-05,
+      "loss": 0.3981,
+      "step": 1786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.963184741627855e-05,
+      "loss": 0.3934,
+      "step": 1787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.965402528276779e-05,
+      "loss": 0.4052,
+      "step": 1788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.967620314925704e-05,
+      "loss": 0.3825,
+      "step": 1789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9698381015746275e-05,
+      "loss": 0.4009,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.972055888223553e-05,
+      "loss": 0.4182,
+      "step": 1791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9742736748724765e-05,
+      "loss": 0.4119,
+      "step": 1792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.976491461521402e-05,
+      "loss": 0.3884,
+      "step": 1793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9787092481703255e-05,
+      "loss": 0.4003,
+      "step": 1794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.98092703481925e-05,
+      "loss": 0.4033,
+      "step": 1795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.983144821468174e-05,
+      "loss": 0.3746,
+      "step": 1796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.985362608117099e-05,
+      "loss": 0.4291,
+      "step": 1797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.987580394766023e-05,
+      "loss": 0.4223,
+      "step": 1798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.989798181414948e-05,
+      "loss": 0.4144,
+      "step": 1799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.992015968063872e-05,
+      "loss": 0.3825,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.994233754712796e-05,
+      "loss": 0.4351,
+      "step": 1801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.996451541361721e-05,
+      "loss": 0.4111,
+      "step": 1802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998669328010645e-05,
+      "loss": 0.3834,
+      "step": 1803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.000887114659569e-05,
+      "loss": 0.369,
+      "step": 1804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.003104901308494e-05,
+      "loss": 0.4253,
+      "step": 1805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.005322687957418e-05,
+      "loss": 0.4218,
+      "step": 1806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0075404746063425e-05,
+      "loss": 0.3667,
+      "step": 1807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.009758261255267e-05,
+      "loss": 0.3556,
+      "step": 1808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0119760479041915e-05,
+      "loss": 0.4055,
+      "step": 1809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.014193834553115e-05,
+      "loss": 0.4141,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0164116212020405e-05,
+      "loss": 0.3824,
+      "step": 1811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.018629407850964e-05,
+      "loss": 0.3872,
+      "step": 1812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.020847194499889e-05,
+      "loss": 0.4362,
+      "step": 1813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.023064981148813e-05,
+      "loss": 0.3926,
+      "step": 1814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.025282767797738e-05,
+      "loss": 0.3712,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0275005544466616e-05,
+      "loss": 0.4297,
+      "step": 1816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.029718341095587e-05,
+      "loss": 0.3962,
+      "step": 1817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0319361277445105e-05,
+      "loss": 0.4008,
+      "step": 1818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.034153914393436e-05,
+      "loss": 0.3959,
+      "step": 1819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0363717010423595e-05,
+      "loss": 0.3883,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.038589487691284e-05,
+      "loss": 0.3668,
+      "step": 1821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.040807274340208e-05,
+      "loss": 0.401,
+      "step": 1822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.043025060989132e-05,
+      "loss": 0.3463,
+      "step": 1823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.045242847638057e-05,
+      "loss": 0.4431,
+      "step": 1824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0474606342869806e-05,
+      "loss": 0.3956,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.049678420935906e-05,
+      "loss": 0.3971,
+      "step": 1826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0518962075848296e-05,
+      "loss": 0.4736,
+      "step": 1827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.054113994233755e-05,
+      "loss": 0.3965,
+      "step": 1828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0563317808826786e-05,
+      "loss": 0.3747,
+      "step": 1829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.058549567531603e-05,
+      "loss": 0.4436,
+      "step": 1830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.060767354180527e-05,
+      "loss": 0.3943,
+      "step": 1831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.062985140829452e-05,
+      "loss": 0.3456,
+      "step": 1832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.065202927478376e-05,
+      "loss": 0.3932,
+      "step": 1833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.067420714127301e-05,
+      "loss": 0.4356,
+      "step": 1834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.069638500776225e-05,
+      "loss": 0.4241,
+      "step": 1835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.071856287425149e-05,
+      "loss": 0.4525,
+      "step": 1836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 0.3909,
+      "step": 1837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.076291860722998e-05,
+      "loss": 0.3942,
+      "step": 1838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.078509647371922e-05,
+      "loss": 0.3801,
+      "step": 1839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.080727434020847e-05,
+      "loss": 0.384,
+      "step": 1840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.082945220669771e-05,
+      "loss": 0.4074,
+      "step": 1841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0851630073186956e-05,
+      "loss": 0.4199,
+      "step": 1842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.08738079396762e-05,
+      "loss": 0.4283,
+      "step": 1843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0895985806165446e-05,
+      "loss": 0.4204,
+      "step": 1844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0918163672654684e-05,
+      "loss": 0.4278,
+      "step": 1845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0940341539143935e-05,
+      "loss": 0.3826,
+      "step": 1846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0962519405633174e-05,
+      "loss": 0.3731,
+      "step": 1847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.098469727212242e-05,
+      "loss": 0.4138,
+      "step": 1848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1006875138611663e-05,
+      "loss": 0.4109,
+      "step": 1849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.102905300510091e-05,
+      "loss": 0.3729,
+      "step": 1850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1051230871590146e-05,
+      "loss": 0.4091,
+      "step": 1851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.10734087380794e-05,
+      "loss": 0.4719,
+      "step": 1852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1095586604568636e-05,
+      "loss": 0.4278,
+      "step": 1853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.111776447105789e-05,
+      "loss": 0.3637,
+      "step": 1854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1139942337547126e-05,
+      "loss": 0.4251,
+      "step": 1855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1162120204036364e-05,
+      "loss": 0.3653,
+      "step": 1856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.118429807052561e-05,
+      "loss": 0.3686,
+      "step": 1857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1206475937014854e-05,
+      "loss": 0.3871,
+      "step": 1858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.12286538035041e-05,
+      "loss": 0.4227,
+      "step": 1859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.125083166999334e-05,
+      "loss": 0.3947,
+      "step": 1860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.127300953648259e-05,
+      "loss": 0.4185,
+      "step": 1861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.129518740297183e-05,
+      "loss": 0.4266,
+      "step": 1862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.131736526946108e-05,
+      "loss": 0.4011,
+      "step": 1863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1339543135950317e-05,
+      "loss": 0.394,
+      "step": 1864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.136172100243956e-05,
+      "loss": 0.4367,
+      "step": 1865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.13838988689288e-05,
+      "loss": 0.3991,
+      "step": 1866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.140607673541805e-05,
+      "loss": 0.4299,
+      "step": 1867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.142825460190729e-05,
+      "loss": 0.4284,
+      "step": 1868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.145043246839654e-05,
+      "loss": 0.3621,
+      "step": 1869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.147261033488578e-05,
+      "loss": 0.3751,
+      "step": 1870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1494788201375024e-05,
+      "loss": 0.4406,
+      "step": 1871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.151696606786427e-05,
+      "loss": 0.4376,
+      "step": 1872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1539143934353514e-05,
+      "loss": 0.3824,
+      "step": 1873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.156132180084275e-05,
+      "loss": 0.3951,
+      "step": 1874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1583499667332004e-05,
+      "loss": 0.39,
+      "step": 1875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.160567753382124e-05,
+      "loss": 0.4368,
+      "step": 1876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.162785540031049e-05,
+      "loss": 0.4374,
+      "step": 1877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.165003326679973e-05,
+      "loss": 0.3617,
+      "step": 1878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1672211133288977e-05,
+      "loss": 0.3956,
+      "step": 1879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1694388999778215e-05,
+      "loss": 0.4256,
+      "step": 1880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1716566866267466e-05,
+      "loss": 0.4083,
+      "step": 1881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1738744732756704e-05,
+      "loss": 0.4335,
+      "step": 1882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.176092259924595e-05,
+      "loss": 0.4233,
+      "step": 1883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1783100465735194e-05,
+      "loss": 0.3685,
+      "step": 1884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.180527833222444e-05,
+      "loss": 0.3543,
+      "step": 1885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.182745619871368e-05,
+      "loss": 0.3965,
+      "step": 1886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.184963406520293e-05,
+      "loss": 0.406,
+      "step": 1887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.187181193169217e-05,
+      "loss": 0.3949,
+      "step": 1888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.189398979818142e-05,
+      "loss": 0.4033,
+      "step": 1889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.191616766467066e-05,
+      "loss": 0.3834,
+      "step": 1890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1938345531159895e-05,
+      "loss": 0.4084,
+      "step": 1891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.196052339764914e-05,
+      "loss": 0.3754,
+      "step": 1892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1982701264138385e-05,
+      "loss": 0.3534,
+      "step": 1893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.200487913062763e-05,
+      "loss": 0.3996,
+      "step": 1894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.202705699711687e-05,
+      "loss": 0.4253,
+      "step": 1895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.204923486360612e-05,
+      "loss": 0.3882,
+      "step": 1896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.207141273009536e-05,
+      "loss": 0.4236,
+      "step": 1897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.209359059658461e-05,
+      "loss": 0.3745,
+      "step": 1898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.211576846307385e-05,
+      "loss": 0.4207,
+      "step": 1899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.213794632956309e-05,
+      "loss": 0.3907,
+      "step": 1900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.216012419605233e-05,
+      "loss": 0.4058,
+      "step": 1901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.218230206254158e-05,
+      "loss": 0.414,
+      "step": 1902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.220447992903082e-05,
+      "loss": 0.4077,
+      "step": 1903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.222665779552007e-05,
+      "loss": 0.3873,
+      "step": 1904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.224883566200931e-05,
+      "loss": 0.3937,
+      "step": 1905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2271013528498555e-05,
+      "loss": 0.4462,
+      "step": 1906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.22931913949878e-05,
+      "loss": 0.4381,
+      "step": 1907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2315369261477045e-05,
+      "loss": 0.376,
+      "step": 1908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.233754712796628e-05,
+      "loss": 0.3852,
+      "step": 1909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2359724994455534e-05,
+      "loss": 0.3951,
+      "step": 1910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.238190286094477e-05,
+      "loss": 0.4097,
+      "step": 1911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.240408072743402e-05,
+      "loss": 0.4309,
+      "step": 1912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.242625859392326e-05,
+      "loss": 0.3872,
+      "step": 1913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.244843646041251e-05,
+      "loss": 0.3953,
+      "step": 1914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2470614326901745e-05,
+      "loss": 0.3926,
+      "step": 1915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2492792193391e-05,
+      "loss": 0.3842,
+      "step": 1916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2514970059880235e-05,
+      "loss": 0.3788,
+      "step": 1917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.253714792636948e-05,
+      "loss": 0.4082,
+      "step": 1918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2559325792858725e-05,
+      "loss": 0.401,
+      "step": 1919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.258150365934797e-05,
+      "loss": 0.3869,
+      "step": 1920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.260368152583721e-05,
+      "loss": 0.4168,
+      "step": 1921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.262585939232646e-05,
+      "loss": 0.3646,
+      "step": 1922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.26480372588157e-05,
+      "loss": 0.3936,
+      "step": 1923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2670215125304936e-05,
+      "loss": 0.3798,
+      "step": 1924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.269239299179419e-05,
+      "loss": 0.423,
+      "step": 1925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2714570858283426e-05,
+      "loss": 0.3878,
+      "step": 1926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.273674872477267e-05,
+      "loss": 0.4194,
+      "step": 1927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2758926591261916e-05,
+      "loss": 0.4067,
+      "step": 1928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278110445775116e-05,
+      "loss": 0.3695,
+      "step": 1929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28032823242404e-05,
+      "loss": 0.4159,
+      "step": 1930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282546019072965e-05,
+      "loss": 0.4267,
+      "step": 1931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284763805721889e-05,
+      "loss": 0.4364,
+      "step": 1932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286981592370814e-05,
+      "loss": 0.4049,
+      "step": 1933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289199379019738e-05,
+      "loss": 0.3884,
+      "step": 1934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291417165668662e-05,
+      "loss": 0.4025,
+      "step": 1935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293634952317586e-05,
+      "loss": 0.4094,
+      "step": 1936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.295852738966511e-05,
+      "loss": 0.3809,
+      "step": 1937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.298070525615435e-05,
+      "loss": 0.3974,
+      "step": 1938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.30028831226436e-05,
+      "loss": 0.3995,
+      "step": 1939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.302506098913284e-05,
+      "loss": 0.3917,
+      "step": 1940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3047238855622086e-05,
+      "loss": 0.353,
+      "step": 1941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.306941672211133e-05,
+      "loss": 0.3952,
+      "step": 1942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3091594588600575e-05,
+      "loss": 0.359,
+      "step": 1943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3113772455089814e-05,
+      "loss": 0.4387,
+      "step": 1944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3135950321579065e-05,
+      "loss": 0.4328,
+      "step": 1945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3158128188068303e-05,
+      "loss": 0.4101,
+      "step": 1946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.318030605455755e-05,
+      "loss": 0.3627,
+      "step": 1947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.320248392104679e-05,
+      "loss": 0.3704,
+      "step": 1948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.322466178753604e-05,
+      "loss": 0.3734,
+      "step": 1949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3246839654025276e-05,
+      "loss": 0.3577,
+      "step": 1950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.326901752051453e-05,
+      "loss": 0.4033,
+      "step": 1951
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3291195387003766e-05,
+      "loss": 0.3811,
+      "step": 1952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.331337325349301e-05,
+      "loss": 0.4206,
+      "step": 1953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3335551119982256e-05,
+      "loss": 0.3746,
+      "step": 1954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.33577289864715e-05,
+      "loss": 0.3957,
+      "step": 1955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.337990685296074e-05,
+      "loss": 0.4022,
+      "step": 1956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3402084719449984e-05,
+      "loss": 0.4199,
+      "step": 1957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.342426258593923e-05,
+      "loss": 0.3617,
+      "step": 1958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.344644045242847e-05,
+      "loss": 0.3435,
+      "step": 1959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.346861831891772e-05,
+      "loss": 0.4322,
+      "step": 1960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3490796185406957e-05,
+      "loss": 0.4005,
+      "step": 1961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.35129740518962e-05,
+      "loss": 0.4151,
+      "step": 1962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3535151918385446e-05,
+      "loss": 0.3781,
+      "step": 1963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.355732978487469e-05,
+      "loss": 0.4005,
+      "step": 1964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.357950765136393e-05,
+      "loss": 0.4172,
+      "step": 1965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.360168551785318e-05,
+      "loss": 0.4104,
+      "step": 1966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.362386338434242e-05,
+      "loss": 0.4233,
+      "step": 1967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.364604125083167e-05,
+      "loss": 0.3864,
+      "step": 1968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.366821911732091e-05,
+      "loss": 0.4639,
+      "step": 1969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3690396983810154e-05,
+      "loss": 0.3664,
+      "step": 1970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.371257485029939e-05,
+      "loss": 0.3809,
+      "step": 1971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3734752716788644e-05,
+      "loss": 0.4005,
+      "step": 1972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.375693058327788e-05,
+      "loss": 0.3829,
+      "step": 1973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3779108449767133e-05,
+      "loss": 0.3817,
+      "step": 1974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.380128631625637e-05,
+      "loss": 0.4017,
+      "step": 1975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3823464182745617e-05,
+      "loss": 0.3718,
+      "step": 1976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.384564204923486e-05,
+      "loss": 0.3781,
+      "step": 1977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3867819915724106e-05,
+      "loss": 0.4075,
+      "step": 1978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3889997782213344e-05,
+      "loss": 0.3991,
+      "step": 1979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3912175648702596e-05,
+      "loss": 0.3755,
+      "step": 1980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3934353515191834e-05,
+      "loss": 0.4144,
+      "step": 1981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.395653138168108e-05,
+      "loss": 0.431,
+      "step": 1982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3978709248170324e-05,
+      "loss": 0.4244,
+      "step": 1983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.400088711465957e-05,
+      "loss": 0.366,
+      "step": 1984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.402306498114881e-05,
+      "loss": 0.3946,
+      "step": 1985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.404524284763806e-05,
+      "loss": 0.3748,
+      "step": 1986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.40674207141273e-05,
+      "loss": 0.3937,
+      "step": 1987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.408959858061654e-05,
+      "loss": 0.3682,
+      "step": 1988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.411177644710579e-05,
+      "loss": 0.4439,
+      "step": 1989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.413395431359503e-05,
+      "loss": 0.3998,
+      "step": 1990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.415613218008427e-05,
+      "loss": 0.3996,
+      "step": 1991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4178310046573515e-05,
+      "loss": 0.3753,
+      "step": 1992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.420048791306276e-05,
+      "loss": 0.3656,
+      "step": 1993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4222665779552e-05,
+      "loss": 0.4083,
+      "step": 1994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.424484364604125e-05,
+      "loss": 0.4128,
+      "step": 1995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.426702151253049e-05,
+      "loss": 0.3982,
+      "step": 1996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.428919937901973e-05,
+      "loss": 0.3916,
+      "step": 1997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.431137724550898e-05,
+      "loss": 0.4461,
+      "step": 1998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.433355511199822e-05,
+      "loss": 0.4077,
+      "step": 1999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.435573297848746e-05,
+      "loss": 0.3486,
+      "step": 2000
+    }
+  ],
+  "max_steps": 450900,
+  "num_train_epochs": 100,
+  "total_flos": 165307148697600.0,
+  "trial_name": null,
+  "trial_params": null
+}