diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,180016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0435223698894185,
+  "global_step": 30000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.38123415046492e-09,
+      "loss": 1.044,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.76246830092984e-09,
+      "loss": 1.2473,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.014370245139476e-08,
+      "loss": 1.2359,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.352493660185968e-08,
+      "loss": 1.1636,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.69061707523246e-08,
+      "loss": 1.0143,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.028740490278952e-08,
+      "loss": 1.2954,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.366863905325444e-08,
+      "loss": 1.3627,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.704987320371936e-08,
+      "loss": 1.0517,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.043110735418428e-08,
+      "loss": 1.3822,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.38123415046492e-08,
+      "loss": 1.1037,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.719357565511412e-08,
+      "loss": 1.2309,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.057480980557904e-08,
+      "loss": 0.9952,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.395604395604396e-08,
+      "loss": 1.1565,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.733727810650888e-08,
+      "loss": 1.146,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.0718512256973804e-08,
+      "loss": 1.1043,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.409974640743872e-08,
+      "loss": 1.0942,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.748098055790364e-08,
+      "loss": 1.1427,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.086221470836856e-08,
+      "loss": 0.9898,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.424344885883348e-08,
+      "loss": 1.3633,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.76246830092984e-08,
+      "loss": 0.9935,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.100591715976332e-08,
+      "loss": 1.1155,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.438715131022823e-08,
+      "loss": 1.0959,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.776838546069315e-08,
+      "loss": 1.1023,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.114961961115808e-08,
+      "loss": 0.8636,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.4530853761623e-08,
+      "loss": 1.3283,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.791208791208792e-08,
+      "loss": 0.9993,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.129332206255283e-08,
+      "loss": 0.9939,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.467455621301776e-08,
+      "loss": 1.0058,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.805579036348268e-08,
+      "loss": 0.9819,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0143702451394761e-07,
+      "loss": 1.0265,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0481825866441251e-07,
+      "loss": 1.0261,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0819949281487745e-07,
+      "loss": 1.0988,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1158072696534235e-07,
+      "loss": 0.9422,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1496196111580728e-07,
+      "loss": 1.0464,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.183431952662722e-07,
+      "loss": 0.8681,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2172442941673712e-07,
+      "loss": 1.0988,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2510566356720204e-07,
+      "loss": 0.871,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2848689771766696e-07,
+      "loss": 0.9233,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3186813186813187e-07,
+      "loss": 1.003,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.352493660185968e-07,
+      "loss": 0.8247,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.386306001690617e-07,
+      "loss": 0.9222,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4201183431952663e-07,
+      "loss": 0.9755,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4539306846999155e-07,
+      "loss": 1.0114,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4877430262045647e-07,
+      "loss": 1.0326,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.521555367709214e-07,
+      "loss": 0.8907,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.555367709213863e-07,
+      "loss": 0.925,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5891800507185122e-07,
+      "loss": 0.9279,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6229923922231617e-07,
+      "loss": 0.9212,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6568047337278109e-07,
+      "loss": 0.8307,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.69061707523246e-07,
+      "loss": 1.0452,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.724429416737109e-07,
+      "loss": 0.9842,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7582417582417584e-07,
+      "loss": 0.8882,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7920540997464076e-07,
+      "loss": 0.8185,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8258664412510565e-07,
+      "loss": 1.0168,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.859678782755706e-07,
+      "loss": 0.8381,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8934911242603552e-07,
+      "loss": 0.7868,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9273034657650046e-07,
+      "loss": 0.9571,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9611158072696535e-07,
+      "loss": 0.9184,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9949281487743027e-07,
+      "loss": 0.9219,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0287404902789522e-07,
+      "loss": 0.8997,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0625528317836014e-07,
+      "loss": 0.8905,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0963651732882503e-07,
+      "loss": 0.9077,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1301775147928995e-07,
+      "loss": 0.9286,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.163989856297549e-07,
+      "loss": 1.0074,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.197802197802198e-07,
+      "loss": 0.9705,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.231614539306847e-07,
+      "loss": 1.0229,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2654268808114965e-07,
+      "loss": 0.7871,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2992392223161456e-07,
+      "loss": 0.8199,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3330515638207946e-07,
+      "loss": 0.8493,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.366863905325444e-07,
+      "loss": 0.812,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.400676246830093e-07,
+      "loss": 1.0022,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4344885883347424e-07,
+      "loss": 0.9493,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4683009298393916e-07,
+      "loss": 0.8035,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.502113271344041e-07,
+      "loss": 1.0423,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.53592561284869e-07,
+      "loss": 0.9456,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.569737954353339e-07,
+      "loss": 0.9312,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6035502958579883e-07,
+      "loss": 0.7828,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6373626373626375e-07,
+      "loss": 0.7983,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6711749788672867e-07,
+      "loss": 0.8176,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.704987320371936e-07,
+      "loss": 0.8656,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.738799661876585e-07,
+      "loss": 0.8289,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.772612003381234e-07,
+      "loss": 0.9372,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.806424344885884e-07,
+      "loss": 0.9458,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8402366863905326e-07,
+      "loss": 0.9233,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.874049027895182e-07,
+      "loss": 0.7682,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.907861369399831e-07,
+      "loss": 0.8549,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9416737109044807e-07,
+      "loss": 0.9339,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9754860524091294e-07,
+      "loss": 0.8467,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0092983939137785e-07,
+      "loss": 0.8848,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.043110735418428e-07,
+      "loss": 0.9205,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-07,
+      "loss": 0.9248,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.110735418427726e-07,
+      "loss": 0.8811,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1445477599323753e-07,
+      "loss": 0.866,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1783601014370245e-07,
+      "loss": 0.9805,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2121724429416736e-07,
+      "loss": 0.8497,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2459847844463234e-07,
+      "loss": 0.8482,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2797971259509725e-07,
+      "loss": 0.73,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3136094674556217e-07,
+      "loss": 0.8424,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.347421808960271e-07,
+      "loss": 0.8124,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.38123415046492e-07,
+      "loss": 0.8997,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.415046491969569e-07,
+      "loss": 0.8663,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.448858833474218e-07,
+      "loss": 0.7591,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4826711749788677e-07,
+      "loss": 0.8234,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.516483516483517e-07,
+      "loss": 0.8597,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.550295857988166e-07,
+      "loss": 0.7928,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.584108199492815e-07,
+      "loss": 0.9301,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6179205409974644e-07,
+      "loss": 0.7374,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.651732882502113e-07,
+      "loss": 0.8473,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.685545224006763e-07,
+      "loss": 0.8838,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.719357565511412e-07,
+      "loss": 0.8113,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.753169907016061e-07,
+      "loss": 0.819,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7869822485207103e-07,
+      "loss": 0.7277,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8207945900253595e-07,
+      "loss": 0.8844,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.854606931530009e-07,
+      "loss": 0.7975,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8884192730346584e-07,
+      "loss": 0.789,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.922231614539307e-07,
+      "loss": 0.8824,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.956043956043956e-07,
+      "loss": 0.7306,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9898562975486054e-07,
+      "loss": 0.7175,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0236686390532546e-07,
+      "loss": 0.7888,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0574809805579043e-07,
+      "loss": 0.8191,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0912933220625535e-07,
+      "loss": 0.8063,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1251056635672027e-07,
+      "loss": 1.0446,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1589180050718514e-07,
+      "loss": 0.7761,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1927303465765005e-07,
+      "loss": 0.811,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2265426880811497e-07,
+      "loss": 0.6968,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.260355029585799e-07,
+      "loss": 0.9311,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2941673710904486e-07,
+      "loss": 0.846,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.327979712595098e-07,
+      "loss": 0.7689,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.361792054099747e-07,
+      "loss": 0.8164,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.395604395604396e-07,
+      "loss": 0.8469,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.429416737109045e-07,
+      "loss": 0.7571,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.463229078613694e-07,
+      "loss": 0.7445,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.497041420118344e-07,
+      "loss": 0.8573,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.530853761622993e-07,
+      "loss": 0.881,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.564666103127642e-07,
+      "loss": 0.6912,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5984784446322913e-07,
+      "loss": 0.6837,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6322907861369405e-07,
+      "loss": 0.8155,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.666103127641589e-07,
+      "loss": 0.8009,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6999154691462383e-07,
+      "loss": 0.7532,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.733727810650888e-07,
+      "loss": 0.7315,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.767540152155537e-07,
+      "loss": 0.7631,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.801352493660186e-07,
+      "loss": 0.8836,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.835164835164836e-07,
+      "loss": 0.817,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.868977176669485e-07,
+      "loss": 0.949,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.902789518174133e-07,
+      "loss": 0.6764,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.936601859678783e-07,
+      "loss": 0.7529,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.970414201183432e-07,
+      "loss": 0.7067,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.004226542688082e-07,
+      "loss": 0.7636,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.038038884192731e-07,
+      "loss": 0.7391,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.07185122569738e-07,
+      "loss": 0.8519,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.10566356720203e-07,
+      "loss": 0.7202,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.139475908706678e-07,
+      "loss": 0.827,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.173288250211327e-07,
+      "loss": 0.8069,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.207100591715977e-07,
+      "loss": 0.8124,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.240912933220626e-07,
+      "loss": 0.7293,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.274725274725275e-07,
+      "loss": 0.749,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.308537616229925e-07,
+      "loss": 0.7903,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.342349957734573e-07,
+      "loss": 0.7935,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.376162299239223e-07,
+      "loss": 0.8065,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.409974640743872e-07,
+      "loss": 0.8539,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.44378698224852e-07,
+      "loss": 0.8677,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.47759932375317e-07,
+      "loss": 0.9651,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.51141166525782e-07,
+      "loss": 0.6743,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.545224006762468e-07,
+      "loss": 0.7376,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.579036348267118e-07,
+      "loss": 0.7932,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.612848689771768e-07,
+      "loss": 0.7048,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.646661031276417e-07,
+      "loss": 0.6668,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.680473372781065e-07,
+      "loss": 0.8138,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 0.8806,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.748098055790364e-07,
+      "loss": 0.7511,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.781910397295013e-07,
+      "loss": 0.6699,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.815722738799662e-07,
+      "loss": 0.8107,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.849535080304312e-07,
+      "loss": 0.8717,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.883347421808961e-07,
+      "loss": 0.7527,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.91715976331361e-07,
+      "loss": 0.8623,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.950972104818259e-07,
+      "loss": 0.7941,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.984784446322908e-07,
+      "loss": 0.7133,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.018596787827557e-07,
+      "loss": 0.7765,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.052409129332207e-07,
+      "loss": 0.7338,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.086221470836857e-07,
+      "loss": 0.7611,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.120033812341505e-07,
+      "loss": 0.7876,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.153846153846155e-07,
+      "loss": 0.7538,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.187658495350804e-07,
+      "loss": 0.8053,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.221470836855452e-07,
+      "loss": 0.7018,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.255283178360103e-07,
+      "loss": 0.7745,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.289095519864751e-07,
+      "loss": 0.7728,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3229078613694e-07,
+      "loss": 0.7238,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.356720202874049e-07,
+      "loss": 0.7579,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.390532544378699e-07,
+      "loss": 0.7371,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.424344885883347e-07,
+      "loss": 0.7594,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.458157227387997e-07,
+      "loss": 0.7023,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.491969568892647e-07,
+      "loss": 0.9141,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.525781910397295e-07,
+      "loss": 0.7301,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.559594251901945e-07,
+      "loss": 0.7573,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.593406593406594e-07,
+      "loss": 0.7944,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.627218934911243e-07,
+      "loss": 0.784,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.661031276415893e-07,
+      "loss": 0.6731,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.694843617920542e-07,
+      "loss": 0.7188,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.728655959425192e-07,
+      "loss": 0.744,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.76246830092984e-07,
+      "loss": 0.6768,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.796280642434489e-07,
+      "loss": 0.7363,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.830092983939138e-07,
+      "loss": 0.7512,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.863905325443787e-07,
+      "loss": 0.7507,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.897717666948436e-07,
+      "loss": 0.7687,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.931530008453086e-07,
+      "loss": 0.7502,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.965342349957735e-07,
+      "loss": 0.7728,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.999154691462384e-07,
+      "loss": 0.7359,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.032967032967034e-07,
+      "loss": 0.8368,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.066779374471682e-07,
+      "loss": 0.901,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.100591715976332e-07,
+      "loss": 0.797,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.134404057480982e-07,
+      "loss": 0.7122,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.16821639898563e-07,
+      "loss": 0.7132,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.20202874049028e-07,
+      "loss": 0.7705,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.235841081994929e-07,
+      "loss": 0.8024,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.269653423499579e-07,
+      "loss": 0.8106,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.303465765004226e-07,
+      "loss": 0.7081,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.337278106508876e-07,
+      "loss": 0.7358,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.371090448013526e-07,
+      "loss": 0.8148,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.404902789518174e-07,
+      "loss": 0.7848,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.438715131022824e-07,
+      "loss": 0.7407,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.472527472527473e-07,
+      "loss": 0.7772,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.506339814032122e-07,
+      "loss": 0.8187,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.540152155536772e-07,
+      "loss": 0.7302,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.573964497041421e-07,
+      "loss": 0.686,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.60777683854607e-07,
+      "loss": 0.7539,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.641589180050719e-07,
+      "loss": 0.7396,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.675401521555369e-07,
+      "loss": 0.7618,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.709213863060018e-07,
+      "loss": 0.6785,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.743026204564667e-07,
+      "loss": 0.7703,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.776838546069317e-07,
+      "loss": 0.8366,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.810650887573964e-07,
+      "loss": 0.7019,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.844463229078614e-07,
+      "loss": 0.7979,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.878275570583263e-07,
+      "loss": 0.7725,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.912087912087912e-07,
+      "loss": 0.7011,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.945900253592561e-07,
+      "loss": 0.7608,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.979712595097211e-07,
+      "loss": 0.8122,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.013524936601861e-07,
+      "loss": 0.6507,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.047337278106509e-07,
+      "loss": 0.7746,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.081149619611159e-07,
+      "loss": 0.8493,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.114961961115809e-07,
+      "loss": 0.881,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.148774302620457e-07,
+      "loss": 0.7417,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.182586644125107e-07,
+      "loss": 0.7637,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.216398985629756e-07,
+      "loss": 0.7517,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.250211327134405e-07,
+      "loss": 0.8223,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.284023668639055e-07,
+      "loss": 0.6736,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.317836010143703e-07,
+      "loss": 0.6762,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.351648351648351e-07,
+      "loss": 0.7017,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.385460693153001e-07,
+      "loss": 0.7404,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.419273034657651e-07,
+      "loss": 0.6485,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.453085376162299e-07,
+      "loss": 0.6642,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.486897717666949e-07,
+      "loss": 0.6952,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.520710059171598e-07,
+      "loss": 0.8184,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.554522400676248e-07,
+      "loss": 0.6438,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.588334742180897e-07,
+      "loss": 0.8623,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.622147083685546e-07,
+      "loss": 0.8131,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.655959425190196e-07,
+      "loss": 0.6798,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.689771766694844e-07,
+      "loss": 0.7557,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.723584108199494e-07,
+      "loss": 0.8157,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.757396449704144e-07,
+      "loss": 0.6925,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.791208791208792e-07,
+      "loss": 0.7528,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.825021132713441e-07,
+      "loss": 0.7218,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.85883347421809e-07,
+      "loss": 0.677,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.892645815722739e-07,
+      "loss": 0.7789,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.926458157227388e-07,
+      "loss": 0.7843,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.960270498732038e-07,
+      "loss": 0.875,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.994082840236687e-07,
+      "loss": 0.6675,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.027895181741336e-07,
+      "loss": 0.6352,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.061707523245986e-07,
+      "loss": 0.7372,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.095519864750635e-07,
+      "loss": 0.7912,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.129332206255284e-07,
+      "loss": 0.7193,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.163144547759934e-07,
+      "loss": 0.7665,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.196956889264583e-07,
+      "loss": 0.7008,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.230769230769232e-07,
+      "loss": 0.7775,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.264581572273881e-07,
+      "loss": 0.675,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.29839391377853e-07,
+      "loss": 0.7113,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.332206255283178e-07,
+      "loss": 0.6505,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.366018596787828e-07,
+      "loss": 0.705,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.399830938292477e-07,
+      "loss": 0.6683,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.433643279797126e-07,
+      "loss": 0.7486,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.467455621301776e-07,
+      "loss": 0.7812,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.501267962806425e-07,
+      "loss": 0.7309,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.535080304311074e-07,
+      "loss": 0.7861,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.568892645815723e-07,
+      "loss": 0.7443,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.602704987320373e-07,
+      "loss": 0.6916,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.636517328825023e-07,
+      "loss": 0.6741,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.670329670329672e-07,
+      "loss": 0.694,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.70414201183432e-07,
+      "loss": 0.7254,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.73795435333897e-07,
+      "loss": 0.7819,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.77176669484362e-07,
+      "loss": 0.823,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.805579036348267e-07,
+      "loss": 0.648,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.839391377852917e-07,
+      "loss": 0.7669,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.873203719357566e-07,
+      "loss": 0.7399,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.907016060862216e-07,
+      "loss": 0.8318,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.940828402366864e-07,
+      "loss": 0.7186,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974640743871513e-07,
+      "loss": 0.6977,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0008453085376163e-06,
+      "loss": 0.7007,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0042265426880813e-06,
+      "loss": 0.6944,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0076077768385462e-06,
+      "loss": 0.7489,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.010989010989011e-06,
+      "loss": 0.6206,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.014370245139476e-06,
+      "loss": 0.6257,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.017751479289941e-06,
+      "loss": 0.7564,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.021132713440406e-06,
+      "loss": 0.7954,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0245139475908709e-06,
+      "loss": 0.8077,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0278951817413357e-06,
+      "loss": 0.7415,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0312764158918006e-06,
+      "loss": 0.7725,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0346576500422654e-06,
+      "loss": 0.7015,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0380388841927304e-06,
+      "loss": 0.725,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0414201183431953e-06,
+      "loss": 0.7751,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0448013524936603e-06,
+      "loss": 0.6168,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0481825866441253e-06,
+      "loss": 0.7957,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.05156382079459e-06,
+      "loss": 0.6805,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.054945054945055e-06,
+      "loss": 0.8246,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.05832628909552e-06,
+      "loss": 0.7255,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.061707523245985e-06,
+      "loss": 0.8323,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.06508875739645e-06,
+      "loss": 0.7925,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0684699915469147e-06,
+      "loss": 0.7194,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0718512256973796e-06,
+      "loss": 0.7941,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0752324598478446e-06,
+      "loss": 0.7818,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0786136939983096e-06,
+      "loss": 0.6823,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0819949281487743e-06,
+      "loss": 0.7141,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0853761622992393e-06,
+      "loss": 0.6917,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.088757396449704e-06,
+      "loss": 0.6631,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.092138630600169e-06,
+      "loss": 0.7529,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.095519864750634e-06,
+      "loss": 0.6815,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.098901098901099e-06,
+      "loss": 0.7501,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.102282333051564e-06,
+      "loss": 0.7967,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.105663567202029e-06,
+      "loss": 0.8273,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1090448013524937e-06,
+      "loss": 0.7219,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1124260355029587e-06,
+      "loss": 0.7574,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1158072696534236e-06,
+      "loss": 0.7088,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1191885038038886e-06,
+      "loss": 0.8241,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1225697379543536e-06,
+      "loss": 0.7538,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1259509721048183e-06,
+      "loss": 0.7367,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1293322062552833e-06,
+      "loss": 0.7138,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.132713440405748e-06,
+      "loss": 0.7495,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.136094674556213e-06,
+      "loss": 0.6726,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.139475908706678e-06,
+      "loss": 0.865,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 0.6507,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1462383770076077e-06,
+      "loss": 0.7574,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1496196111580727e-06,
+      "loss": 0.759,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1530008453085377e-06,
+      "loss": 0.7111,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1563820794590027e-06,
+      "loss": 0.5966,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1597633136094676e-06,
+      "loss": 0.7498,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1631445477599324e-06,
+      "loss": 0.7162,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1665257819103974e-06,
+      "loss": 0.6735,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1699070160608623e-06,
+      "loss": 0.7362,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1732882502113273e-06,
+      "loss": 0.6096,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1766694843617923e-06,
+      "loss": 0.703,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.180050718512257e-06,
+      "loss": 0.7395,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.183431952662722e-06,
+      "loss": 0.7608,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1868131868131868e-06,
+      "loss": 0.7648,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1901944209636517e-06,
+      "loss": 0.7354,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1935756551141167e-06,
+      "loss": 0.6375,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1969568892645817e-06,
+      "loss": 0.6231,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2003381234150467e-06,
+      "loss": 0.8056,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2037193575655114e-06,
+      "loss": 0.6971,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2071005917159764e-06,
+      "loss": 0.7238,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2104818258664414e-06,
+      "loss": 0.6676,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2138630600169063e-06,
+      "loss": 0.7232,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2172442941673713e-06,
+      "loss": 0.7646,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.220625528317836e-06,
+      "loss": 0.7065,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.224006762468301e-06,
+      "loss": 0.7776,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.227387996618766e-06,
+      "loss": 0.6491,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.230769230769231e-06,
+      "loss": 0.7202,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2341504649196957e-06,
+      "loss": 0.6629,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2375316990701607e-06,
+      "loss": 0.7622,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2409129332206257e-06,
+      "loss": 0.7476,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2442941673710904e-06,
+      "loss": 0.7644,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2476754015215554e-06,
+      "loss": 0.7107,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2510566356720206e-06,
+      "loss": 0.6292,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2544378698224854e-06,
+      "loss": 0.7471,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2578191039729501e-06,
+      "loss": 0.7732,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.261200338123415e-06,
+      "loss": 0.6406,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.26458157227388e-06,
+      "loss": 0.7354,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2679628064243448e-06,
+      "loss": 0.6424,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2713440405748098e-06,
+      "loss": 0.667,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2747252747252748e-06,
+      "loss": 0.6243,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2781065088757397e-06,
+      "loss": 0.8539,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2814877430262047e-06,
+      "loss": 0.6953,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2848689771766695e-06,
+      "loss": 0.6877,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2882502113271344e-06,
+      "loss": 0.6707,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2916314454775994e-06,
+      "loss": 0.6852,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2950126796280644e-06,
+      "loss": 0.6928,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2983939137785293e-06,
+      "loss": 0.6713,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.301775147928994e-06,
+      "loss": 0.6856,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.305156382079459e-06,
+      "loss": 0.7163,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.308537616229924e-06,
+      "loss": 0.7558,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.311918850380389e-06,
+      "loss": 0.7646,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.315300084530854e-06,
+      "loss": 0.6887,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3186813186813187e-06,
+      "loss": 0.7559,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3220625528317837e-06,
+      "loss": 0.6629,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3254437869822487e-06,
+      "loss": 0.7712,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3288250211327137e-06,
+      "loss": 0.7747,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3322062552831786e-06,
+      "loss": 0.6981,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3355874894336434e-06,
+      "loss": 0.7184,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3389687235841084e-06,
+      "loss": 0.6239,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3423499577345733e-06,
+      "loss": 0.7045,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3457311918850383e-06,
+      "loss": 0.7092,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3491124260355033e-06,
+      "loss": 0.6732,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.352493660185968e-06,
+      "loss": 0.716,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.355874894336433e-06,
+      "loss": 0.7831,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3592561284868978e-06,
+      "loss": 0.6553,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3626373626373625e-06,
+      "loss": 0.6947,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3660185967878275e-06,
+      "loss": 0.7755,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3693998309382925e-06,
+      "loss": 0.7837,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3727810650887574e-06,
+      "loss": 0.6858,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3761622992392224e-06,
+      "loss": 0.6991,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3795435333896872e-06,
+      "loss": 0.6839,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3829247675401521e-06,
+      "loss": 0.6691,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3863060016906171e-06,
+      "loss": 0.7162,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.389687235841082e-06,
+      "loss": 0.5976,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.393068469991547e-06,
+      "loss": 0.6646,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3964497041420118e-06,
+      "loss": 0.8182,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3998309382924768e-06,
+      "loss": 0.6572,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4032121724429418e-06,
+      "loss": 0.679,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4065934065934067e-06,
+      "loss": 0.6038,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4099746407438717e-06,
+      "loss": 0.6549,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4133558748943365e-06,
+      "loss": 0.7016,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4167371090448014e-06,
+      "loss": 0.7055,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4201183431952664e-06,
+      "loss": 0.6206,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4234995773457314e-06,
+      "loss": 0.8575,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4268808114961964e-06,
+      "loss": 0.6753,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4302620456466611e-06,
+      "loss": 0.7286,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.433643279797126e-06,
+      "loss": 0.7764,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.437024513947591e-06,
+      "loss": 0.6337,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.440405748098056e-06,
+      "loss": 0.6921,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.443786982248521e-06,
+      "loss": 0.7644,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4471682163989858e-06,
+      "loss": 0.6838,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4505494505494507e-06,
+      "loss": 0.7146,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4539306846999157e-06,
+      "loss": 0.6303,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4573119188503805e-06,
+      "loss": 0.6206,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4606931530008452e-06,
+      "loss": 0.7175,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4640743871513102e-06,
+      "loss": 0.8261,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4674556213017752e-06,
+      "loss": 0.733,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4708368554522401e-06,
+      "loss": 0.6831,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4742180896027051e-06,
+      "loss": 0.702,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4775993237531699e-06,
+      "loss": 0.6673,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4809805579036348e-06,
+      "loss": 0.7119,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4843617920540998e-06,
+      "loss": 0.673,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4877430262045648e-06,
+      "loss": 0.6511,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4911242603550298e-06,
+      "loss": 0.6498,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4945054945054945e-06,
+      "loss": 0.6299,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4978867286559595e-06,
+      "loss": 0.5941,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5012679628064245e-06,
+      "loss": 0.6394,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5046491969568894e-06,
+      "loss": 0.6897,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5080304311073544e-06,
+      "loss": 0.6937,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5114116652578192e-06,
+      "loss": 0.7291,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5147928994082841e-06,
+      "loss": 0.6921,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.518174133558749e-06,
+      "loss": 0.7368,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.521555367709214e-06,
+      "loss": 0.7673,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.524936601859679e-06,
+      "loss": 0.7008,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5283178360101438e-06,
+      "loss": 0.7177,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5316990701606088e-06,
+      "loss": 0.7052,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5350803043110737e-06,
+      "loss": 0.6517,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 0.6981,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5418427726120037e-06,
+      "loss": 0.6331,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5452240067624684e-06,
+      "loss": 0.8254,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5486052409129334e-06,
+      "loss": 0.7175,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5519864750633984e-06,
+      "loss": 0.7352,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5553677092138634e-06,
+      "loss": 0.6529,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.558748943364328e-06,
+      "loss": 0.5712,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5621301775147929e-06,
+      "loss": 0.6677,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5655114116652579e-06,
+      "loss": 0.6823,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5688926458157228e-06,
+      "loss": 0.731,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5722738799661876e-06,
+      "loss": 0.5844,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5756551141166526e-06,
+      "loss": 0.6419,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5790363482671175e-06,
+      "loss": 0.6974,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5824175824175825e-06,
+      "loss": 0.6109,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5857988165680475e-06,
+      "loss": 0.7857,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5891800507185122e-06,
+      "loss": 0.696,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5925612848689772e-06,
+      "loss": 0.6868,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5959425190194422e-06,
+      "loss": 0.7722,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5993237531699071e-06,
+      "loss": 0.6192,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6027049873203721e-06,
+      "loss": 0.6807,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6060862214708369e-06,
+      "loss": 0.6363,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6094674556213018e-06,
+      "loss": 0.6276,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6128486897717668e-06,
+      "loss": 0.6295,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6162299239222318e-06,
+      "loss": 0.6319,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6196111580726968e-06,
+      "loss": 0.8256,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6229923922231617e-06,
+      "loss": 0.5766,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6263736263736265e-06,
+      "loss": 0.6721,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6297548605240915e-06,
+      "loss": 0.6339,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6331360946745564e-06,
+      "loss": 0.6538,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6365173288250214e-06,
+      "loss": 0.6438,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6398985629754864e-06,
+      "loss": 0.675,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6432797971259511e-06,
+      "loss": 0.6415,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6466610312764161e-06,
+      "loss": 0.7785,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.650042265426881e-06,
+      "loss": 0.6313,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.653423499577346e-06,
+      "loss": 0.6944,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.656804733727811e-06,
+      "loss": 0.6694,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6601859678782756e-06,
+      "loss": 0.6537,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6635672020287405e-06,
+      "loss": 0.6894,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6669484361792055e-06,
+      "loss": 0.6845,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6703296703296703e-06,
+      "loss": 0.6865,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6737109044801352e-06,
+      "loss": 0.7409,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6770921386306002e-06,
+      "loss": 0.7037,
+      "step": 496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6804733727810652e-06,
+      "loss": 0.6469,
+      "step": 497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6838546069315302e-06,
+      "loss": 0.6976,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.687235841081995e-06,
+      "loss": 0.6626,
+      "step": 499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6906170752324599e-06,
+      "loss": 0.5992,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6939983093829249e-06,
+      "loss": 0.6739,
+      "step": 501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6973795435333898e-06,
+      "loss": 0.6419,
+      "step": 502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7007607776838548e-06,
+      "loss": 0.6347,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7041420118343196e-06,
+      "loss": 0.7845,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7075232459847845e-06,
+      "loss": 0.7164,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7109044801352495e-06,
+      "loss": 0.661,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 0.6857,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7176669484361795e-06,
+      "loss": 0.7538,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7210481825866442e-06,
+      "loss": 0.6975,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7244294167371092e-06,
+      "loss": 0.7843,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7278106508875742e-06,
+      "loss": 0.7198,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7311918850380391e-06,
+      "loss": 0.6599,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.734573119188504e-06,
+      "loss": 0.6375,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7379543533389689e-06,
+      "loss": 0.7289,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7413355874894338e-06,
+      "loss": 0.6436,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7447168216398988e-06,
+      "loss": 0.7382,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7480980557903638e-06,
+      "loss": 0.7496,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7514792899408287e-06,
+      "loss": 0.7047,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7548605240912935e-06,
+      "loss": 0.774,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7582417582417585e-06,
+      "loss": 0.6602,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7616229923922232e-06,
+      "loss": 0.7271,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7650042265426882e-06,
+      "loss": 0.6862,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.768385460693153e-06,
+      "loss": 0.6593,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.771766694843618e-06,
+      "loss": 0.6725,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.775147928994083e-06,
+      "loss": 0.6337,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7785291631445479e-06,
+      "loss": 0.6587,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7819103972950129e-06,
+      "loss": 0.6488,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7852916314454776e-06,
+      "loss": 0.6341,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7886728655959426e-06,
+      "loss": 0.7121,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7920540997464076e-06,
+      "loss": 0.593,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7954353338968725e-06,
+      "loss": 0.5929,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7988165680473375e-06,
+      "loss": 0.7372,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8021978021978023e-06,
+      "loss": 0.6685,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8055790363482672e-06,
+      "loss": 0.629,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8089602704987322e-06,
+      "loss": 0.6132,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8123415046491972e-06,
+      "loss": 0.6883,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8157227387996621e-06,
+      "loss": 0.639,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.819103972950127e-06,
+      "loss": 0.6708,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8224852071005919e-06,
+      "loss": 0.6451,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8258664412510568e-06,
+      "loss": 0.6265,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8292476754015218e-06,
+      "loss": 0.6602,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8326289095519868e-06,
+      "loss": 0.6963,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8360101437024515e-06,
+      "loss": 0.7581,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8393913778529165e-06,
+      "loss": 0.6489,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8427726120033815e-06,
+      "loss": 0.7183,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8461538461538465e-06,
+      "loss": 0.6725,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8495350803043114e-06,
+      "loss": 0.6419,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8529163144547762e-06,
+      "loss": 0.6565,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8562975486052412e-06,
+      "loss": 0.715,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.859678782755706e-06,
+      "loss": 0.6485,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8630600169061707e-06,
+      "loss": 0.7385,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8664412510566357e-06,
+      "loss": 0.7119,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8698224852071006e-06,
+      "loss": 0.6927,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8732037193575656e-06,
+      "loss": 0.7489,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8765849535080306e-06,
+      "loss": 0.5998,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8799661876584953e-06,
+      "loss": 0.6468,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8833474218089603e-06,
+      "loss": 0.6863,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8867286559594253e-06,
+      "loss": 0.7612,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8901098901098902e-06,
+      "loss": 0.7607,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8934911242603552e-06,
+      "loss": 0.7788,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.89687235841082e-06,
+      "loss": 0.6988,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.900253592561285e-06,
+      "loss": 0.698,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.90363482671175e-06,
+      "loss": 0.7756,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9070160608622149e-06,
+      "loss": 0.6738,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9103972950126796e-06,
+      "loss": 0.6353,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9137785291631446e-06,
+      "loss": 0.6593,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9171597633136096e-06,
+      "loss": 0.6569,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9205409974640746e-06,
+      "loss": 0.6703,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9239222316145395e-06,
+      "loss": 0.6555,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9273034657650045e-06,
+      "loss": 0.6634,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9306846999154695e-06,
+      "loss": 0.6879,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9340659340659344e-06,
+      "loss": 0.7171,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9374471682163994e-06,
+      "loss": 0.6854,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.940828402366864e-06,
+      "loss": 0.739,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.944209636517329e-06,
+      "loss": 0.6325,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.947590870667794e-06,
+      "loss": 0.7725,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.950972104818259e-06,
+      "loss": 0.7761,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.954353338968724e-06,
+      "loss": 0.6419,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.957734573119189e-06,
+      "loss": 0.6681,
+      "step": 579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9611158072696534e-06,
+      "loss": 0.6779,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9644970414201183e-06,
+      "loss": 0.6521,
+      "step": 581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9678782755705833e-06,
+      "loss": 0.6212,
+      "step": 582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9712595097210483e-06,
+      "loss": 0.6352,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9746407438715133e-06,
+      "loss": 0.6431,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9780219780219782e-06,
+      "loss": 0.7774,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981403212172443e-06,
+      "loss": 0.7052,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9847844463229077e-06,
+      "loss": 0.6484,
+      "step": 587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9881656804733727e-06,
+      "loss": 0.6303,
+      "step": 588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9915469146238377e-06,
+      "loss": 0.711,
+      "step": 589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949281487743027e-06,
+      "loss": 0.6805,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9983093829247676e-06,
+      "loss": 0.6583,
+      "step": 591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0016906170752326e-06,
+      "loss": 0.6542,
+      "step": 592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0050718512256976e-06,
+      "loss": 0.7289,
+      "step": 593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0084530853761625e-06,
+      "loss": 0.6762,
+      "step": 594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0118343195266275e-06,
+      "loss": 0.5694,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0152155536770925e-06,
+      "loss": 0.6703,
+      "step": 596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.018596787827557e-06,
+      "loss": 0.6077,
+      "step": 597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.021978021978022e-06,
+      "loss": 0.7669,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.025359256128487e-06,
+      "loss": 0.6345,
+      "step": 599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.028740490278952e-06,
+      "loss": 0.7296,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.032121724429417e-06,
+      "loss": 0.6609,
+      "step": 601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.035502958579882e-06,
+      "loss": 0.8128,
+      "step": 602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.038884192730347e-06,
+      "loss": 0.6622,
+      "step": 603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.042265426880812e-06,
+      "loss": 0.7406,
+      "step": 604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.045646661031277e-06,
+      "loss": 0.5885,
+      "step": 605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0490278951817418e-06,
+      "loss": 0.6246,
+      "step": 606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0524091293322063e-06,
+      "loss": 0.8367,
+      "step": 607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0557903634826713e-06,
+      "loss": 0.6493,
+      "step": 608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0591715976331363e-06,
+      "loss": 0.6977,
+      "step": 609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0625528317836012e-06,
+      "loss": 0.748,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.065934065934066e-06,
+      "loss": 0.6473,
+      "step": 611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0693153000845308e-06,
+      "loss": 0.6085,
+      "step": 612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0726965342349957e-06,
+      "loss": 0.5958,
+      "step": 613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0760777683854607e-06,
+      "loss": 0.7222,
+      "step": 614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0794590025359257e-06,
+      "loss": 0.6004,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0828402366863907e-06,
+      "loss": 0.6906,
+      "step": 616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0862214708368556e-06,
+      "loss": 0.6257,
+      "step": 617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0896027049873206e-06,
+      "loss": 0.6513,
+      "step": 618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0929839391377856e-06,
+      "loss": 0.6803,
+      "step": 619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0963651732882505e-06,
+      "loss": 0.7363,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.099746407438715e-06,
+      "loss": 0.8088,
+      "step": 621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.10312764158918e-06,
+      "loss": 0.7729,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.106508875739645e-06,
+      "loss": 0.7027,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.10989010989011e-06,
+      "loss": 0.7291,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.113271344040575e-06,
+      "loss": 0.6706,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.11665257819104e-06,
+      "loss": 0.6745,
+      "step": 626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.120033812341505e-06,
+      "loss": 0.6618,
+      "step": 627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.12341504649197e-06,
+      "loss": 0.6517,
+      "step": 628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.126796280642435e-06,
+      "loss": 0.7892,
+      "step": 629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1301775147929e-06,
+      "loss": 0.8368,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1335587489433644e-06,
+      "loss": 0.5665,
+      "step": 631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1369399830938293e-06,
+      "loss": 0.6781,
+      "step": 632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1403212172442943e-06,
+      "loss": 0.6193,
+      "step": 633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1437024513947593e-06,
+      "loss": 0.7785,
+      "step": 634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1470836855452243e-06,
+      "loss": 0.6836,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1504649196956892e-06,
+      "loss": 0.6902,
+      "step": 636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.153846153846154e-06,
+      "loss": 0.6219,
+      "step": 637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.157227387996619e-06,
+      "loss": 0.7328,
+      "step": 638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.160608622147084e-06,
+      "loss": 0.627,
+      "step": 639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1639898562975487e-06,
+      "loss": 0.7518,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1673710904480137e-06,
+      "loss": 0.6305,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1707523245984786e-06,
+      "loss": 0.6444,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1741335587489436e-06,
+      "loss": 0.6238,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.177514792899408e-06,
+      "loss": 0.6198,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.180896027049873e-06,
+      "loss": 0.7269,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.184277261200338e-06,
+      "loss": 0.674,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.187658495350803e-06,
+      "loss": 0.7181,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191039729501268e-06,
+      "loss": 0.7025,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.194420963651733e-06,
+      "loss": 0.675,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.197802197802198e-06,
+      "loss": 0.5911,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.201183431952663e-06,
+      "loss": 0.7825,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.204564666103128e-06,
+      "loss": 0.732,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.207945900253593e-06,
+      "loss": 0.7057,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.211327134404058e-06,
+      "loss": 0.6455,
+      "step": 654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2147083685545224e-06,
+      "loss": 0.7833,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2180896027049874e-06,
+      "loss": 0.6476,
+      "step": 656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2214708368554524e-06,
+      "loss": 0.6768,
+      "step": 657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2248520710059173e-06,
+      "loss": 0.6604,
+      "step": 658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2282333051563823e-06,
+      "loss": 0.7093,
+      "step": 659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2316145393068473e-06,
+      "loss": 0.6321,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2349957734573122e-06,
+      "loss": 0.6781,
+      "step": 661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2383770076077772e-06,
+      "loss": 0.6762,
+      "step": 662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.241758241758242e-06,
+      "loss": 0.5986,
+      "step": 663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.245139475908707e-06,
+      "loss": 0.6806,
+      "step": 664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2485207100591717e-06,
+      "loss": 0.7361,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2519019442096367e-06,
+      "loss": 0.708,
+      "step": 666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2552831783601017e-06,
+      "loss": 0.6796,
+      "step": 667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2586644125105666e-06,
+      "loss": 0.6934,
+      "step": 668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.262045646661031e-06,
+      "loss": 0.7248,
+      "step": 669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.265426880811496e-06,
+      "loss": 0.5709,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.268808114961961e-06,
+      "loss": 0.7377,
+      "step": 671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.272189349112426e-06,
+      "loss": 0.6894,
+      "step": 672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.275570583262891e-06,
+      "loss": 0.7917,
+      "step": 673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.278951817413356e-06,
+      "loss": 0.6434,
+      "step": 674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.282333051563821e-06,
+      "loss": 0.63,
+      "step": 675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 0.6353,
+      "step": 676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.289095519864751e-06,
+      "loss": 0.6548,
+      "step": 677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2924767540152155e-06,
+      "loss": 0.6622,
+      "step": 678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2958579881656805e-06,
+      "loss": 0.6066,
+      "step": 679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2992392223161454e-06,
+      "loss": 0.6572,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3026204564666104e-06,
+      "loss": 0.6226,
+      "step": 681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3060016906170754e-06,
+      "loss": 0.6852,
+      "step": 682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3093829247675404e-06,
+      "loss": 0.6202,
+      "step": 683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3127641589180053e-06,
+      "loss": 0.6237,
+      "step": 684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3161453930684703e-06,
+      "loss": 0.6322,
+      "step": 685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3195266272189353e-06,
+      "loss": 0.6655,
+      "step": 686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3229078613694002e-06,
+      "loss": 0.7982,
+      "step": 687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3262890955198648e-06,
+      "loss": 0.6227,
+      "step": 688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3296703296703298e-06,
+      "loss": 0.7828,
+      "step": 689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3330515638207947e-06,
+      "loss": 0.6588,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3364327979712597e-06,
+      "loss": 0.6854,
+      "step": 691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3398140321217247e-06,
+      "loss": 0.6765,
+      "step": 692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3431952662721896e-06,
+      "loss": 0.6893,
+      "step": 693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3465765004226546e-06,
+      "loss": 0.6526,
+      "step": 694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3499577345731196e-06,
+      "loss": 0.6975,
+      "step": 695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3533389687235846e-06,
+      "loss": 0.6867,
+      "step": 696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3567202028740495e-06,
+      "loss": 0.6214,
+      "step": 697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.360101437024514e-06,
+      "loss": 0.7268,
+      "step": 698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.363482671174979e-06,
+      "loss": 0.6508,
+      "step": 699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.366863905325444e-06,
+      "loss": 0.7094,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.370245139475909e-06,
+      "loss": 0.7387,
+      "step": 701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3736263736263735e-06,
+      "loss": 0.6063,
+      "step": 702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3770076077768385e-06,
+      "loss": 0.6965,
+      "step": 703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3803888419273035e-06,
+      "loss": 0.6296,
+      "step": 704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3837700760777685e-06,
+      "loss": 0.763,
+      "step": 705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3871513102282334e-06,
+      "loss": 0.6599,
+      "step": 706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3905325443786984e-06,
+      "loss": 0.6146,
+      "step": 707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3939137785291634e-06,
+      "loss": 0.6733,
+      "step": 708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3972950126796283e-06,
+      "loss": 0.6983,
+      "step": 709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4006762468300933e-06,
+      "loss": 0.6781,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4040574809805583e-06,
+      "loss": 0.6878,
+      "step": 711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.407438715131023e-06,
+      "loss": 0.7422,
+      "step": 712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.410819949281488e-06,
+      "loss": 0.64,
+      "step": 713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4142011834319528e-06,
+      "loss": 0.7299,
+      "step": 714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4175824175824177e-06,
+      "loss": 0.614,
+      "step": 715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4209636517328827e-06,
+      "loss": 0.6531,
+      "step": 716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4243448858833477e-06,
+      "loss": 0.6705,
+      "step": 717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4277261200338127e-06,
+      "loss": 0.6361,
+      "step": 718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4311073541842776e-06,
+      "loss": 0.7511,
+      "step": 719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4344885883347426e-06,
+      "loss": 0.7065,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4378698224852076e-06,
+      "loss": 0.6615,
+      "step": 721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.441251056635672e-06,
+      "loss": 0.6286,
+      "step": 722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.444632290786137e-06,
+      "loss": 0.6769,
+      "step": 723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.448013524936602e-06,
+      "loss": 0.74,
+      "step": 724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.451394759087067e-06,
+      "loss": 0.6304,
+      "step": 725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.454775993237532e-06,
+      "loss": 0.6181,
+      "step": 726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.458157227387997e-06,
+      "loss": 0.7595,
+      "step": 727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.461538461538462e-06,
+      "loss": 0.706,
+      "step": 728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4649196956889265e-06,
+      "loss": 0.6376,
+      "step": 729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4683009298393915e-06,
+      "loss": 0.7041,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4716821639898564e-06,
+      "loss": 0.7065,
+      "step": 731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4750633981403214e-06,
+      "loss": 0.7281,
+      "step": 732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4784446322907864e-06,
+      "loss": 0.7293,
+      "step": 733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4818258664412514e-06,
+      "loss": 0.6847,
+      "step": 734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.485207100591716e-06,
+      "loss": 0.6399,
+      "step": 735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.488588334742181e-06,
+      "loss": 0.7018,
+      "step": 736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.491969568892646e-06,
+      "loss": 0.6518,
+      "step": 737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.495350803043111e-06,
+      "loss": 0.6587,
+      "step": 738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4987320371935758e-06,
+      "loss": 0.6164,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.502113271344041e-06,
+      "loss": 0.6664,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5054945054945057e-06,
+      "loss": 0.5703,
+      "step": 741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5088757396449707e-06,
+      "loss": 0.5991,
+      "step": 742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5122569737954357e-06,
+      "loss": 0.6942,
+      "step": 743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5156382079459002e-06,
+      "loss": 0.6342,
+      "step": 744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.519019442096365e-06,
+      "loss": 0.6315,
+      "step": 745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.52240067624683e-06,
+      "loss": 0.8238,
+      "step": 746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.525781910397295e-06,
+      "loss": 0.8171,
+      "step": 747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.52916314454776e-06,
+      "loss": 0.7052,
+      "step": 748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.532544378698225e-06,
+      "loss": 0.774,
+      "step": 749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5359256128486896e-06,
+      "loss": 0.7388,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.539306846999155e-06,
+      "loss": 0.6783,
+      "step": 751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5426880811496196e-06,
+      "loss": 0.6584,
+      "step": 752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.546069315300085e-06,
+      "loss": 0.7775,
+      "step": 753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5494505494505495e-06,
+      "loss": 0.7119,
+      "step": 754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5528317836010145e-06,
+      "loss": 0.6931,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5562130177514795e-06,
+      "loss": 0.596,
+      "step": 756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5595942519019444e-06,
+      "loss": 0.6726,
+      "step": 757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5629754860524094e-06,
+      "loss": 0.6574,
+      "step": 758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5663567202028744e-06,
+      "loss": 0.5497,
+      "step": 759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.569737954353339e-06,
+      "loss": 0.662,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5731191885038043e-06,
+      "loss": 0.7095,
+      "step": 761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.576500422654269e-06,
+      "loss": 0.7207,
+      "step": 762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5798816568047343e-06,
+      "loss": 0.6644,
+      "step": 763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.583262890955199e-06,
+      "loss": 0.6834,
+      "step": 764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5866441251056638e-06,
+      "loss": 0.6731,
+      "step": 765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5900253592561287e-06,
+      "loss": 0.6109,
+      "step": 766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5934065934065937e-06,
+      "loss": 0.6777,
+      "step": 767
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5967878275570587e-06,
+      "loss": 0.6926,
+      "step": 768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6001690617075237e-06,
+      "loss": 0.7438,
+      "step": 769
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.603550295857988e-06,
+      "loss": 0.6364,
+      "step": 770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6069315300084536e-06,
+      "loss": 0.6425,
+      "step": 771
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.610312764158918e-06,
+      "loss": 0.737,
+      "step": 772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6136939983093827e-06,
+      "loss": 0.7169,
+      "step": 773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.617075232459848e-06,
+      "loss": 0.6769,
+      "step": 774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6204564666103126e-06,
+      "loss": 0.6068,
+      "step": 775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.623837700760778e-06,
+      "loss": 0.6964,
+      "step": 776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6272189349112426e-06,
+      "loss": 0.6195,
+      "step": 777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.630600169061708e-06,
+      "loss": 0.6645,
+      "step": 778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6339814032121725e-06,
+      "loss": 0.6778,
+      "step": 779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6373626373626375e-06,
+      "loss": 0.6596,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6407438715131025e-06,
+      "loss": 0.7635,
+      "step": 781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6441251056635674e-06,
+      "loss": 0.6367,
+      "step": 782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.647506339814032e-06,
+      "loss": 0.6545,
+      "step": 783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6508875739644974e-06,
+      "loss": 0.6973,
+      "step": 784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.654268808114962e-06,
+      "loss": 0.5745,
+      "step": 785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6576500422654273e-06,
+      "loss": 0.766,
+      "step": 786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.661031276415892e-06,
+      "loss": 0.5939,
+      "step": 787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6644125105663573e-06,
+      "loss": 0.6514,
+      "step": 788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.667793744716822e-06,
+      "loss": 0.7316,
+      "step": 789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.671174978867287e-06,
+      "loss": 0.6906,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6745562130177518e-06,
+      "loss": 0.6702,
+      "step": 791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6779374471682167e-06,
+      "loss": 0.6382,
+      "step": 792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6813186813186813e-06,
+      "loss": 0.6369,
+      "step": 793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6846999154691467e-06,
+      "loss": 0.6747,
+      "step": 794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6880811496196112e-06,
+      "loss": 0.7745,
+      "step": 795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6914623837700766e-06,
+      "loss": 0.7084,
+      "step": 796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.694843617920541e-06,
+      "loss": 0.6266,
+      "step": 797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6982248520710066e-06,
+      "loss": 0.6839,
+      "step": 798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.701606086221471e-06,
+      "loss": 0.6253,
+      "step": 799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.704987320371936e-06,
+      "loss": 0.6055,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.708368554522401e-06,
+      "loss": 0.6685,
+      "step": 801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.711749788672866e-06,
+      "loss": 0.5879,
+      "step": 802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7151310228233306e-06,
+      "loss": 0.6599,
+      "step": 803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7185122569737955e-06,
+      "loss": 0.5956,
+      "step": 804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7218934911242605e-06,
+      "loss": 0.7602,
+      "step": 805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.725274725274725e-06,
+      "loss": 0.6035,
+      "step": 806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7286559594251905e-06,
+      "loss": 0.6683,
+      "step": 807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.732037193575655e-06,
+      "loss": 0.5944,
+      "step": 808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7354184277261204e-06,
+      "loss": 0.5951,
+      "step": 809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.738799661876585e-06,
+      "loss": 0.6393,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7421808960270503e-06,
+      "loss": 0.6745,
+      "step": 811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.745562130177515e-06,
+      "loss": 0.6384,
+      "step": 812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.74894336432798e-06,
+      "loss": 0.6719,
+      "step": 813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.752324598478445e-06,
+      "loss": 0.6869,
+      "step": 814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.75570583262891e-06,
+      "loss": 0.6594,
+      "step": 815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7590870667793744e-06,
+      "loss": 0.6272,
+      "step": 816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7624683009298397e-06,
+      "loss": 0.6888,
+      "step": 817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7658495350803043e-06,
+      "loss": 0.6952,
+      "step": 818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7692307692307697e-06,
+      "loss": 0.7145,
+      "step": 819
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7726120033812342e-06,
+      "loss": 0.6802,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7759932375316996e-06,
+      "loss": 0.7171,
+      "step": 821
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.779374471682164e-06,
+      "loss": 0.7621,
+      "step": 822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.782755705832629e-06,
+      "loss": 0.6667,
+      "step": 823
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.786136939983094e-06,
+      "loss": 0.6855,
+      "step": 824
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.789518174133559e-06,
+      "loss": 0.7034,
+      "step": 825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7928994082840236e-06,
+      "loss": 0.6762,
+      "step": 826
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.796280642434489e-06,
+      "loss": 0.6043,
+      "step": 827
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7996618765849536e-06,
+      "loss": 0.6405,
+      "step": 828
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.803043110735419e-06,
+      "loss": 0.6803,
+      "step": 829
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8064243448858835e-06,
+      "loss": 0.6394,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.809805579036349e-06,
+      "loss": 0.6743,
+      "step": 831
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8131868131868135e-06,
+      "loss": 0.6993,
+      "step": 832
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.816568047337278e-06,
+      "loss": 0.6044,
+      "step": 833
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8199492814877434e-06,
+      "loss": 0.6217,
+      "step": 834
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.823330515638208e-06,
+      "loss": 0.7508,
+      "step": 835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.826711749788673e-06,
+      "loss": 0.7388,
+      "step": 836
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.830092983939138e-06,
+      "loss": 0.5985,
+      "step": 837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.833474218089603e-06,
+      "loss": 0.7405,
+      "step": 838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.836855452240068e-06,
+      "loss": 0.6618,
+      "step": 839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.840236686390533e-06,
+      "loss": 0.7212,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8436179205409974e-06,
+      "loss": 0.669,
+      "step": 841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8469991546914628e-06,
+      "loss": 0.7337,
+      "step": 842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8503803888419273e-06,
+      "loss": 0.6523,
+      "step": 843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8537616229923927e-06,
+      "loss": 0.6807,
+      "step": 844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.6467,
+      "step": 845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8605240912933222e-06,
+      "loss": 0.5955,
+      "step": 846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.863905325443787e-06,
+      "loss": 0.5979,
+      "step": 847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.867286559594252e-06,
+      "loss": 0.6267,
+      "step": 848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.870667793744717e-06,
+      "loss": 0.6998,
+      "step": 849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.874049027895182e-06,
+      "loss": 0.6098,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8774302620456467e-06,
+      "loss": 0.6214,
+      "step": 851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.880811496196112e-06,
+      "loss": 0.7644,
+      "step": 852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8841927303465766e-06,
+      "loss": 0.6674,
+      "step": 853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.887573964497042e-06,
+      "loss": 0.701,
+      "step": 854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8909551986475065e-06,
+      "loss": 0.7075,
+      "step": 855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8943364327979715e-06,
+      "loss": 0.6775,
+      "step": 856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8977176669484365e-06,
+      "loss": 0.755,
+      "step": 857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9010989010989015e-06,
+      "loss": 0.6463,
+      "step": 858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9044801352493664e-06,
+      "loss": 0.584,
+      "step": 859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9078613693998314e-06,
+      "loss": 0.6721,
+      "step": 860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.911242603550296e-06,
+      "loss": 0.7198,
+      "step": 861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.914623837700761e-06,
+      "loss": 0.6284,
+      "step": 862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.918005071851226e-06,
+      "loss": 0.614,
+      "step": 863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9213863060016904e-06,
+      "loss": 0.6406,
+      "step": 864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.924767540152156e-06,
+      "loss": 0.6762,
+      "step": 865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9281487743026204e-06,
+      "loss": 0.6552,
+      "step": 866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9315300084530858e-06,
+      "loss": 0.5797,
+      "step": 867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9349112426035503e-06,
+      "loss": 0.5822,
+      "step": 868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9382924767540157e-06,
+      "loss": 0.6212,
+      "step": 869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9416737109044803e-06,
+      "loss": 0.6132,
+      "step": 870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9450549450549452e-06,
+      "loss": 0.6599,
+      "step": 871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9484361792054102e-06,
+      "loss": 0.7397,
+      "step": 872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.951817413355875e-06,
+      "loss": 0.663,
+      "step": 873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9551986475063397e-06,
+      "loss": 0.5567,
+      "step": 874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.958579881656805e-06,
+      "loss": 0.6774,
+      "step": 875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9619611158072697e-06,
+      "loss": 0.6384,
+      "step": 876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.965342349957735e-06,
+      "loss": 0.7716,
+      "step": 877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9687235841081996e-06,
+      "loss": 0.6332,
+      "step": 878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.972104818258665e-06,
+      "loss": 0.7161,
+      "step": 879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9754860524091296e-06,
+      "loss": 0.6598,
+      "step": 880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9788672865595945e-06,
+      "loss": 0.6838,
+      "step": 881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9822485207100595e-06,
+      "loss": 0.6047,
+      "step": 882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9856297548605245e-06,
+      "loss": 0.6542,
+      "step": 883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.989010989010989e-06,
+      "loss": 0.6585,
+      "step": 884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9923922231614544e-06,
+      "loss": 0.6183,
+      "step": 885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.995773457311919e-06,
+      "loss": 0.6322,
+      "step": 886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9991546914623844e-06,
+      "loss": 0.6692,
+      "step": 887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.002535925612849e-06,
+      "loss": 0.7321,
+      "step": 888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0059171597633143e-06,
+      "loss": 0.587,
+      "step": 889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.009298393913779e-06,
+      "loss": 0.7156,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.012679628064244e-06,
+      "loss": 0.7063,
+      "step": 891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.016060862214709e-06,
+      "loss": 0.6375,
+      "step": 892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0194420963651733e-06,
+      "loss": 0.6187,
+      "step": 893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0228233305156383e-06,
+      "loss": 0.5756,
+      "step": 894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0262045646661033e-06,
+      "loss": 0.6114,
+      "step": 895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0295857988165683e-06,
+      "loss": 0.6344,
+      "step": 896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.032967032967033e-06,
+      "loss": 0.6376,
+      "step": 897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.036348267117498e-06,
+      "loss": 0.6289,
+      "step": 898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0397295012679627e-06,
+      "loss": 0.7089,
+      "step": 899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.043110735418428e-06,
+      "loss": 0.5804,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0464919695688927e-06,
+      "loss": 0.5615,
+      "step": 901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.049873203719358e-06,
+      "loss": 0.693,
+      "step": 902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0532544378698226e-06,
+      "loss": 0.7137,
+      "step": 903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0566356720202876e-06,
+      "loss": 0.6853,
+      "step": 904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0600169061707526e-06,
+      "loss": 0.709,
+      "step": 905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0633981403212175e-06,
+      "loss": 0.7378,
+      "step": 906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.066779374471682e-06,
+      "loss": 0.7192,
+      "step": 907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0701606086221475e-06,
+      "loss": 0.6165,
+      "step": 908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.073541842772612e-06,
+      "loss": 0.6109,
+      "step": 909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.6233,
+      "step": 910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.080304311073542e-06,
+      "loss": 0.6088,
+      "step": 911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0836855452240074e-06,
+      "loss": 0.6201,
+      "step": 912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.087066779374472e-06,
+      "loss": 0.7977,
+      "step": 913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.090448013524937e-06,
+      "loss": 0.6758,
+      "step": 914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.093829247675402e-06,
+      "loss": 0.6061,
+      "step": 915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.097210481825867e-06,
+      "loss": 0.698,
+      "step": 916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1005917159763314e-06,
+      "loss": 0.5986,
+      "step": 917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1039729501267968e-06,
+      "loss": 0.6419,
+      "step": 918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1073541842772613e-06,
+      "loss": 0.6861,
+      "step": 919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1107354184277267e-06,
+      "loss": 0.6109,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1141166525781913e-06,
+      "loss": 0.7206,
+      "step": 921
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.117497886728656e-06,
+      "loss": 0.6596,
+      "step": 922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1208791208791212e-06,
+      "loss": 0.656,
+      "step": 923
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1242603550295858e-06,
+      "loss": 0.6617,
+      "step": 924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.127641589180051e-06,
+      "loss": 0.6678,
+      "step": 925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1310228233305157e-06,
+      "loss": 0.675,
+      "step": 926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1344040574809807e-06,
+      "loss": 0.6086,
+      "step": 927
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1377852916314457e-06,
+      "loss": 0.6263,
+      "step": 928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1411665257819106e-06,
+      "loss": 0.6776,
+      "step": 929
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.144547759932375e-06,
+      "loss": 0.6012,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1479289940828406e-06,
+      "loss": 0.738,
+      "step": 931
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.151310228233305e-06,
+      "loss": 0.7042,
+      "step": 932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1546914623837705e-06,
+      "loss": 0.6595,
+      "step": 933
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.158072696534235e-06,
+      "loss": 0.6674,
+      "step": 934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1614539306847005e-06,
+      "loss": 0.6434,
+      "step": 935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.164835164835165e-06,
+      "loss": 0.6393,
+      "step": 936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.16821639898563e-06,
+      "loss": 0.6066,
+      "step": 937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.171597633136095e-06,
+      "loss": 0.6448,
+      "step": 938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.17497886728656e-06,
+      "loss": 0.7405,
+      "step": 939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1783601014370245e-06,
+      "loss": 0.6435,
+      "step": 940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.18174133558749e-06,
+      "loss": 0.6802,
+      "step": 941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1851225697379544e-06,
+      "loss": 0.6232,
+      "step": 942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.18850380388842e-06,
+      "loss": 0.6351,
+      "step": 943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1918850380388843e-06,
+      "loss": 0.7418,
+      "step": 944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1952662721893497e-06,
+      "loss": 0.6506,
+      "step": 945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1986475063398143e-06,
+      "loss": 0.6484,
+      "step": 946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2020287404902793e-06,
+      "loss": 0.6905,
+      "step": 947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2054099746407442e-06,
+      "loss": 0.5959,
+      "step": 948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.208791208791209e-06,
+      "loss": 0.6381,
+      "step": 949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2121724429416738e-06,
+      "loss": 0.6161,
+      "step": 950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.215553677092139e-06,
+      "loss": 0.6044,
+      "step": 951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2189349112426037e-06,
+      "loss": 0.6232,
+      "step": 952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2223161453930687e-06,
+      "loss": 0.6371,
+      "step": 953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2256973795435336e-06,
+      "loss": 0.6564,
+      "step": 954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.229078613693998e-06,
+      "loss": 0.7741,
+      "step": 955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2324598478444636e-06,
+      "loss": 0.6981,
+      "step": 956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.235841081994928e-06,
+      "loss": 0.6092,
+      "step": 957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2392223161453935e-06,
+      "loss": 0.6727,
+      "step": 958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.242603550295858e-06,
+      "loss": 0.6686,
+      "step": 959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2459847844463235e-06,
+      "loss": 0.6497,
+      "step": 960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.249366018596788e-06,
+      "loss": 0.649,
+      "step": 961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.252747252747253e-06,
+      "loss": 0.5959,
+      "step": 962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.256128486897718e-06,
+      "loss": 0.6148,
+      "step": 963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.259509721048183e-06,
+      "loss": 0.6551,
+      "step": 964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2628909551986475e-06,
+      "loss": 0.5917,
+      "step": 965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.266272189349113e-06,
+      "loss": 0.5681,
+      "step": 966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2696534234995774e-06,
+      "loss": 0.6658,
+      "step": 967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.273034657650043e-06,
+      "loss": 0.6454,
+      "step": 968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2764158918005074e-06,
+      "loss": 0.6036,
+      "step": 969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2797971259509728e-06,
+      "loss": 0.6393,
+      "step": 970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2831783601014373e-06,
+      "loss": 0.6988,
+      "step": 971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2865595942519023e-06,
+      "loss": 0.6659,
+      "step": 972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2899408284023672e-06,
+      "loss": 0.6724,
+      "step": 973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2933220625528322e-06,
+      "loss": 0.5963,
+      "step": 974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2967032967032968e-06,
+      "loss": 0.5466,
+      "step": 975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.300084530853762e-06,
+      "loss": 0.6079,
+      "step": 976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3034657650042267e-06,
+      "loss": 0.7467,
+      "step": 977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.306846999154692e-06,
+      "loss": 0.6978,
+      "step": 978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3102282333051567e-06,
+      "loss": 0.5843,
+      "step": 979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.313609467455622e-06,
+      "loss": 0.639,
+      "step": 980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3169907016060866e-06,
+      "loss": 0.6913,
+      "step": 981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.320371935756551e-06,
+      "loss": 0.7485,
+      "step": 982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3237531699070165e-06,
+      "loss": 0.5873,
+      "step": 983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.327134404057481e-06,
+      "loss": 0.5725,
+      "step": 984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.330515638207946e-06,
+      "loss": 0.6727,
+      "step": 985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.333896872358411e-06,
+      "loss": 0.6123,
+      "step": 986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.337278106508876e-06,
+      "loss": 0.7735,
+      "step": 987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3406593406593406e-06,
+      "loss": 0.5746,
+      "step": 988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.344040574809806e-06,
+      "loss": 0.7001,
+      "step": 989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3474218089602705e-06,
+      "loss": 0.6761,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.350803043110736e-06,
+      "loss": 0.6989,
+      "step": 991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3541842772612004e-06,
+      "loss": 0.5918,
+      "step": 992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.357565511411666e-06,
+      "loss": 0.5458,
+      "step": 993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3609467455621304e-06,
+      "loss": 0.605,
+      "step": 994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3643279797125954e-06,
+      "loss": 0.6215,
+      "step": 995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3677092138630603e-06,
+      "loss": 0.6193,
+      "step": 996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3710904480135253e-06,
+      "loss": 0.696,
+      "step": 997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.37447168216399e-06,
+      "loss": 0.5864,
+      "step": 998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3778529163144552e-06,
+      "loss": 0.6866,
+      "step": 999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3812341504649198e-06,
+      "loss": 0.6796,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.384615384615385e-06,
+      "loss": 0.7596,
+      "step": 1001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3879966187658497e-06,
+      "loss": 0.6051,
+      "step": 1002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.391377852916315e-06,
+      "loss": 0.561,
+      "step": 1003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3947590870667797e-06,
+      "loss": 0.5877,
+      "step": 1004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3981403212172446e-06,
+      "loss": 0.6473,
+      "step": 1005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4015215553677096e-06,
+      "loss": 0.6746,
+      "step": 1006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4049027895181746e-06,
+      "loss": 0.6852,
+      "step": 1007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.408284023668639e-06,
+      "loss": 0.6886,
+      "step": 1008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4116652578191045e-06,
+      "loss": 0.7638,
+      "step": 1009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.415046491969569e-06,
+      "loss": 0.6111,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4184277261200336e-06,
+      "loss": 0.7245,
+      "step": 1011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.421808960270499e-06,
+      "loss": 0.7653,
+      "step": 1012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4251901944209636e-06,
+      "loss": 0.6087,
+      "step": 1013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 0.6543,
+      "step": 1014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4319526627218935e-06,
+      "loss": 0.6482,
+      "step": 1015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.435333896872359e-06,
+      "loss": 0.7491,
+      "step": 1016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4387151310228235e-06,
+      "loss": 0.613,
+      "step": 1017
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4420963651732884e-06,
+      "loss": 0.6795,
+      "step": 1018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4454775993237534e-06,
+      "loss": 0.659,
+      "step": 1019
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4488588334742184e-06,
+      "loss": 0.6417,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.452240067624683e-06,
+      "loss": 0.7572,
+      "step": 1021
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4556213017751483e-06,
+      "loss": 0.6322,
+      "step": 1022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.459002535925613e-06,
+      "loss": 0.6275,
+      "step": 1023
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4623837700760783e-06,
+      "loss": 0.5968,
+      "step": 1024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.465765004226543e-06,
+      "loss": 0.5789,
+      "step": 1025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.469146238377008e-06,
+      "loss": 0.719,
+      "step": 1026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4725274725274727e-06,
+      "loss": 0.6197,
+      "step": 1027
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4759087066779377e-06,
+      "loss": 0.6773,
+      "step": 1028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4792899408284027e-06,
+      "loss": 0.6181,
+      "step": 1029
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4826711749788677e-06,
+      "loss": 0.6383,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.486052409129332e-06,
+      "loss": 0.6635,
+      "step": 1031
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4894336432797976e-06,
+      "loss": 0.5997,
+      "step": 1032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.492814877430262e-06,
+      "loss": 0.6812,
+      "step": 1033
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4961961115807275e-06,
+      "loss": 0.6545,
+      "step": 1034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.499577345731192e-06,
+      "loss": 0.6117,
+      "step": 1035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5029585798816575e-06,
+      "loss": 0.7843,
+      "step": 1036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.506339814032122e-06,
+      "loss": 0.7074,
+      "step": 1037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.509721048182587e-06,
+      "loss": 0.7514,
+      "step": 1038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.513102282333052e-06,
+      "loss": 0.6492,
+      "step": 1039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.516483516483517e-06,
+      "loss": 0.8227,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5198647506339815e-06,
+      "loss": 0.6342,
+      "step": 1041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5232459847844465e-06,
+      "loss": 0.608,
+      "step": 1042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5266272189349114e-06,
+      "loss": 0.6371,
+      "step": 1043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5300084530853764e-06,
+      "loss": 0.6959,
+      "step": 1044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5333896872358414e-06,
+      "loss": 0.6227,
+      "step": 1045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.536770921386306e-06,
+      "loss": 0.6345,
+      "step": 1046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5401521555367713e-06,
+      "loss": 0.622,
+      "step": 1047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.543533389687236e-06,
+      "loss": 0.5797,
+      "step": 1048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5469146238377013e-06,
+      "loss": 0.7905,
+      "step": 1049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.550295857988166e-06,
+      "loss": 0.6759,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5536770921386308e-06,
+      "loss": 0.6553,
+      "step": 1051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5570583262890958e-06,
+      "loss": 0.6205,
+      "step": 1052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5604395604395607e-06,
+      "loss": 0.6434,
+      "step": 1053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5638207945900257e-06,
+      "loss": 0.6045,
+      "step": 1054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5672020287404907e-06,
+      "loss": 0.6407,
+      "step": 1055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5705832628909552e-06,
+      "loss": 0.6648,
+      "step": 1056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5739644970414206e-06,
+      "loss": 0.6519,
+      "step": 1057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.577345731191885e-06,
+      "loss": 0.6322,
+      "step": 1058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5807269653423506e-06,
+      "loss": 0.8056,
+      "step": 1059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.584108199492815e-06,
+      "loss": 0.599,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.58748943364328e-06,
+      "loss": 0.6726,
+      "step": 1061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.590870667793745e-06,
+      "loss": 0.6619,
+      "step": 1062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.59425190194421e-06,
+      "loss": 0.6233,
+      "step": 1063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.597633136094675e-06,
+      "loss": 0.6481,
+      "step": 1064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.60101437024514e-06,
+      "loss": 0.6098,
+      "step": 1065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6043956043956045e-06,
+      "loss": 0.5662,
+      "step": 1066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.60777683854607e-06,
+      "loss": 0.643,
+      "step": 1067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6111580726965345e-06,
+      "loss": 0.756,
+      "step": 1068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.614539306847e-06,
+      "loss": 0.7128,
+      "step": 1069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6179205409974644e-06,
+      "loss": 0.6895,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.621301775147929e-06,
+      "loss": 0.6649,
+      "step": 1071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6246830092983943e-06,
+      "loss": 0.673,
+      "step": 1072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.628064243448859e-06,
+      "loss": 0.6627,
+      "step": 1073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6314454775993243e-06,
+      "loss": 0.712,
+      "step": 1074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.634826711749789e-06,
+      "loss": 0.6463,
+      "step": 1075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.638207945900254e-06,
+      "loss": 0.7168,
+      "step": 1076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6415891800507188e-06,
+      "loss": 0.7384,
+      "step": 1077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6449704142011837e-06,
+      "loss": 0.6184,
+      "step": 1078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6483516483516483e-06,
+      "loss": 0.6601,
+      "step": 1079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6517328825021137e-06,
+      "loss": 0.6729,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6551141166525782e-06,
+      "loss": 0.715,
+      "step": 1081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6584953508030436e-06,
+      "loss": 0.6748,
+      "step": 1082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.661876584953508e-06,
+      "loss": 0.6189,
+      "step": 1083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6652578191039736e-06,
+      "loss": 0.7026,
+      "step": 1084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.668639053254438e-06,
+      "loss": 0.7316,
+      "step": 1085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.672020287404903e-06,
+      "loss": 0.6609,
+      "step": 1086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.675401521555368e-06,
+      "loss": 0.5781,
+      "step": 1087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.678782755705833e-06,
+      "loss": 0.7089,
+      "step": 1088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6821639898562976e-06,
+      "loss": 0.6126,
+      "step": 1089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.685545224006763e-06,
+      "loss": 0.6683,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6889264581572275e-06,
+      "loss": 0.7136,
+      "step": 1091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.692307692307693e-06,
+      "loss": 0.5879,
+      "step": 1092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6956889264581575e-06,
+      "loss": 0.6577,
+      "step": 1093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.699070160608623e-06,
+      "loss": 0.643,
+      "step": 1094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7024513947590874e-06,
+      "loss": 0.671,
+      "step": 1095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7058326289095524e-06,
+      "loss": 0.636,
+      "step": 1096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7092138630600174e-06,
+      "loss": 0.6025,
+      "step": 1097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7125950972104823e-06,
+      "loss": 0.5935,
+      "step": 1098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.715976331360947e-06,
+      "loss": 0.6697,
+      "step": 1099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.719357565511412e-06,
+      "loss": 0.6391,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.722738799661877e-06,
+      "loss": 0.6861,
+      "step": 1101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7261200338123414e-06,
+      "loss": 0.5739,
+      "step": 1102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7295012679628068e-06,
+      "loss": 0.6584,
+      "step": 1103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7328825021132713e-06,
+      "loss": 0.6194,
+      "step": 1104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7362637362637367e-06,
+      "loss": 0.6004,
+      "step": 1105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7396449704142013e-06,
+      "loss": 0.6664,
+      "step": 1106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7430262045646666e-06,
+      "loss": 0.7241,
+      "step": 1107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.746407438715131e-06,
+      "loss": 0.6691,
+      "step": 1108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.749788672865596e-06,
+      "loss": 0.6303,
+      "step": 1109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.753169907016061e-06,
+      "loss": 0.6651,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.756551141166526e-06,
+      "loss": 0.5751,
+      "step": 1111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7599323753169907e-06,
+      "loss": 0.649,
+      "step": 1112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.763313609467456e-06,
+      "loss": 0.653,
+      "step": 1113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7666948436179206e-06,
+      "loss": 0.6475,
+      "step": 1114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.770076077768386e-06,
+      "loss": 0.6573,
+      "step": 1115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7734573119188505e-06,
+      "loss": 0.6343,
+      "step": 1116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.776838546069316e-06,
+      "loss": 0.6408,
+      "step": 1117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7802197802197805e-06,
+      "loss": 0.6813,
+      "step": 1118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7836010143702455e-06,
+      "loss": 0.6556,
+      "step": 1119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7869822485207104e-06,
+      "loss": 0.7072,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7903634826711754e-06,
+      "loss": 0.7031,
+      "step": 1121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.79374471682164e-06,
+      "loss": 0.6385,
+      "step": 1122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7971259509721053e-06,
+      "loss": 0.692,
+      "step": 1123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.80050718512257e-06,
+      "loss": 0.6475,
+      "step": 1124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8038884192730353e-06,
+      "loss": 0.6852,
+      "step": 1125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8072696534235e-06,
+      "loss": 0.684,
+      "step": 1126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8106508875739652e-06,
+      "loss": 0.7024,
+      "step": 1127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8140321217244298e-06,
+      "loss": 0.647,
+      "step": 1128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.817413355874895e-06,
+      "loss": 0.7157,
+      "step": 1129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.820794590025359e-06,
+      "loss": 0.6574,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.824175824175824e-06,
+      "loss": 0.6413,
+      "step": 1131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.827557058326289e-06,
+      "loss": 0.6217,
+      "step": 1132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.830938292476754e-06,
+      "loss": 0.6425,
+      "step": 1133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.834319526627219e-06,
+      "loss": 0.5628,
+      "step": 1134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.837700760777684e-06,
+      "loss": 0.5563,
+      "step": 1135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.841081994928149e-06,
+      "loss": 0.653,
+      "step": 1136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.844463229078614e-06,
+      "loss": 0.6728,
+      "step": 1137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.847844463229079e-06,
+      "loss": 0.5265,
+      "step": 1138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.851225697379544e-06,
+      "loss": 0.6944,
+      "step": 1139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.854606931530009e-06,
+      "loss": 0.6481,
+      "step": 1140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8579881656804736e-06,
+      "loss": 0.703,
+      "step": 1141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.861369399830939e-06,
+      "loss": 0.751,
+      "step": 1142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8647506339814035e-06,
+      "loss": 0.7499,
+      "step": 1143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.868131868131869e-06,
+      "loss": 0.7542,
+      "step": 1144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8715131022823334e-06,
+      "loss": 0.7598,
+      "step": 1145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.874894336432799e-06,
+      "loss": 0.6852,
+      "step": 1146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.878275570583263e-06,
+      "loss": 0.6097,
+      "step": 1147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.881656804733728e-06,
+      "loss": 0.6319,
+      "step": 1148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.885038038884193e-06,
+      "loss": 0.661,
+      "step": 1149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.888419273034658e-06,
+      "loss": 0.6181,
+      "step": 1150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8918005071851224e-06,
+      "loss": 0.6239,
+      "step": 1151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.895181741335588e-06,
+      "loss": 0.6184,
+      "step": 1152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.898562975486052e-06,
+      "loss": 0.6276,
+      "step": 1153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.901944209636518e-06,
+      "loss": 0.6735,
+      "step": 1154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.905325443786982e-06,
+      "loss": 0.7429,
+      "step": 1155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908706677937448e-06,
+      "loss": 0.5925,
+      "step": 1156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912087912087912e-06,
+      "loss": 0.5907,
+      "step": 1157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915469146238378e-06,
+      "loss": 0.6859,
+      "step": 1158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918850380388842e-06,
+      "loss": 0.6114,
+      "step": 1159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922231614539307e-06,
+      "loss": 0.6882,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.925612848689772e-06,
+      "loss": 0.6332,
+      "step": 1161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.928994082840237e-06,
+      "loss": 0.6265,
+      "step": 1162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.932375316990702e-06,
+      "loss": 0.5956,
+      "step": 1163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.935756551141167e-06,
+      "loss": 0.657,
+      "step": 1164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.939137785291632e-06,
+      "loss": 0.6657,
+      "step": 1165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9425190194420966e-06,
+      "loss": 0.7517,
+      "step": 1166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.945900253592562e-06,
+      "loss": 0.6071,
+      "step": 1167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9492814877430265e-06,
+      "loss": 0.5771,
+      "step": 1168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.952662721893492e-06,
+      "loss": 0.6442,
+      "step": 1169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9560439560439565e-06,
+      "loss": 0.6146,
+      "step": 1170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.959425190194421e-06,
+      "loss": 0.6416,
+      "step": 1171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.962806424344886e-06,
+      "loss": 0.6331,
+      "step": 1172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.966187658495351e-06,
+      "loss": 0.6771,
+      "step": 1173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9695688926458155e-06,
+      "loss": 0.6337,
+      "step": 1174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.972950126796281e-06,
+      "loss": 0.6242,
+      "step": 1175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9763313609467454e-06,
+      "loss": 0.762,
+      "step": 1176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.979712595097211e-06,
+      "loss": 0.6501,
+      "step": 1177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.983093829247675e-06,
+      "loss": 0.6164,
+      "step": 1178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.986475063398141e-06,
+      "loss": 0.6047,
+      "step": 1179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.989856297548605e-06,
+      "loss": 0.6359,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.993237531699071e-06,
+      "loss": 0.6941,
+      "step": 1181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.996618765849535e-06,
+      "loss": 0.5931,
+      "step": 1182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.6784,
+      "step": 1183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.003381234150465e-06,
+      "loss": 0.5991,
+      "step": 1184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.006762468300931e-06,
+      "loss": 0.7158,
+      "step": 1185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.010143702451395e-06,
+      "loss": 0.6322,
+      "step": 1186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0135249366018606e-06,
+      "loss": 0.7099,
+      "step": 1187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.016906170752325e-06,
+      "loss": 0.6504,
+      "step": 1188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.02028740490279e-06,
+      "loss": 0.6111,
+      "step": 1189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.023668639053255e-06,
+      "loss": 0.6253,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.02704987320372e-06,
+      "loss": 0.6253,
+      "step": 1191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.030431107354185e-06,
+      "loss": 0.6329,
+      "step": 1192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0338123415046495e-06,
+      "loss": 0.5794,
+      "step": 1193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.037193575655114e-06,
+      "loss": 0.6448,
+      "step": 1194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0405748098055795e-06,
+      "loss": 0.6262,
+      "step": 1195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.043956043956044e-06,
+      "loss": 0.6413,
+      "step": 1196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.047337278106509e-06,
+      "loss": 0.6216,
+      "step": 1197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.050718512256974e-06,
+      "loss": 0.6438,
+      "step": 1198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0540997464074385e-06,
+      "loss": 0.7409,
+      "step": 1199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.057480980557904e-06,
+      "loss": 0.6087,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0608622147083685e-06,
+      "loss": 0.6984,
+      "step": 1201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.064243448858834e-06,
+      "loss": 0.5932,
+      "step": 1202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.067624683009298e-06,
+      "loss": 0.5904,
+      "step": 1203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.071005917159764e-06,
+      "loss": 0.6392,
+      "step": 1204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.074387151310228e-06,
+      "loss": 0.6552,
+      "step": 1205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.077768385460694e-06,
+      "loss": 0.6193,
+      "step": 1206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.081149619611158e-06,
+      "loss": 0.7799,
+      "step": 1207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.084530853761624e-06,
+      "loss": 0.7823,
+      "step": 1208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.087912087912088e-06,
+      "loss": 0.6288,
+      "step": 1209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.091293322062554e-06,
+      "loss": 0.6568,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.094674556213018e-06,
+      "loss": 0.6414,
+      "step": 1211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0980557903634836e-06,
+      "loss": 0.6269,
+      "step": 1212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.101437024513948e-06,
+      "loss": 0.6786,
+      "step": 1213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.104818258664413e-06,
+      "loss": 0.6322,
+      "step": 1214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.108199492814878e-06,
+      "loss": 0.6073,
+      "step": 1215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.111580726965343e-06,
+      "loss": 0.7428,
+      "step": 1216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.114961961115808e-06,
+      "loss": 0.6075,
+      "step": 1217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1183431952662725e-06,
+      "loss": 0.6746,
+      "step": 1218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.121724429416737e-06,
+      "loss": 0.749,
+      "step": 1219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1251056635672025e-06,
+      "loss": 0.588,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.128486897717667e-06,
+      "loss": 0.6403,
+      "step": 1221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.131868131868132e-06,
+      "loss": 0.6101,
+      "step": 1222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.135249366018597e-06,
+      "loss": 0.701,
+      "step": 1223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1386306001690615e-06,
+      "loss": 0.6562,
+      "step": 1224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.142011834319527e-06,
+      "loss": 0.6441,
+      "step": 1225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1453930684699915e-06,
+      "loss": 0.6126,
+      "step": 1226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.148774302620457e-06,
+      "loss": 0.6741,
+      "step": 1227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.152155536770921e-06,
+      "loss": 0.6097,
+      "step": 1228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.155536770921387e-06,
+      "loss": 0.7086,
+      "step": 1229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.158918005071851e-06,
+      "loss": 0.595,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.162299239222317e-06,
+      "loss": 0.5943,
+      "step": 1231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.165680473372781e-06,
+      "loss": 0.6102,
+      "step": 1232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.169061707523247e-06,
+      "loss": 0.7018,
+      "step": 1233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.172442941673711e-06,
+      "loss": 0.7524,
+      "step": 1234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.175824175824177e-06,
+      "loss": 0.6281,
+      "step": 1235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.179205409974641e-06,
+      "loss": 0.7056,
+      "step": 1236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.182586644125107e-06,
+      "loss": 0.7101,
+      "step": 1237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.185967878275571e-06,
+      "loss": 0.6227,
+      "step": 1238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.189349112426036e-06,
+      "loss": 0.5595,
+      "step": 1239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.192730346576501e-06,
+      "loss": 0.5929,
+      "step": 1240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.196111580726966e-06,
+      "loss": 0.663,
+      "step": 1241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.19949281487743e-06,
+      "loss": 0.5868,
+      "step": 1242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2028740490278956e-06,
+      "loss": 0.6544,
+      "step": 1243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.20625528317836e-06,
+      "loss": 0.6399,
+      "step": 1244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2096365173288255e-06,
+      "loss": 0.652,
+      "step": 1245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.21301775147929e-06,
+      "loss": 0.6817,
+      "step": 1246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2163989856297555e-06,
+      "loss": 0.5841,
+      "step": 1247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.21978021978022e-06,
+      "loss": 0.595,
+      "step": 1248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2231614539306845e-06,
+      "loss": 0.5709,
+      "step": 1249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.22654268808115e-06,
+      "loss": 0.5828,
+      "step": 1250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2299239222316145e-06,
+      "loss": 0.7815,
+      "step": 1251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.23330515638208e-06,
+      "loss": 0.6196,
+      "step": 1252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2366863905325444e-06,
+      "loss": 0.7131,
+      "step": 1253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.24006762468301e-06,
+      "loss": 0.6651,
+      "step": 1254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.243448858833474e-06,
+      "loss": 0.6833,
+      "step": 1255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.24683009298394e-06,
+      "loss": 0.6604,
+      "step": 1256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.250211327134404e-06,
+      "loss": 0.6554,
+      "step": 1257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.25359256128487e-06,
+      "loss": 0.5592,
+      "step": 1258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.256973795435334e-06,
+      "loss": 0.5931,
+      "step": 1259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2603550295858e-06,
+      "loss": 0.6131,
+      "step": 1260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.263736263736264e-06,
+      "loss": 0.6863,
+      "step": 1261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.267117497886729e-06,
+      "loss": 0.642,
+      "step": 1262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.270498732037194e-06,
+      "loss": 0.6344,
+      "step": 1263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.273879966187659e-06,
+      "loss": 0.6568,
+      "step": 1264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.277261200338123e-06,
+      "loss": 0.7444,
+      "step": 1265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.280642434488589e-06,
+      "loss": 0.7241,
+      "step": 1266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.284023668639053e-06,
+      "loss": 0.546,
+      "step": 1267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.287404902789519e-06,
+      "loss": 0.6998,
+      "step": 1268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.290786136939983e-06,
+      "loss": 0.5608,
+      "step": 1269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2941673710904485e-06,
+      "loss": 0.6236,
+      "step": 1270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.297548605240913e-06,
+      "loss": 0.6194,
+      "step": 1271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3009298393913785e-06,
+      "loss": 0.7186,
+      "step": 1272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.304311073541843e-06,
+      "loss": 0.6509,
+      "step": 1273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.307692307692308e-06,
+      "loss": 0.6599,
+      "step": 1274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.311073541842773e-06,
+      "loss": 0.6086,
+      "step": 1275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.314454775993238e-06,
+      "loss": 0.5864,
+      "step": 1276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.317836010143703e-06,
+      "loss": 0.6314,
+      "step": 1277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.321217244294168e-06,
+      "loss": 0.6123,
+      "step": 1278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.324598478444633e-06,
+      "loss": 0.7127,
+      "step": 1279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.327979712595097e-06,
+      "loss": 0.5206,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.331360946745563e-06,
+      "loss": 0.5546,
+      "step": 1281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.334742180896027e-06,
+      "loss": 0.62,
+      "step": 1282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.338123415046493e-06,
+      "loss": 0.6482,
+      "step": 1283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.341504649196957e-06,
+      "loss": 0.6938,
+      "step": 1284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.344885883347422e-06,
+      "loss": 0.5864,
+      "step": 1285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.348267117497887e-06,
+      "loss": 0.6234,
+      "step": 1286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.351648351648352e-06,
+      "loss": 0.6999,
+      "step": 1287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.355029585798816e-06,
+      "loss": 0.6667,
+      "step": 1288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.358410819949282e-06,
+      "loss": 0.6573,
+      "step": 1289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.361792054099746e-06,
+      "loss": 0.6747,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.365173288250212e-06,
+      "loss": 0.5982,
+      "step": 1291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.368554522400676e-06,
+      "loss": 0.6078,
+      "step": 1292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.371935756551142e-06,
+      "loss": 0.6389,
+      "step": 1293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.375316990701606e-06,
+      "loss": 0.6655,
+      "step": 1294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3786982248520715e-06,
+      "loss": 0.6175,
+      "step": 1295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.382079459002536e-06,
+      "loss": 0.6225,
+      "step": 1296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3854606931530015e-06,
+      "loss": 0.6277,
+      "step": 1297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.388841927303466e-06,
+      "loss": 0.7246,
+      "step": 1298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3922231614539314e-06,
+      "loss": 0.6956,
+      "step": 1299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.395604395604396e-06,
+      "loss": 0.689,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.398985629754861e-06,
+      "loss": 0.6081,
+      "step": 1301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.402366863905326e-06,
+      "loss": 0.6021,
+      "step": 1302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.405748098055791e-06,
+      "loss": 0.5589,
+      "step": 1303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.409129332206256e-06,
+      "loss": 0.6228,
+      "step": 1304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.41251056635672e-06,
+      "loss": 0.6982,
+      "step": 1305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.415891800507186e-06,
+      "loss": 0.6345,
+      "step": 1306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.41927303465765e-06,
+      "loss": 0.5686,
+      "step": 1307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.422654268808116e-06,
+      "loss": 0.6147,
+      "step": 1308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.42603550295858e-06,
+      "loss": 0.6336,
+      "step": 1309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.429416737109045e-06,
+      "loss": 0.6112,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.43279797125951e-06,
+      "loss": 0.6537,
+      "step": 1311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.436179205409975e-06,
+      "loss": 0.6371,
+      "step": 1312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.439560439560439e-06,
+      "loss": 0.676,
+      "step": 1313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.442941673710905e-06,
+      "loss": 0.6954,
+      "step": 1314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.446322907861369e-06,
+      "loss": 0.6007,
+      "step": 1315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.449704142011835e-06,
+      "loss": 0.5613,
+      "step": 1316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.453085376162299e-06,
+      "loss": 0.7026,
+      "step": 1317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.456466610312765e-06,
+      "loss": 0.6077,
+      "step": 1318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.459847844463229e-06,
+      "loss": 0.6956,
+      "step": 1319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4632290786136946e-06,
+      "loss": 0.7073,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.466610312764159e-06,
+      "loss": 0.7095,
+      "step": 1321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4699915469146245e-06,
+      "loss": 0.6394,
+      "step": 1322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.473372781065089e-06,
+      "loss": 0.7546,
+      "step": 1323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4767540152155544e-06,
+      "loss": 0.6577,
+      "step": 1324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.480135249366019e-06,
+      "loss": 0.6603,
+      "step": 1325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.483516483516484e-06,
+      "loss": 0.5489,
+      "step": 1326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.486897717666949e-06,
+      "loss": 0.5757,
+      "step": 1327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.490278951817414e-06,
+      "loss": 0.6506,
+      "step": 1328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.493660185967879e-06,
+      "loss": 0.6329,
+      "step": 1329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.497041420118343e-06,
+      "loss": 0.6554,
+      "step": 1330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.500422654268809e-06,
+      "loss": 0.6409,
+      "step": 1331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.503803888419273e-06,
+      "loss": 0.6304,
+      "step": 1332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.507185122569738e-06,
+      "loss": 0.6257,
+      "step": 1333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.510566356720203e-06,
+      "loss": 0.6193,
+      "step": 1334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.513947590870668e-06,
+      "loss": 0.7251,
+      "step": 1335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.517328825021133e-06,
+      "loss": 0.6582,
+      "step": 1336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.520710059171598e-06,
+      "loss": 0.6808,
+      "step": 1337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.524091293322062e-06,
+      "loss": 0.593,
+      "step": 1338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.527472527472528e-06,
+      "loss": 0.6509,
+      "step": 1339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.530853761622992e-06,
+      "loss": 0.595,
+      "step": 1340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.534234995773458e-06,
+      "loss": 0.5925,
+      "step": 1341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.537616229923922e-06,
+      "loss": 0.6502,
+      "step": 1342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.540997464074388e-06,
+      "loss": 0.7177,
+      "step": 1343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.544378698224852e-06,
+      "loss": 0.6702,
+      "step": 1344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5477599323753176e-06,
+      "loss": 0.63,
+      "step": 1345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.551141166525782e-06,
+      "loss": 0.5475,
+      "step": 1346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5545224006762475e-06,
+      "loss": 0.5816,
+      "step": 1347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.557903634826712e-06,
+      "loss": 0.6352,
+      "step": 1348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5612848689771775e-06,
+      "loss": 0.6491,
+      "step": 1349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.564666103127642e-06,
+      "loss": 0.5763,
+      "step": 1350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.568047337278107e-06,
+      "loss": 0.7149,
+      "step": 1351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 0.5677,
+      "step": 1352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5748098055790365e-06,
+      "loss": 0.6388,
+      "step": 1353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.578191039729502e-06,
+      "loss": 0.613,
+      "step": 1354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5815722738799664e-06,
+      "loss": 0.5913,
+      "step": 1355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.584953508030431e-06,
+      "loss": 0.6517,
+      "step": 1356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.588334742180896e-06,
+      "loss": 0.6535,
+      "step": 1357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.591715976331361e-06,
+      "loss": 0.6707,
+      "step": 1358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.595097210481826e-06,
+      "loss": 0.6224,
+      "step": 1359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.598478444632291e-06,
+      "loss": 0.6831,
+      "step": 1360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.601859678782756e-06,
+      "loss": 0.7149,
+      "step": 1361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.605240912933221e-06,
+      "loss": 0.5961,
+      "step": 1362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.608622147083686e-06,
+      "loss": 0.6076,
+      "step": 1363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.612003381234151e-06,
+      "loss": 0.609,
+      "step": 1364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.6572,
+      "step": 1365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.618765849535081e-06,
+      "loss": 0.6046,
+      "step": 1366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.622147083685546e-06,
+      "loss": 0.7084,
+      "step": 1367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.625528317836011e-06,
+      "loss": 0.6625,
+      "step": 1368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.628909551986475e-06,
+      "loss": 0.5918,
+      "step": 1369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.632290786136941e-06,
+      "loss": 0.6098,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.635672020287405e-06,
+      "loss": 0.5838,
+      "step": 1371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6390532544378705e-06,
+      "loss": 0.6616,
+      "step": 1372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.642434488588335e-06,
+      "loss": 0.6551,
+      "step": 1373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6458157227388005e-06,
+      "loss": 0.6107,
+      "step": 1374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.649196956889265e-06,
+      "loss": 0.6297,
+      "step": 1375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6525781910397296e-06,
+      "loss": 0.5532,
+      "step": 1376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.655959425190195e-06,
+      "loss": 0.6524,
+      "step": 1377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6593406593406595e-06,
+      "loss": 0.6476,
+      "step": 1378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.662721893491124e-06,
+      "loss": 0.6599,
+      "step": 1379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6661031276415895e-06,
+      "loss": 0.604,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.669484361792054e-06,
+      "loss": 0.6032,
+      "step": 1381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.672865595942519e-06,
+      "loss": 0.605,
+      "step": 1382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.676246830092984e-06,
+      "loss": 0.6348,
+      "step": 1383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.679628064243449e-06,
+      "loss": 0.5905,
+      "step": 1384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.683009298393914e-06,
+      "loss": 0.6494,
+      "step": 1385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.686390532544379e-06,
+      "loss": 0.5405,
+      "step": 1386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.689771766694844e-06,
+      "loss": 0.675,
+      "step": 1387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.693153000845309e-06,
+      "loss": 0.6394,
+      "step": 1388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.696534234995774e-06,
+      "loss": 0.6676,
+      "step": 1389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.699915469146239e-06,
+      "loss": 0.6126,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.703296703296704e-06,
+      "loss": 0.629,
+      "step": 1391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.706677937447169e-06,
+      "loss": 0.6977,
+      "step": 1392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.710059171597634e-06,
+      "loss": 0.6696,
+      "step": 1393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.713440405748099e-06,
+      "loss": 0.5939,
+      "step": 1394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.716821639898564e-06,
+      "loss": 0.6334,
+      "step": 1395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.720202874049028e-06,
+      "loss": 0.6455,
+      "step": 1396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7235841081994935e-06,
+      "loss": 0.5377,
+      "step": 1397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.726965342349958e-06,
+      "loss": 0.6305,
+      "step": 1398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.730346576500423e-06,
+      "loss": 0.6665,
+      "step": 1399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.733727810650888e-06,
+      "loss": 0.6215,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.737109044801353e-06,
+      "loss": 0.6056,
+      "step": 1401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.740490278951818e-06,
+      "loss": 0.7001,
+      "step": 1402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7438715131022825e-06,
+      "loss": 0.5965,
+      "step": 1403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.747252747252747e-06,
+      "loss": 0.5882,
+      "step": 1404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7506339814032125e-06,
+      "loss": 0.5542,
+      "step": 1405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.754015215553677e-06,
+      "loss": 0.5891,
+      "step": 1406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.757396449704142e-06,
+      "loss": 0.6133,
+      "step": 1407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.760777683854607e-06,
+      "loss": 0.7081,
+      "step": 1408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764158918005072e-06,
+      "loss": 0.6225,
+      "step": 1409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767540152155537e-06,
+      "loss": 0.7168,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770921386306002e-06,
+      "loss": 0.5955,
+      "step": 1411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774302620456467e-06,
+      "loss": 0.5793,
+      "step": 1412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777683854606932e-06,
+      "loss": 0.6066,
+      "step": 1413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.781065088757397e-06,
+      "loss": 0.6805,
+      "step": 1414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.784446322907862e-06,
+      "loss": 0.6802,
+      "step": 1415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.787827557058327e-06,
+      "loss": 0.5793,
+      "step": 1416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.791208791208792e-06,
+      "loss": 0.6612,
+      "step": 1417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.794590025359257e-06,
+      "loss": 0.5774,
+      "step": 1418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.797971259509721e-06,
+      "loss": 0.6785,
+      "step": 1419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.801352493660187e-06,
+      "loss": 0.6491,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.804733727810651e-06,
+      "loss": 0.6626,
+      "step": 1421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8081149619611166e-06,
+      "loss": 0.6464,
+      "step": 1422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.811496196111581e-06,
+      "loss": 0.5536,
+      "step": 1423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.814877430262046e-06,
+      "loss": 0.5655,
+      "step": 1424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.818258664412511e-06,
+      "loss": 0.7282,
+      "step": 1425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.821639898562976e-06,
+      "loss": 0.5875,
+      "step": 1426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.82502113271344e-06,
+      "loss": 0.6515,
+      "step": 1427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8284023668639055e-06,
+      "loss": 0.6163,
+      "step": 1428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.83178360101437e-06,
+      "loss": 0.603,
+      "step": 1429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8351648351648355e-06,
+      "loss": 0.6422,
+      "step": 1430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8385460693153e-06,
+      "loss": 0.6095,
+      "step": 1431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8419273034657654e-06,
+      "loss": 0.6085,
+      "step": 1432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.84530853761623e-06,
+      "loss": 0.7064,
+      "step": 1433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.848689771766695e-06,
+      "loss": 0.5367,
+      "step": 1434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.85207100591716e-06,
+      "loss": 0.6077,
+      "step": 1435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.855452240067625e-06,
+      "loss": 0.63,
+      "step": 1436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.85883347421809e-06,
+      "loss": 0.6554,
+      "step": 1437
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.862214708368555e-06,
+      "loss": 0.656,
+      "step": 1438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.86559594251902e-06,
+      "loss": 0.6404,
+      "step": 1439
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.868977176669485e-06,
+      "loss": 0.5996,
+      "step": 1440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.87235841081995e-06,
+      "loss": 0.5821,
+      "step": 1441
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.875739644970415e-06,
+      "loss": 0.5541,
+      "step": 1442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.87912087912088e-06,
+      "loss": 0.5711,
+      "step": 1443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.882502113271344e-06,
+      "loss": 0.5783,
+      "step": 1444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.88588334742181e-06,
+      "loss": 0.5853,
+      "step": 1445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.889264581572274e-06,
+      "loss": 0.6161,
+      "step": 1446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.892645815722739e-06,
+      "loss": 0.6538,
+      "step": 1447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.896027049873204e-06,
+      "loss": 0.6726,
+      "step": 1448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.899408284023669e-06,
+      "loss": 0.6095,
+      "step": 1449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.902789518174134e-06,
+      "loss": 0.7126,
+      "step": 1450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.906170752324599e-06,
+      "loss": 0.6326,
+      "step": 1451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909551986475064e-06,
+      "loss": 0.6804,
+      "step": 1452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9129332206255286e-06,
+      "loss": 0.5357,
+      "step": 1453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.916314454775994e-06,
+      "loss": 0.7505,
+      "step": 1454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9196956889264585e-06,
+      "loss": 0.6251,
+      "step": 1455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.923076923076924e-06,
+      "loss": 0.6289,
+      "step": 1456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9264581572273884e-06,
+      "loss": 0.631,
+      "step": 1457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929839391377853e-06,
+      "loss": 0.5918,
+      "step": 1458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.933220625528318e-06,
+      "loss": 0.6466,
+      "step": 1459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.936601859678783e-06,
+      "loss": 0.6119,
+      "step": 1460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.939983093829248e-06,
+      "loss": 0.5902,
+      "step": 1461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.943364327979713e-06,
+      "loss": 0.5358,
+      "step": 1462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.946745562130178e-06,
+      "loss": 0.571,
+      "step": 1463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.950126796280643e-06,
+      "loss": 0.6329,
+      "step": 1464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.953508030431108e-06,
+      "loss": 0.5697,
+      "step": 1465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.956889264581573e-06,
+      "loss": 0.6178,
+      "step": 1466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.960270498732037e-06,
+      "loss": 0.6522,
+      "step": 1467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.963651732882503e-06,
+      "loss": 0.631,
+      "step": 1468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967032967032967e-06,
+      "loss": 0.6887,
+      "step": 1469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.970414201183432e-06,
+      "loss": 0.7064,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.973795435333897e-06,
+      "loss": 0.5947,
+      "step": 1471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.977176669484362e-06,
+      "loss": 0.6001,
+      "step": 1472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.980557903634827e-06,
+      "loss": 0.6512,
+      "step": 1473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.983939137785292e-06,
+      "loss": 0.573,
+      "step": 1474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.987320371935757e-06,
+      "loss": 0.5742,
+      "step": 1475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.990701606086222e-06,
+      "loss": 0.6365,
+      "step": 1476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.994082840236687e-06,
+      "loss": 0.6448,
+      "step": 1477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9974640743871516e-06,
+      "loss": 0.6428,
+      "step": 1478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.000845308537616e-06,
+      "loss": 0.6566,
+      "step": 1479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.004226542688082e-06,
+      "loss": 0.6146,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.007607776838547e-06,
+      "loss": 0.6191,
+      "step": 1481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.0109890109890115e-06,
+      "loss": 0.6561,
+      "step": 1482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.014370245139476e-06,
+      "loss": 0.651,
+      "step": 1483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.017751479289941e-06,
+      "loss": 0.6358,
+      "step": 1484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.021132713440407e-06,
+      "loss": 0.6108,
+      "step": 1485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.024513947590871e-06,
+      "loss": 0.5257,
+      "step": 1486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.027895181741336e-06,
+      "loss": 0.6019,
+      "step": 1487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.0312764158918004e-06,
+      "loss": 0.6078,
+      "step": 1488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.034657650042266e-06,
+      "loss": 0.6178,
+      "step": 1489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.03803888419273e-06,
+      "loss": 0.6957,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.041420118343196e-06,
+      "loss": 0.6154,
+      "step": 1491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.04480135249366e-06,
+      "loss": 0.6059,
+      "step": 1492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.048182586644126e-06,
+      "loss": 0.7283,
+      "step": 1493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.05156382079459e-06,
+      "loss": 0.6332,
+      "step": 1494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.054945054945055e-06,
+      "loss": 0.6879,
+      "step": 1495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.05832628909552e-06,
+      "loss": 0.6639,
+      "step": 1496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.061707523245986e-06,
+      "loss": 0.6941,
+      "step": 1497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.06508875739645e-06,
+      "loss": 0.5976,
+      "step": 1498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.068469991546915e-06,
+      "loss": 0.6253,
+      "step": 1499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.071851225697379e-06,
+      "loss": 0.6147,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.0752324598478455e-06,
+      "loss": 0.6209,
+      "step": 1501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.07861369399831e-06,
+      "loss": 0.6512,
+      "step": 1502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.081994928148775e-06,
+      "loss": 0.6739,
+      "step": 1503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.085376162299239e-06,
+      "loss": 0.6043,
+      "step": 1504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.088757396449705e-06,
+      "loss": 0.6565,
+      "step": 1505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.09213863060017e-06,
+      "loss": 0.6154,
+      "step": 1506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.0955198647506345e-06,
+      "loss": 0.6712,
+      "step": 1507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.098901098901099e-06,
+      "loss": 0.5959,
+      "step": 1508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.102282333051564e-06,
+      "loss": 0.5645,
+      "step": 1509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.105663567202029e-06,
+      "loss": 0.6175,
+      "step": 1510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.109044801352494e-06,
+      "loss": 0.6405,
+      "step": 1511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.112426035502959e-06,
+      "loss": 0.7057,
+      "step": 1512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.115807269653424e-06,
+      "loss": 0.6639,
+      "step": 1513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.119188503803889e-06,
+      "loss": 0.6749,
+      "step": 1514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.122569737954353e-06,
+      "loss": 0.7111,
+      "step": 1515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.125950972104819e-06,
+      "loss": 0.6188,
+      "step": 1516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.129332206255283e-06,
+      "loss": 0.658,
+      "step": 1517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.132713440405749e-06,
+      "loss": 0.6618,
+      "step": 1518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.136094674556213e-06,
+      "loss": 0.6751,
+      "step": 1519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.139475908706678e-06,
+      "loss": 0.677,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 0.536,
+      "step": 1521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.146238377007609e-06,
+      "loss": 0.6743,
+      "step": 1522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.149619611158073e-06,
+      "loss": 0.673,
+      "step": 1523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.153000845308538e-06,
+      "loss": 0.598,
+      "step": 1524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.156382079459002e-06,
+      "loss": 0.6327,
+      "step": 1525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.1597633136094685e-06,
+      "loss": 0.6378,
+      "step": 1526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.163144547759933e-06,
+      "loss": 0.5536,
+      "step": 1527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.166525781910398e-06,
+      "loss": 0.6779,
+      "step": 1528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.169907016060862e-06,
+      "loss": 0.5618,
+      "step": 1529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.1732882502113275e-06,
+      "loss": 0.594,
+      "step": 1530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.176669484361793e-06,
+      "loss": 0.661,
+      "step": 1531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.1800507185122575e-06,
+      "loss": 0.6774,
+      "step": 1532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.183431952662722e-06,
+      "loss": 0.6414,
+      "step": 1533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.1868131868131874e-06,
+      "loss": 0.7229,
+      "step": 1534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.190194420963652e-06,
+      "loss": 0.5528,
+      "step": 1535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.193575655114117e-06,
+      "loss": 0.5911,
+      "step": 1536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.196956889264582e-06,
+      "loss": 0.634,
+      "step": 1537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.200338123415047e-06,
+      "loss": 0.5896,
+      "step": 1538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.203719357565512e-06,
+      "loss": 0.5409,
+      "step": 1539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.207100591715976e-06,
+      "loss": 0.6044,
+      "step": 1540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.210481825866441e-06,
+      "loss": 0.6022,
+      "step": 1541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.213863060016907e-06,
+      "loss": 0.6507,
+      "step": 1542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.217244294167372e-06,
+      "loss": 0.6401,
+      "step": 1543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.220625528317836e-06,
+      "loss": 0.6471,
+      "step": 1544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.224006762468301e-06,
+      "loss": 0.5824,
+      "step": 1545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.227387996618765e-06,
+      "loss": 0.6791,
+      "step": 1546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.230769230769232e-06,
+      "loss": 0.5593,
+      "step": 1547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.234150464919696e-06,
+      "loss": 0.6706,
+      "step": 1548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.237531699070161e-06,
+      "loss": 0.6699,
+      "step": 1549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.240912933220625e-06,
+      "loss": 0.7567,
+      "step": 1550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.2442941673710915e-06,
+      "loss": 0.7194,
+      "step": 1551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.247675401521556e-06,
+      "loss": 0.672,
+      "step": 1552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.251056635672021e-06,
+      "loss": 0.6671,
+      "step": 1553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.254437869822485e-06,
+      "loss": 0.6,
+      "step": 1554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.2578191039729506e-06,
+      "loss": 0.5877,
+      "step": 1555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.261200338123416e-06,
+      "loss": 0.5874,
+      "step": 1556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.2645815722738805e-06,
+      "loss": 0.7631,
+      "step": 1557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.267962806424345e-06,
+      "loss": 0.6221,
+      "step": 1558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.2713440405748105e-06,
+      "loss": 0.5766,
+      "step": 1559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.274725274725275e-06,
+      "loss": 0.728,
+      "step": 1560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.2781065088757395e-06,
+      "loss": 0.675,
+      "step": 1561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.281487743026205e-06,
+      "loss": 0.6573,
+      "step": 1562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.28486897717667e-06,
+      "loss": 0.5611,
+      "step": 1563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.288250211327135e-06,
+      "loss": 0.765,
+      "step": 1564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.2916314454775994e-06,
+      "loss": 0.6227,
+      "step": 1565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.295012679628064e-06,
+      "loss": 0.528,
+      "step": 1566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.29839391377853e-06,
+      "loss": 0.6519,
+      "step": 1567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.301775147928995e-06,
+      "loss": 0.6776,
+      "step": 1568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.305156382079459e-06,
+      "loss": 0.6223,
+      "step": 1569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.308537616229924e-06,
+      "loss": 0.5869,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.31191885038039e-06,
+      "loss": 0.5418,
+      "step": 1571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.315300084530855e-06,
+      "loss": 0.7038,
+      "step": 1572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.318681318681319e-06,
+      "loss": 0.6088,
+      "step": 1573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.322062552831784e-06,
+      "loss": 0.6452,
+      "step": 1574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.325443786982249e-06,
+      "loss": 0.7117,
+      "step": 1575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.3288250211327145e-06,
+      "loss": 0.6293,
+      "step": 1576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.332206255283179e-06,
+      "loss": 0.6466,
+      "step": 1577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.335587489433644e-06,
+      "loss": 0.5972,
+      "step": 1578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.338968723584108e-06,
+      "loss": 0.6115,
+      "step": 1579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.342349957734574e-06,
+      "loss": 0.5653,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.345731191885038e-06,
+      "loss": 0.5882,
+      "step": 1581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.3491124260355035e-06,
+      "loss": 0.5406,
+      "step": 1582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.352493660185968e-06,
+      "loss": 0.5659,
+      "step": 1583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.3558748943364335e-06,
+      "loss": 0.7085,
+      "step": 1584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.359256128486898e-06,
+      "loss": 0.6593,
+      "step": 1585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.3626373626373626e-06,
+      "loss": 0.735,
+      "step": 1586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.366018596787828e-06,
+      "loss": 0.6749,
+      "step": 1587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.369399830938293e-06,
+      "loss": 0.8156,
+      "step": 1588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.372781065088758e-06,
+      "loss": 0.6274,
+      "step": 1589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.3761622992392224e-06,
+      "loss": 0.7006,
+      "step": 1590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.379543533389687e-06,
+      "loss": 0.6136,
+      "step": 1591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.382924767540153e-06,
+      "loss": 0.6659,
+      "step": 1592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.386306001690618e-06,
+      "loss": 0.6309,
+      "step": 1593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.389687235841082e-06,
+      "loss": 0.5999,
+      "step": 1594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.393068469991547e-06,
+      "loss": 0.5833,
+      "step": 1595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.396449704142013e-06,
+      "loss": 0.6705,
+      "step": 1596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.399830938292478e-06,
+      "loss": 0.56,
+      "step": 1597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.403212172442942e-06,
+      "loss": 0.6038,
+      "step": 1598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.406593406593407e-06,
+      "loss": 0.5845,
+      "step": 1599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.409974640743872e-06,
+      "loss": 0.6399,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.413355874894337e-06,
+      "loss": 0.7292,
+      "step": 1601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.416737109044802e-06,
+      "loss": 0.5455,
+      "step": 1602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.420118343195267e-06,
+      "loss": 0.6417,
+      "step": 1603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.423499577345732e-06,
+      "loss": 0.6383,
+      "step": 1604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.426880811496197e-06,
+      "loss": 0.6148,
+      "step": 1605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.430262045646661e-06,
+      "loss": 0.6207,
+      "step": 1606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4336432797971265e-06,
+      "loss": 0.6796,
+      "step": 1607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.437024513947591e-06,
+      "loss": 0.6879,
+      "step": 1608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4404057480980565e-06,
+      "loss": 0.5945,
+      "step": 1609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.443786982248521e-06,
+      "loss": 0.6485,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.447168216398986e-06,
+      "loss": 0.6212,
+      "step": 1611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.45054945054945e-06,
+      "loss": 0.5914,
+      "step": 1612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.453930684699916e-06,
+      "loss": 0.6925,
+      "step": 1613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.457311918850381e-06,
+      "loss": 0.6941,
+      "step": 1614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4606931530008455e-06,
+      "loss": 0.5994,
+      "step": 1615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.46407438715131e-06,
+      "loss": 0.6375,
+      "step": 1616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.467455621301776e-06,
+      "loss": 0.7011,
+      "step": 1617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.470836855452241e-06,
+      "loss": 0.6272,
+      "step": 1618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.474218089602705e-06,
+      "loss": 0.6617,
+      "step": 1619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.47759932375317e-06,
+      "loss": 0.5753,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.480980557903635e-06,
+      "loss": 0.6315,
+      "step": 1621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.484361792054101e-06,
+      "loss": 0.6963,
+      "step": 1622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.487743026204565e-06,
+      "loss": 0.7065,
+      "step": 1623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.49112426035503e-06,
+      "loss": 0.6719,
+      "step": 1624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.494505494505495e-06,
+      "loss": 0.6154,
+      "step": 1625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.49788672865596e-06,
+      "loss": 0.7065,
+      "step": 1626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.501267962806425e-06,
+      "loss": 0.5761,
+      "step": 1627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.50464919695689e-06,
+      "loss": 0.6598,
+      "step": 1628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.508030431107355e-06,
+      "loss": 0.7129,
+      "step": 1629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.51141166525782e-06,
+      "loss": 0.7543,
+      "step": 1630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.514792899408284e-06,
+      "loss": 0.6567,
+      "step": 1631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.518174133558749e-06,
+      "loss": 0.612,
+      "step": 1632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.521555367709215e-06,
+      "loss": 0.6858,
+      "step": 1633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.5249366018596795e-06,
+      "loss": 0.5638,
+      "step": 1634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.528317836010144e-06,
+      "loss": 0.6598,
+      "step": 1635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.531699070160609e-06,
+      "loss": 0.663,
+      "step": 1636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.535080304311073e-06,
+      "loss": 0.6659,
+      "step": 1637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.538461538461539e-06,
+      "loss": 0.8334,
+      "step": 1638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.541842772612004e-06,
+      "loss": 0.7372,
+      "step": 1639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.5452240067624685e-06,
+      "loss": 0.6694,
+      "step": 1640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.548605240912933e-06,
+      "loss": 0.7003,
+      "step": 1641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.551986475063399e-06,
+      "loss": 0.6044,
+      "step": 1642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.555367709213864e-06,
+      "loss": 0.7048,
+      "step": 1643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.558748943364328e-06,
+      "loss": 0.5744,
+      "step": 1644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.562130177514793e-06,
+      "loss": 0.6104,
+      "step": 1645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.565511411665258e-06,
+      "loss": 0.7245,
+      "step": 1646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.568892645815724e-06,
+      "loss": 0.6104,
+      "step": 1647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.572273879966188e-06,
+      "loss": 0.6947,
+      "step": 1648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.575655114116653e-06,
+      "loss": 0.5866,
+      "step": 1649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.579036348267118e-06,
+      "loss": 0.6335,
+      "step": 1650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.582417582417583e-06,
+      "loss": 0.7089,
+      "step": 1651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.585798816568047e-06,
+      "loss": 0.6296,
+      "step": 1652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.589180050718513e-06,
+      "loss": 0.5619,
+      "step": 1653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.592561284868978e-06,
+      "loss": 0.6445,
+      "step": 1654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.595942519019443e-06,
+      "loss": 0.6342,
+      "step": 1655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.599323753169907e-06,
+      "loss": 0.6643,
+      "step": 1656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.602704987320372e-06,
+      "loss": 0.6427,
+      "step": 1657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.606086221470838e-06,
+      "loss": 0.5683,
+      "step": 1658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.6094674556213025e-06,
+      "loss": 0.6317,
+      "step": 1659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.612848689771767e-06,
+      "loss": 0.6488,
+      "step": 1660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.616229923922232e-06,
+      "loss": 0.6064,
+      "step": 1661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.619611158072698e-06,
+      "loss": 0.5104,
+      "step": 1662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.622992392223162e-06,
+      "loss": 0.6036,
+      "step": 1663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.626373626373627e-06,
+      "loss": 0.6302,
+      "step": 1664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.6297548605240915e-06,
+      "loss": 0.641,
+      "step": 1665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.633136094674556e-06,
+      "loss": 0.6156,
+      "step": 1666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.636517328825022e-06,
+      "loss": 0.5377,
+      "step": 1667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.639898562975487e-06,
+      "loss": 0.6959,
+      "step": 1668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.643279797125951e-06,
+      "loss": 0.5859,
+      "step": 1669
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.646661031276416e-06,
+      "loss": 0.6017,
+      "step": 1670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.650042265426881e-06,
+      "loss": 0.6402,
+      "step": 1671
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.653423499577346e-06,
+      "loss": 0.6199,
+      "step": 1672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.656804733727811e-06,
+      "loss": 0.743,
+      "step": 1673
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.660185967878276e-06,
+      "loss": 0.6304,
+      "step": 1674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.663567202028741e-06,
+      "loss": 0.6123,
+      "step": 1675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.666948436179206e-06,
+      "loss": 0.6201,
+      "step": 1676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.67032967032967e-06,
+      "loss": 0.7237,
+      "step": 1677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.673710904480136e-06,
+      "loss": 0.585,
+      "step": 1678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.677092138630601e-06,
+      "loss": 0.6436,
+      "step": 1679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.680473372781066e-06,
+      "loss": 0.6769,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.68385460693153e-06,
+      "loss": 0.7305,
+      "step": 1681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.687235841081995e-06,
+      "loss": 0.7683,
+      "step": 1682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.690617075232461e-06,
+      "loss": 0.6539,
+      "step": 1683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.6939983093829255e-06,
+      "loss": 0.5366,
+      "step": 1684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.69737954353339e-06,
+      "loss": 0.6044,
+      "step": 1685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.700760777683855e-06,
+      "loss": 0.6418,
+      "step": 1686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.704142011834321e-06,
+      "loss": 0.6186,
+      "step": 1687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.707523245984785e-06,
+      "loss": 0.6726,
+      "step": 1688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.71090448013525e-06,
+      "loss": 0.6538,
+      "step": 1689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.5316,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.71766694843618e-06,
+      "loss": 0.6503,
+      "step": 1691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.7210481825866445e-06,
+      "loss": 0.5915,
+      "step": 1692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.72442941673711e-06,
+      "loss": 0.5731,
+      "step": 1693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.727810650887574e-06,
+      "loss": 0.7033,
+      "step": 1694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.731191885038039e-06,
+      "loss": 0.5905,
+      "step": 1695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.734573119188504e-06,
+      "loss": 0.7415,
+      "step": 1696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.737954353338969e-06,
+      "loss": 0.6135,
+      "step": 1697
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.741335587489434e-06,
+      "loss": 0.6278,
+      "step": 1698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.744716821639899e-06,
+      "loss": 0.7824,
+      "step": 1699
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.748098055790364e-06,
+      "loss": 0.624,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.751479289940829e-06,
+      "loss": 0.6115,
+      "step": 1701
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.754860524091293e-06,
+      "loss": 0.6937,
+      "step": 1702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.758241758241758e-06,
+      "loss": 0.6234,
+      "step": 1703
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.761622992392224e-06,
+      "loss": 0.6846,
+      "step": 1704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.765004226542689e-06,
+      "loss": 0.6798,
+      "step": 1705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.768385460693153e-06,
+      "loss": 0.6255,
+      "step": 1706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.771766694843618e-06,
+      "loss": 0.6471,
+      "step": 1707
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.775147928994084e-06,
+      "loss": 0.6522,
+      "step": 1708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.7785291631445485e-06,
+      "loss": 0.6409,
+      "step": 1709
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.781910397295013e-06,
+      "loss": 0.6682,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.785291631445478e-06,
+      "loss": 0.5923,
+      "step": 1711
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.788672865595943e-06,
+      "loss": 0.6444,
+      "step": 1712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.7920540997464084e-06,
+      "loss": 0.6475,
+      "step": 1713
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.795435333896873e-06,
+      "loss": 0.5798,
+      "step": 1714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.7988165680473375e-06,
+      "loss": 0.5491,
+      "step": 1715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.802197802197803e-06,
+      "loss": 0.613,
+      "step": 1716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.8055790363482675e-06,
+      "loss": 0.6061,
+      "step": 1717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.808960270498733e-06,
+      "loss": 0.5956,
+      "step": 1718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.812341504649197e-06,
+      "loss": 0.7173,
+      "step": 1719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.815722738799663e-06,
+      "loss": 0.7137,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.819103972950127e-06,
+      "loss": 0.6046,
+      "step": 1721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.822485207100592e-06,
+      "loss": 0.5604,
+      "step": 1722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.8258664412510565e-06,
+      "loss": 0.554,
+      "step": 1723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.829247675401522e-06,
+      "loss": 0.5616,
+      "step": 1724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.832628909551987e-06,
+      "loss": 0.6528,
+      "step": 1725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.836010143702452e-06,
+      "loss": 0.5657,
+      "step": 1726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.839391377852916e-06,
+      "loss": 0.6477,
+      "step": 1727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.842772612003381e-06,
+      "loss": 0.6454,
+      "step": 1728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.846153846153847e-06,
+      "loss": 0.5433,
+      "step": 1729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.849535080304312e-06,
+      "loss": 0.5277,
+      "step": 1730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.852916314454776e-06,
+      "loss": 0.598,
+      "step": 1731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.856297548605241e-06,
+      "loss": 0.6083,
+      "step": 1732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.859678782755707e-06,
+      "loss": 0.5513,
+      "step": 1733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8630600169061716e-06,
+      "loss": 0.6639,
+      "step": 1734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.866441251056636e-06,
+      "loss": 0.5743,
+      "step": 1735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.869822485207101e-06,
+      "loss": 0.7499,
+      "step": 1736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.873203719357566e-06,
+      "loss": 0.5983,
+      "step": 1737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8765849535080314e-06,
+      "loss": 0.6096,
+      "step": 1738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.879966187658496e-06,
+      "loss": 0.6822,
+      "step": 1739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8833474218089605e-06,
+      "loss": 0.5815,
+      "step": 1740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.886728655959426e-06,
+      "loss": 0.6882,
+      "step": 1741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8901098901098905e-06,
+      "loss": 0.6939,
+      "step": 1742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.893491124260355e-06,
+      "loss": 0.6491,
+      "step": 1743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8968723584108204e-06,
+      "loss": 0.5978,
+      "step": 1744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.900253592561286e-06,
+      "loss": 0.7046,
+      "step": 1745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.90363482671175e-06,
+      "loss": 0.6151,
+      "step": 1746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.907016060862215e-06,
+      "loss": 0.608,
+      "step": 1747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9103972950126795e-06,
+      "loss": 0.6889,
+      "step": 1748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.913778529163146e-06,
+      "loss": 0.5841,
+      "step": 1749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.91715976331361e-06,
+      "loss": 0.6237,
+      "step": 1750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.920540997464075e-06,
+      "loss": 0.6509,
+      "step": 1751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.923922231614539e-06,
+      "loss": 0.5536,
+      "step": 1752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.927303465765006e-06,
+      "loss": 0.8218,
+      "step": 1753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.93068469991547e-06,
+      "loss": 0.6203,
+      "step": 1754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.934065934065935e-06,
+      "loss": 0.5965,
+      "step": 1755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.937447168216399e-06,
+      "loss": 0.6061,
+      "step": 1756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.940828402366864e-06,
+      "loss": 0.742,
+      "step": 1757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.94420963651733e-06,
+      "loss": 0.6596,
+      "step": 1758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.947590870667795e-06,
+      "loss": 0.756,
+      "step": 1759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.950972104818259e-06,
+      "loss": 0.5815,
+      "step": 1760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.954353338968724e-06,
+      "loss": 0.6171,
+      "step": 1761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.957734573119189e-06,
+      "loss": 0.6825,
+      "step": 1762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.961115807269654e-06,
+      "loss": 0.6034,
+      "step": 1763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.964497041420119e-06,
+      "loss": 0.5473,
+      "step": 1764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9678782755705836e-06,
+      "loss": 0.6602,
+      "step": 1765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971259509721049e-06,
+      "loss": 0.6882,
+      "step": 1766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9746407438715135e-06,
+      "loss": 0.5596,
+      "step": 1767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.978021978021978e-06,
+      "loss": 0.6526,
+      "step": 1768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.981403212172443e-06,
+      "loss": 0.692,
+      "step": 1769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.984784446322909e-06,
+      "loss": 0.6286,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.988165680473373e-06,
+      "loss": 0.5973,
+      "step": 1771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.991546914623838e-06,
+      "loss": 0.7146,
+      "step": 1772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9949281487743025e-06,
+      "loss": 0.7044,
+      "step": 1773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.998309382924769e-06,
+      "loss": 0.6177,
+      "step": 1774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.001690617075233e-06,
+      "loss": 0.6416,
+      "step": 1775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.005071851225698e-06,
+      "loss": 0.5639,
+      "step": 1776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.008453085376162e-06,
+      "loss": 0.6188,
+      "step": 1777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.011834319526629e-06,
+      "loss": 0.643,
+      "step": 1778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.015215553677093e-06,
+      "loss": 0.614,
+      "step": 1779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.018596787827558e-06,
+      "loss": 0.658,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.021978021978022e-06,
+      "loss": 0.6232,
+      "step": 1781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.025359256128488e-06,
+      "loss": 0.5751,
+      "step": 1782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.028740490278952e-06,
+      "loss": 0.5474,
+      "step": 1783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.032121724429418e-06,
+      "loss": 0.6592,
+      "step": 1784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.035502958579882e-06,
+      "loss": 0.584,
+      "step": 1785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.038884192730347e-06,
+      "loss": 0.6019,
+      "step": 1786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.042265426880812e-06,
+      "loss": 0.6101,
+      "step": 1787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.045646661031277e-06,
+      "loss": 0.5756,
+      "step": 1788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.049027895181742e-06,
+      "loss": 0.7522,
+      "step": 1789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.0524091293322066e-06,
+      "loss": 0.6017,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.055790363482672e-06,
+      "loss": 0.6445,
+      "step": 1791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.0591715976331365e-06,
+      "loss": 0.6441,
+      "step": 1792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.062552831783601e-06,
+      "loss": 0.7413,
+      "step": 1793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.065934065934066e-06,
+      "loss": 0.6026,
+      "step": 1794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.069315300084532e-06,
+      "loss": 0.6677,
+      "step": 1795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.072696534234996e-06,
+      "loss": 0.7231,
+      "step": 1796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.076077768385461e-06,
+      "loss": 0.6674,
+      "step": 1797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.0794590025359255e-06,
+      "loss": 0.6,
+      "step": 1798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.082840236686392e-06,
+      "loss": 0.603,
+      "step": 1799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.086221470836856e-06,
+      "loss": 0.5592,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.089602704987321e-06,
+      "loss": 0.7593,
+      "step": 1801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.092983939137785e-06,
+      "loss": 0.7244,
+      "step": 1802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.096365173288251e-06,
+      "loss": 0.6423,
+      "step": 1803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.099746407438716e-06,
+      "loss": 0.6677,
+      "step": 1804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.103127641589181e-06,
+      "loss": 0.6365,
+      "step": 1805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.106508875739645e-06,
+      "loss": 0.6808,
+      "step": 1806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.109890109890111e-06,
+      "loss": 0.5875,
+      "step": 1807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.113271344040575e-06,
+      "loss": 0.5777,
+      "step": 1808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.116652578191041e-06,
+      "loss": 0.7058,
+      "step": 1809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.120033812341505e-06,
+      "loss": 0.6537,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.1234150464919706e-06,
+      "loss": 0.5739,
+      "step": 1811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.126796280642435e-06,
+      "loss": 0.676,
+      "step": 1812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.1301775147929e-06,
+      "loss": 0.6837,
+      "step": 1813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.133558748943364e-06,
+      "loss": 0.6736,
+      "step": 1814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.13693998309383e-06,
+      "loss": 0.6545,
+      "step": 1815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.140321217244295e-06,
+      "loss": 0.6042,
+      "step": 1816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.1437024513947595e-06,
+      "loss": 0.6216,
+      "step": 1817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.147083685545224e-06,
+      "loss": 0.6809,
+      "step": 1818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.150464919695689e-06,
+      "loss": 0.7022,
+      "step": 1819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.5951,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.157227387996619e-06,
+      "loss": 0.6454,
+      "step": 1821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.160608622147084e-06,
+      "loss": 0.5828,
+      "step": 1822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.1639898562975485e-06,
+      "loss": 0.6523,
+      "step": 1823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.167371090448015e-06,
+      "loss": 0.6308,
+      "step": 1824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.170752324598479e-06,
+      "loss": 0.6107,
+      "step": 1825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.174133558748944e-06,
+      "loss": 0.5744,
+      "step": 1826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.177514792899408e-06,
+      "loss": 0.6098,
+      "step": 1827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.180896027049874e-06,
+      "loss": 0.6579,
+      "step": 1828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.184277261200339e-06,
+      "loss": 0.5685,
+      "step": 1829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.187658495350804e-06,
+      "loss": 0.651,
+      "step": 1830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.191039729501268e-06,
+      "loss": 0.6621,
+      "step": 1831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.194420963651734e-06,
+      "loss": 0.5658,
+      "step": 1832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.197802197802198e-06,
+      "loss": 0.5722,
+      "step": 1833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.201183431952663e-06,
+      "loss": 0.61,
+      "step": 1834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.204564666103128e-06,
+      "loss": 0.5863,
+      "step": 1835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.2079459002535936e-06,
+      "loss": 0.6186,
+      "step": 1836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.211327134404058e-06,
+      "loss": 0.6833,
+      "step": 1837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.214708368554523e-06,
+      "loss": 0.6506,
+      "step": 1838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.218089602704987e-06,
+      "loss": 0.6301,
+      "step": 1839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.2214708368554535e-06,
+      "loss": 0.5812,
+      "step": 1840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.224852071005918e-06,
+      "loss": 0.563,
+      "step": 1841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.2282333051563825e-06,
+      "loss": 0.6459,
+      "step": 1842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.231614539306847e-06,
+      "loss": 0.6001,
+      "step": 1843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.234995773457312e-06,
+      "loss": 0.6643,
+      "step": 1844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.238377007607778e-06,
+      "loss": 0.7093,
+      "step": 1845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.2417582417582424e-06,
+      "loss": 0.6942,
+      "step": 1846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.245139475908707e-06,
+      "loss": 0.7077,
+      "step": 1847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.2485207100591715e-06,
+      "loss": 0.6453,
+      "step": 1848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.251901944209638e-06,
+      "loss": 0.5418,
+      "step": 1849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.255283178360102e-06,
+      "loss": 0.6949,
+      "step": 1850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.258664412510567e-06,
+      "loss": 0.5589,
+      "step": 1851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.262045646661031e-06,
+      "loss": 0.6265,
+      "step": 1852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.265426880811497e-06,
+      "loss": 0.6397,
+      "step": 1853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.268808114961961e-06,
+      "loss": 0.6561,
+      "step": 1854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.272189349112427e-06,
+      "loss": 0.6472,
+      "step": 1855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.275570583262891e-06,
+      "loss": 0.5877,
+      "step": 1856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.278951817413357e-06,
+      "loss": 0.6079,
+      "step": 1857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.282333051563821e-06,
+      "loss": 0.6624,
+      "step": 1858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 0.6943,
+      "step": 1859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.28909551986475e-06,
+      "loss": 0.5543,
+      "step": 1860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.292476754015217e-06,
+      "loss": 0.5641,
+      "step": 1861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.295857988165681e-06,
+      "loss": 0.6267,
+      "step": 1862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.299239222316146e-06,
+      "loss": 0.5572,
+      "step": 1863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.30262045646661e-06,
+      "loss": 0.6396,
+      "step": 1864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.3060016906170765e-06,
+      "loss": 0.649,
+      "step": 1865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.309382924767541e-06,
+      "loss": 0.7517,
+      "step": 1866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.3127641589180056e-06,
+      "loss": 0.5881,
+      "step": 1867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.31614539306847e-06,
+      "loss": 0.7343,
+      "step": 1868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.319526627218936e-06,
+      "loss": 0.5657,
+      "step": 1869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.322907861369401e-06,
+      "loss": 0.5574,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.3262890955198655e-06,
+      "loss": 0.6938,
+      "step": 1871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.32967032967033e-06,
+      "loss": 0.7071,
+      "step": 1872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.3330515638207945e-06,
+      "loss": 0.6323,
+      "step": 1873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.33643279797126e-06,
+      "loss": 0.5904,
+      "step": 1874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.339814032121725e-06,
+      "loss": 0.6042,
+      "step": 1875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.34319526627219e-06,
+      "loss": 0.6074,
+      "step": 1876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.3465765004226544e-06,
+      "loss": 0.6798,
+      "step": 1877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.34995773457312e-06,
+      "loss": 0.606,
+      "step": 1878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.353338968723584e-06,
+      "loss": 0.6371,
+      "step": 1879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.356720202874049e-06,
+      "loss": 0.6131,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.360101437024514e-06,
+      "loss": 0.6597,
+      "step": 1881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.36348267117498e-06,
+      "loss": 0.7011,
+      "step": 1882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.366863905325444e-06,
+      "loss": 0.5749,
+      "step": 1883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.370245139475909e-06,
+      "loss": 0.5828,
+      "step": 1884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.373626373626373e-06,
+      "loss": 0.5987,
+      "step": 1885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.37700760777684e-06,
+      "loss": 0.6105,
+      "step": 1886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.380388841927304e-06,
+      "loss": 0.5562,
+      "step": 1887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.383770076077769e-06,
+      "loss": 0.6681,
+      "step": 1888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.387151310228233e-06,
+      "loss": 0.5828,
+      "step": 1889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.3905325443786995e-06,
+      "loss": 0.5962,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.393913778529164e-06,
+      "loss": 0.6068,
+      "step": 1891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.397295012679629e-06,
+      "loss": 0.752,
+      "step": 1892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.400676246830093e-06,
+      "loss": 0.6367,
+      "step": 1893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.4040574809805585e-06,
+      "loss": 0.5792,
+      "step": 1894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.407438715131024e-06,
+      "loss": 0.6676,
+      "step": 1895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.4108199492814885e-06,
+      "loss": 0.6535,
+      "step": 1896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.414201183431953e-06,
+      "loss": 0.545,
+      "step": 1897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.417582417582418e-06,
+      "loss": 0.6566,
+      "step": 1898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.420963651732883e-06,
+      "loss": 0.641,
+      "step": 1899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.4243448858833475e-06,
+      "loss": 0.6169,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.427726120033813e-06,
+      "loss": 0.5386,
+      "step": 1901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.431107354184278e-06,
+      "loss": 0.5766,
+      "step": 1902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.434488588334743e-06,
+      "loss": 0.6305,
+      "step": 1903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.437869822485207e-06,
+      "loss": 0.6482,
+      "step": 1904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.441251056635672e-06,
+      "loss": 0.6352,
+      "step": 1905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.444632290786137e-06,
+      "loss": 0.6633,
+      "step": 1906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.448013524936603e-06,
+      "loss": 0.594,
+      "step": 1907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.451394759087067e-06,
+      "loss": 0.6692,
+      "step": 1908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.454775993237532e-06,
+      "loss": 0.6495,
+      "step": 1909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.458157227387996e-06,
+      "loss": 0.5678,
+      "step": 1910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.461538461538463e-06,
+      "loss": 0.5563,
+      "step": 1911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.464919695688927e-06,
+      "loss": 0.6665,
+      "step": 1912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.468300929839392e-06,
+      "loss": 0.5662,
+      "step": 1913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.471682163989856e-06,
+      "loss": 0.6225,
+      "step": 1914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.4750633981403225e-06,
+      "loss": 0.6317,
+      "step": 1915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.478444632290787e-06,
+      "loss": 0.6212,
+      "step": 1916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.481825866441252e-06,
+      "loss": 0.6807,
+      "step": 1917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.485207100591716e-06,
+      "loss": 0.7292,
+      "step": 1918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.4885883347421815e-06,
+      "loss": 0.662,
+      "step": 1919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.491969568892647e-06,
+      "loss": 0.6049,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.4953508030431115e-06,
+      "loss": 0.5591,
+      "step": 1921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.498732037193576e-06,
+      "loss": 0.6362,
+      "step": 1922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.5021132713440414e-06,
+      "loss": 0.6039,
+      "step": 1923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.505494505494506e-06,
+      "loss": 0.5442,
+      "step": 1924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.5088757396449705e-06,
+      "loss": 0.6412,
+      "step": 1925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.512256973795436e-06,
+      "loss": 0.5554,
+      "step": 1926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.515638207945901e-06,
+      "loss": 0.6289,
+      "step": 1927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.519019442096366e-06,
+      "loss": 0.5558,
+      "step": 1928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.52240067624683e-06,
+      "loss": 0.5816,
+      "step": 1929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.525781910397295e-06,
+      "loss": 0.6362,
+      "step": 1930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.529163144547761e-06,
+      "loss": 0.6685,
+      "step": 1931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.532544378698226e-06,
+      "loss": 0.567,
+      "step": 1932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.53592561284869e-06,
+      "loss": 0.6839,
+      "step": 1933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.539306846999155e-06,
+      "loss": 0.6273,
+      "step": 1934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.542688081149619e-06,
+      "loss": 0.5909,
+      "step": 1935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.546069315300086e-06,
+      "loss": 0.6385,
+      "step": 1936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.54945054945055e-06,
+      "loss": 0.7171,
+      "step": 1937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.552831783601015e-06,
+      "loss": 0.6221,
+      "step": 1938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.556213017751479e-06,
+      "loss": 0.5849,
+      "step": 1939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.5595942519019455e-06,
+      "loss": 0.5509,
+      "step": 1940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.56297548605241e-06,
+      "loss": 0.6849,
+      "step": 1941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.566356720202875e-06,
+      "loss": 0.5794,
+      "step": 1942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.569737954353339e-06,
+      "loss": 0.569,
+      "step": 1943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.5731191885038046e-06,
+      "loss": 0.6041,
+      "step": 1944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.576500422654269e-06,
+      "loss": 0.5889,
+      "step": 1945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.5798816568047345e-06,
+      "loss": 0.5302,
+      "step": 1946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.583262890955199e-06,
+      "loss": 0.6803,
+      "step": 1947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.5866441251056644e-06,
+      "loss": 0.594,
+      "step": 1948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.590025359256129e-06,
+      "loss": 0.5674,
+      "step": 1949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.5934065934065935e-06,
+      "loss": 0.5971,
+      "step": 1950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.596787827557058e-06,
+      "loss": 0.6354,
+      "step": 1951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.600169061707524e-06,
+      "loss": 0.589,
+      "step": 1952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.603550295857989e-06,
+      "loss": 0.5693,
+      "step": 1953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.606931530008453e-06,
+      "loss": 0.5798,
+      "step": 1954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.610312764158918e-06,
+      "loss": 0.6048,
+      "step": 1955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.613693998309384e-06,
+      "loss": 0.7898,
+      "step": 1956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.617075232459849e-06,
+      "loss": 0.5543,
+      "step": 1957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.620456466610313e-06,
+      "loss": 0.7422,
+      "step": 1958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.623837700760778e-06,
+      "loss": 0.5556,
+      "step": 1959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.627218934911244e-06,
+      "loss": 0.6537,
+      "step": 1960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.630600169061709e-06,
+      "loss": 0.6845,
+      "step": 1961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.633981403212173e-06,
+      "loss": 0.6157,
+      "step": 1962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.637362637362638e-06,
+      "loss": 0.5948,
+      "step": 1963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.640743871513102e-06,
+      "loss": 0.612,
+      "step": 1964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.644125105663568e-06,
+      "loss": 0.6378,
+      "step": 1965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.647506339814033e-06,
+      "loss": 0.6434,
+      "step": 1966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.650887573964498e-06,
+      "loss": 0.6784,
+      "step": 1967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.654268808114962e-06,
+      "loss": 0.6134,
+      "step": 1968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.6576500422654276e-06,
+      "loss": 0.5586,
+      "step": 1969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.661031276415892e-06,
+      "loss": 0.6455,
+      "step": 1970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.664412510566357e-06,
+      "loss": 0.6939,
+      "step": 1971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.667793744716822e-06,
+      "loss": 0.6073,
+      "step": 1972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.6711749788672875e-06,
+      "loss": 0.6836,
+      "step": 1973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.674556213017752e-06,
+      "loss": 0.6448,
+      "step": 1974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.6779374471682166e-06,
+      "loss": 0.6633,
+      "step": 1975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.681318681318681e-06,
+      "loss": 0.567,
+      "step": 1976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.684699915469147e-06,
+      "loss": 0.6908,
+      "step": 1977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.688081149619612e-06,
+      "loss": 0.5888,
+      "step": 1978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.6914623837700764e-06,
+      "loss": 0.7216,
+      "step": 1979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.694843617920541e-06,
+      "loss": 0.7164,
+      "step": 1980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.698224852071007e-06,
+      "loss": 0.6333,
+      "step": 1981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.701606086221472e-06,
+      "loss": 0.5908,
+      "step": 1982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.704987320371936e-06,
+      "loss": 0.5253,
+      "step": 1983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.708368554522401e-06,
+      "loss": 0.5881,
+      "step": 1984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.711749788672866e-06,
+      "loss": 0.6101,
+      "step": 1985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.715131022823332e-06,
+      "loss": 0.6174,
+      "step": 1986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.718512256973796e-06,
+      "loss": 0.6051,
+      "step": 1987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.721893491124261e-06,
+      "loss": 0.6357,
+      "step": 1988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.725274725274726e-06,
+      "loss": 0.5742,
+      "step": 1989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.728655959425191e-06,
+      "loss": 0.6011,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.732037193575655e-06,
+      "loss": 0.638,
+      "step": 1991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.735418427726121e-06,
+      "loss": 0.6341,
+      "step": 1992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.738799661876585e-06,
+      "loss": 0.6227,
+      "step": 1993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.742180896027051e-06,
+      "loss": 0.7338,
+      "step": 1994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.745562130177515e-06,
+      "loss": 0.5382,
+      "step": 1995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.74894336432798e-06,
+      "loss": 0.5705,
+      "step": 1996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.752324598478445e-06,
+      "loss": 0.6165,
+      "step": 1997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.7557058326289105e-06,
+      "loss": 0.5305,
+      "step": 1998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.759087066779375e-06,
+      "loss": 0.6591,
+      "step": 1999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.7624683009298396e-06,
+      "loss": 0.7175,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.765849535080304e-06,
+      "loss": 0.6195,
+      "step": 2001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.76923076923077e-06,
+      "loss": 0.6018,
+      "step": 2002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.772612003381235e-06,
+      "loss": 0.6275,
+      "step": 2003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.7759932375316995e-06,
+      "loss": 0.6968,
+      "step": 2004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.779374471682164e-06,
+      "loss": 0.5996,
+      "step": 2005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.78275570583263e-06,
+      "loss": 0.6652,
+      "step": 2006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.786136939983095e-06,
+      "loss": 0.6439,
+      "step": 2007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.789518174133559e-06,
+      "loss": 0.6188,
+      "step": 2008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.792899408284024e-06,
+      "loss": 0.5934,
+      "step": 2009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.796280642434489e-06,
+      "loss": 0.66,
+      "step": 2010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.799661876584954e-06,
+      "loss": 0.6269,
+      "step": 2011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.803043110735419e-06,
+      "loss": 0.5816,
+      "step": 2012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.806424344885884e-06,
+      "loss": 0.5693,
+      "step": 2013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.809805579036349e-06,
+      "loss": 0.629,
+      "step": 2014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.813186813186814e-06,
+      "loss": 0.6991,
+      "step": 2015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.816568047337278e-06,
+      "loss": 0.6428,
+      "step": 2016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.819949281487744e-06,
+      "loss": 0.5753,
+      "step": 2017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.823330515638209e-06,
+      "loss": 0.5853,
+      "step": 2018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.826711749788674e-06,
+      "loss": 0.6473,
+      "step": 2019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.830092983939138e-06,
+      "loss": 0.6276,
+      "step": 2020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.833474218089603e-06,
+      "loss": 0.6512,
+      "step": 2021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.836855452240067e-06,
+      "loss": 0.6554,
+      "step": 2022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.8402366863905335e-06,
+      "loss": 0.6444,
+      "step": 2023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.843617920540998e-06,
+      "loss": 0.717,
+      "step": 2024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.846999154691463e-06,
+      "loss": 0.7072,
+      "step": 2025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.850380388841927e-06,
+      "loss": 0.5665,
+      "step": 2026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.853761622992393e-06,
+      "loss": 0.6995,
+      "step": 2027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 0.6859,
+      "step": 2028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.8605240912933225e-06,
+      "loss": 0.6042,
+      "step": 2029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.863905325443787e-06,
+      "loss": 0.5669,
+      "step": 2030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.867286559594252e-06,
+      "loss": 0.5164,
+      "step": 2031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.870667793744718e-06,
+      "loss": 0.6646,
+      "step": 2032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.874049027895182e-06,
+      "loss": 0.5358,
+      "step": 2033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.877430262045647e-06,
+      "loss": 0.5591,
+      "step": 2034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.880811496196112e-06,
+      "loss": 0.6789,
+      "step": 2035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.884192730346577e-06,
+      "loss": 0.6355,
+      "step": 2036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.887573964497042e-06,
+      "loss": 0.6145,
+      "step": 2037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.890955198647507e-06,
+      "loss": 0.4946,
+      "step": 2038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.894336432797972e-06,
+      "loss": 0.6289,
+      "step": 2039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.897717666948437e-06,
+      "loss": 0.6331,
+      "step": 2040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.901098901098901e-06,
+      "loss": 0.6668,
+      "step": 2041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.904480135249366e-06,
+      "loss": 0.6599,
+      "step": 2042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.907861369399832e-06,
+      "loss": 0.6687,
+      "step": 2043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.911242603550297e-06,
+      "loss": 0.5898,
+      "step": 2044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.914623837700761e-06,
+      "loss": 0.6601,
+      "step": 2045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.918005071851226e-06,
+      "loss": 0.6585,
+      "step": 2046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.921386306001692e-06,
+      "loss": 0.5462,
+      "step": 2047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.9247675401521565e-06,
+      "loss": 0.6163,
+      "step": 2048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.928148774302621e-06,
+      "loss": 0.5693,
+      "step": 2049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.931530008453086e-06,
+      "loss": 0.5968,
+      "step": 2050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.93491124260355e-06,
+      "loss": 0.6061,
+      "step": 2051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.938292476754016e-06,
+      "loss": 0.5484,
+      "step": 2052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.941673710904481e-06,
+      "loss": 0.6886,
+      "step": 2053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.9450549450549455e-06,
+      "loss": 0.5773,
+      "step": 2054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.94843617920541e-06,
+      "loss": 0.5941,
+      "step": 2055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.9518174133558754e-06,
+      "loss": 0.5621,
+      "step": 2056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.955198647506341e-06,
+      "loss": 0.5499,
+      "step": 2057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.958579881656805e-06,
+      "loss": 0.6246,
+      "step": 2058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.96196111580727e-06,
+      "loss": 0.6152,
+      "step": 2059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.965342349957735e-06,
+      "loss": 0.5818,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.9687235841082e-06,
+      "loss": 0.5525,
+      "step": 2061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.972104818258664e-06,
+      "loss": 0.685,
+      "step": 2062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.97548605240913e-06,
+      "loss": 0.6251,
+      "step": 2063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.978867286559595e-06,
+      "loss": 0.5711,
+      "step": 2064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.98224852071006e-06,
+      "loss": 0.5915,
+      "step": 2065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.985629754860524e-06,
+      "loss": 0.629,
+      "step": 2066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.989010989010989e-06,
+      "loss": 0.6254,
+      "step": 2067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.992392223161455e-06,
+      "loss": 0.578,
+      "step": 2068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.99577345731192e-06,
+      "loss": 0.553,
+      "step": 2069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.999154691462384e-06,
+      "loss": 0.5628,
+      "step": 2070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.002535925612849e-06,
+      "loss": 0.5989,
+      "step": 2071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.005917159763315e-06,
+      "loss": 0.5915,
+      "step": 2072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.0092983939137795e-06,
+      "loss": 0.6486,
+      "step": 2073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.012679628064244e-06,
+      "loss": 0.6208,
+      "step": 2074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.016060862214709e-06,
+      "loss": 0.6526,
+      "step": 2075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.019442096365174e-06,
+      "loss": 0.614,
+      "step": 2076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.022823330515639e-06,
+      "loss": 0.6537,
+      "step": 2077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.026204564666104e-06,
+      "loss": 0.6387,
+      "step": 2078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.0295857988165685e-06,
+      "loss": 0.5776,
+      "step": 2079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.032967032967034e-06,
+      "loss": 0.596,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.0363482671174984e-06,
+      "loss": 0.5909,
+      "step": 2081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.039729501267963e-06,
+      "loss": 0.5508,
+      "step": 2082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.043110735418428e-06,
+      "loss": 0.6524,
+      "step": 2083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.046491969568893e-06,
+      "loss": 0.5429,
+      "step": 2084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.049873203719358e-06,
+      "loss": 0.7515,
+      "step": 2085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.053254437869823e-06,
+      "loss": 0.649,
+      "step": 2086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.0566356720202874e-06,
+      "loss": 0.5832,
+      "step": 2087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.060016906170753e-06,
+      "loss": 0.5827,
+      "step": 2088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.063398140321218e-06,
+      "loss": 0.5984,
+      "step": 2089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.066779374471683e-06,
+      "loss": 0.6128,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.070160608622147e-06,
+      "loss": 0.6581,
+      "step": 2091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.073541842772612e-06,
+      "loss": 0.6505,
+      "step": 2092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.076923076923078e-06,
+      "loss": 0.6006,
+      "step": 2093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.080304311073543e-06,
+      "loss": 0.6242,
+      "step": 2094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.083685545224007e-06,
+      "loss": 0.5895,
+      "step": 2095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.087066779374472e-06,
+      "loss": 0.574,
+      "step": 2096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.090448013524938e-06,
+      "loss": 0.6309,
+      "step": 2097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.0938292476754025e-06,
+      "loss": 0.6404,
+      "step": 2098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.097210481825867e-06,
+      "loss": 0.6637,
+      "step": 2099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.100591715976332e-06,
+      "loss": 0.6147,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.103972950126797e-06,
+      "loss": 0.4893,
+      "step": 2101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.1073541842772616e-06,
+      "loss": 0.6778,
+      "step": 2102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.110735418427727e-06,
+      "loss": 0.5578,
+      "step": 2103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.1141166525781915e-06,
+      "loss": 0.6319,
+      "step": 2104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.117497886728657e-06,
+      "loss": 0.5742,
+      "step": 2105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.1208791208791215e-06,
+      "loss": 0.6007,
+      "step": 2106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.124260355029586e-06,
+      "loss": 0.6422,
+      "step": 2107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.127641589180051e-06,
+      "loss": 0.6257,
+      "step": 2108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.131022823330517e-06,
+      "loss": 0.593,
+      "step": 2109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.134404057480981e-06,
+      "loss": 0.5969,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.137785291631446e-06,
+      "loss": 0.5715,
+      "step": 2111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.1411665257819104e-06,
+      "loss": 0.5311,
+      "step": 2112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.144547759932375e-06,
+      "loss": 0.587,
+      "step": 2113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.147928994082841e-06,
+      "loss": 0.5481,
+      "step": 2114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.151310228233306e-06,
+      "loss": 0.6224,
+      "step": 2115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.15469146238377e-06,
+      "loss": 0.6595,
+      "step": 2116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.158072696534235e-06,
+      "loss": 0.5797,
+      "step": 2117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.161453930684701e-06,
+      "loss": 0.527,
+      "step": 2118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.164835164835166e-06,
+      "loss": 0.6745,
+      "step": 2119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.16821639898563e-06,
+      "loss": 0.6059,
+      "step": 2120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.171597633136095e-06,
+      "loss": 0.5989,
+      "step": 2121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.17497886728656e-06,
+      "loss": 0.5849,
+      "step": 2122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.1783601014370256e-06,
+      "loss": 0.5732,
+      "step": 2123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.18174133558749e-06,
+      "loss": 0.594,
+      "step": 2124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.185122569737955e-06,
+      "loss": 0.5168,
+      "step": 2125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.18850380388842e-06,
+      "loss": 0.6099,
+      "step": 2126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.191885038038885e-06,
+      "loss": 0.6978,
+      "step": 2127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.19526627218935e-06,
+      "loss": 0.6196,
+      "step": 2128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.1986475063398145e-06,
+      "loss": 0.591,
+      "step": 2129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.20202874049028e-06,
+      "loss": 0.5777,
+      "step": 2130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.2054099746407445e-06,
+      "loss": 0.5221,
+      "step": 2131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.208791208791209e-06,
+      "loss": 0.565,
+      "step": 2132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.2121724429416736e-06,
+      "loss": 0.6326,
+      "step": 2133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.21555367709214e-06,
+      "loss": 0.5986,
+      "step": 2134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.218934911242604e-06,
+      "loss": 0.5853,
+      "step": 2135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.222316145393069e-06,
+      "loss": 0.5553,
+      "step": 2136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.2256973795435335e-06,
+      "loss": 0.5744,
+      "step": 2137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.229078613694e-06,
+      "loss": 0.6055,
+      "step": 2138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.232459847844464e-06,
+      "loss": 0.7061,
+      "step": 2139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.235841081994929e-06,
+      "loss": 0.5819,
+      "step": 2140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.239222316145393e-06,
+      "loss": 0.5886,
+      "step": 2141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.242603550295858e-06,
+      "loss": 0.6673,
+      "step": 2142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.245984784446324e-06,
+      "loss": 0.6972,
+      "step": 2143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.249366018596789e-06,
+      "loss": 0.6475,
+      "step": 2144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.252747252747253e-06,
+      "loss": 0.5461,
+      "step": 2145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.256128486897718e-06,
+      "loss": 0.618,
+      "step": 2146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.259509721048183e-06,
+      "loss": 0.707,
+      "step": 2147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.2628909551986486e-06,
+      "loss": 0.6332,
+      "step": 2148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.266272189349113e-06,
+      "loss": 0.6018,
+      "step": 2149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.269653423499578e-06,
+      "loss": 0.6303,
+      "step": 2150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.273034657650043e-06,
+      "loss": 0.5707,
+      "step": 2151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.276415891800508e-06,
+      "loss": 0.5809,
+      "step": 2152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.279797125950972e-06,
+      "loss": 0.6205,
+      "step": 2153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.2831783601014375e-06,
+      "loss": 0.6674,
+      "step": 2154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.286559594251903e-06,
+      "loss": 0.6416,
+      "step": 2155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.2899408284023675e-06,
+      "loss": 0.5753,
+      "step": 2156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.293322062552832e-06,
+      "loss": 0.7845,
+      "step": 2157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.296703296703297e-06,
+      "loss": 0.5789,
+      "step": 2158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.300084530853763e-06,
+      "loss": 0.5824,
+      "step": 2159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.303465765004227e-06,
+      "loss": 0.5511,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.306846999154692e-06,
+      "loss": 0.6227,
+      "step": 2161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.3102282333051565e-06,
+      "loss": 0.6869,
+      "step": 2162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.313609467455623e-06,
+      "loss": 0.7173,
+      "step": 2163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.316990701606087e-06,
+      "loss": 0.5158,
+      "step": 2164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.320371935756552e-06,
+      "loss": 0.6116,
+      "step": 2165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.323753169907016e-06,
+      "loss": 0.6828,
+      "step": 2166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.327134404057482e-06,
+      "loss": 0.6332,
+      "step": 2167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.330515638207947e-06,
+      "loss": 0.6841,
+      "step": 2168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.333896872358412e-06,
+      "loss": 0.5527,
+      "step": 2169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.337278106508876e-06,
+      "loss": 0.5308,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.340659340659341e-06,
+      "loss": 0.5561,
+      "step": 2171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.344040574809806e-06,
+      "loss": 0.6103,
+      "step": 2172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.347421808960271e-06,
+      "loss": 0.5927,
+      "step": 2173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.350803043110736e-06,
+      "loss": 0.6455,
+      "step": 2174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.354184277261201e-06,
+      "loss": 0.661,
+      "step": 2175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.357565511411666e-06,
+      "loss": 0.5777,
+      "step": 2176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.360946745562131e-06,
+      "loss": 0.6216,
+      "step": 2177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.364327979712595e-06,
+      "loss": 0.5473,
+      "step": 2178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.3677092138630606e-06,
+      "loss": 0.6016,
+      "step": 2179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.371090448013526e-06,
+      "loss": 0.6723,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.3744716821639905e-06,
+      "loss": 0.634,
+      "step": 2181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.377852916314455e-06,
+      "loss": 0.6607,
+      "step": 2182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.38123415046492e-06,
+      "loss": 0.5845,
+      "step": 2183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.384615384615386e-06,
+      "loss": 0.6506,
+      "step": 2184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.38799661876585e-06,
+      "loss": 0.729,
+      "step": 2185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.391377852916315e-06,
+      "loss": 0.6995,
+      "step": 2186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.3947590870667795e-06,
+      "loss": 0.5917,
+      "step": 2187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.398140321217246e-06,
+      "loss": 0.6832,
+      "step": 2188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.40152155536771e-06,
+      "loss": 0.5904,
+      "step": 2189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.404902789518175e-06,
+      "loss": 0.6438,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.408284023668639e-06,
+      "loss": 0.7224,
+      "step": 2191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.411665257819105e-06,
+      "loss": 0.6995,
+      "step": 2192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.415046491969569e-06,
+      "loss": 0.6422,
+      "step": 2193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.418427726120035e-06,
+      "loss": 0.6642,
+      "step": 2194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.421808960270499e-06,
+      "loss": 0.6344,
+      "step": 2195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.425190194420965e-06,
+      "loss": 0.6656,
+      "step": 2196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 0.6007,
+      "step": 2197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.431952662721894e-06,
+      "loss": 0.5549,
+      "step": 2198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.435333896872359e-06,
+      "loss": 0.5749,
+      "step": 2199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.438715131022824e-06,
+      "loss": 0.6454,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.442096365173289e-06,
+      "loss": 0.5241,
+      "step": 2201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.445477599323754e-06,
+      "loss": 0.7111,
+      "step": 2202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.448858833474218e-06,
+      "loss": 0.5723,
+      "step": 2203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.452240067624683e-06,
+      "loss": 0.6105,
+      "step": 2204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.455621301775149e-06,
+      "loss": 0.6165,
+      "step": 2205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.4590025359256135e-06,
+      "loss": 0.6596,
+      "step": 2206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.462383770076078e-06,
+      "loss": 0.58,
+      "step": 2207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.465765004226543e-06,
+      "loss": 0.5983,
+      "step": 2208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.469146238377009e-06,
+      "loss": 0.5957,
+      "step": 2209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.472527472527473e-06,
+      "loss": 0.6153,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.475908706677938e-06,
+      "loss": 0.6211,
+      "step": 2211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.4792899408284025e-06,
+      "loss": 0.7102,
+      "step": 2212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.482671174978868e-06,
+      "loss": 0.5834,
+      "step": 2213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.486052409129333e-06,
+      "loss": 0.589,
+      "step": 2214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.489433643279798e-06,
+      "loss": 0.7109,
+      "step": 2215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.492814877430262e-06,
+      "loss": 0.6344,
+      "step": 2216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.496196111580728e-06,
+      "loss": 0.5628,
+      "step": 2217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.499577345731192e-06,
+      "loss": 0.5615,
+      "step": 2218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.502958579881658e-06,
+      "loss": 0.6892,
+      "step": 2219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.506339814032122e-06,
+      "loss": 0.5288,
+      "step": 2220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.509721048182588e-06,
+      "loss": 0.6518,
+      "step": 2221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.513102282333052e-06,
+      "loss": 0.6468,
+      "step": 2222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.516483516483517e-06,
+      "loss": 0.614,
+      "step": 2223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.519864750633981e-06,
+      "loss": 0.6412,
+      "step": 2224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.5232459847844476e-06,
+      "loss": 0.7041,
+      "step": 2225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.526627218934912e-06,
+      "loss": 0.6304,
+      "step": 2226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.530008453085377e-06,
+      "loss": 0.5898,
+      "step": 2227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.533389687235841e-06,
+      "loss": 0.5978,
+      "step": 2228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.5367709213863074e-06,
+      "loss": 0.5897,
+      "step": 2229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.540152155536772e-06,
+      "loss": 0.7151,
+      "step": 2230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.5435333896872365e-06,
+      "loss": 0.5727,
+      "step": 2231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.546914623837701e-06,
+      "loss": 0.7111,
+      "step": 2232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.550295857988166e-06,
+      "loss": 0.6291,
+      "step": 2233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.553677092138632e-06,
+      "loss": 0.6394,
+      "step": 2234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.5570583262890964e-06,
+      "loss": 0.5955,
+      "step": 2235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.560439560439561e-06,
+      "loss": 0.6133,
+      "step": 2236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.5638207945900255e-06,
+      "loss": 0.6384,
+      "step": 2237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.567202028740491e-06,
+      "loss": 0.5559,
+      "step": 2238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.570583262890956e-06,
+      "loss": 0.4948,
+      "step": 2239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.573964497041421e-06,
+      "loss": 0.6983,
+      "step": 2240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.577345731191885e-06,
+      "loss": 0.5721,
+      "step": 2241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.580726965342351e-06,
+      "loss": 0.5819,
+      "step": 2242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.584108199492815e-06,
+      "loss": 0.5669,
+      "step": 2243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.58748943364328e-06,
+      "loss": 0.5255,
+      "step": 2244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.590870667793745e-06,
+      "loss": 0.5918,
+      "step": 2245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.594251901944211e-06,
+      "loss": 0.5742,
+      "step": 2246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.597633136094675e-06,
+      "loss": 0.597,
+      "step": 2247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.60101437024514e-06,
+      "loss": 0.617,
+      "step": 2248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.604395604395604e-06,
+      "loss": 0.6282,
+      "step": 2249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.6077768385460706e-06,
+      "loss": 0.7607,
+      "step": 2250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.611158072696535e-06,
+      "loss": 0.5801,
+      "step": 2251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.614539306847e-06,
+      "loss": 0.5919,
+      "step": 2252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.617920540997464e-06,
+      "loss": 0.6123,
+      "step": 2253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.6213017751479305e-06,
+      "loss": 0.6045,
+      "step": 2254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.624683009298395e-06,
+      "loss": 0.584,
+      "step": 2255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.6280642434488596e-06,
+      "loss": 0.5839,
+      "step": 2256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.631445477599323e-06,
+      "loss": 0.6013,
+      "step": 2257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.63482671174979e-06,
+      "loss": 0.6668,
+      "step": 2258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.638207945900254e-06,
+      "loss": 0.617,
+      "step": 2259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.641589180050719e-06,
+      "loss": 0.6501,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.644970414201183e-06,
+      "loss": 0.5564,
+      "step": 2261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.648351648351648e-06,
+      "loss": 0.5802,
+      "step": 2262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.651732882502114e-06,
+      "loss": 0.5948,
+      "step": 2263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.655114116652578e-06,
+      "loss": 0.687,
+      "step": 2264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.658495350803043e-06,
+      "loss": 0.6836,
+      "step": 2265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.661876584953508e-06,
+      "loss": 0.7586,
+      "step": 2266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.665257819103974e-06,
+      "loss": 0.6012,
+      "step": 2267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.668639053254438e-06,
+      "loss": 0.6415,
+      "step": 2268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.672020287404903e-06,
+      "loss": 0.6229,
+      "step": 2269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.675401521555367e-06,
+      "loss": 0.6083,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.678782755705834e-06,
+      "loss": 0.6605,
+      "step": 2271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.682163989856298e-06,
+      "loss": 0.6548,
+      "step": 2272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.685545224006763e-06,
+      "loss": 0.6238,
+      "step": 2273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.688926458157227e-06,
+      "loss": 0.5822,
+      "step": 2274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.6131,
+      "step": 2275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.695688926458158e-06,
+      "loss": 0.6622,
+      "step": 2276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.699070160608623e-06,
+      "loss": 0.6158,
+      "step": 2277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.702451394759087e-06,
+      "loss": 0.6496,
+      "step": 2278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.705832628909553e-06,
+      "loss": 0.6467,
+      "step": 2279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.709213863060018e-06,
+      "loss": 0.6977,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.712595097210483e-06,
+      "loss": 0.5569,
+      "step": 2281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.715976331360947e-06,
+      "loss": 0.581,
+      "step": 2282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.719357565511413e-06,
+      "loss": 0.5858,
+      "step": 2283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.722738799661878e-06,
+      "loss": 0.6023,
+      "step": 2284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.726120033812342e-06,
+      "loss": 0.586,
+      "step": 2285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.729501267962807e-06,
+      "loss": 0.6672,
+      "step": 2286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.732882502113273e-06,
+      "loss": 0.6285,
+      "step": 2287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.736263736263738e-06,
+      "loss": 0.6767,
+      "step": 2288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.739644970414202e-06,
+      "loss": 0.6878,
+      "step": 2289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.743026204564667e-06,
+      "loss": 0.6525,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.746407438715131e-06,
+      "loss": 0.5804,
+      "step": 2291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.749788672865598e-06,
+      "loss": 0.6507,
+      "step": 2292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.753169907016062e-06,
+      "loss": 0.6157,
+      "step": 2293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.756551141166527e-06,
+      "loss": 0.6754,
+      "step": 2294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.759932375316991e-06,
+      "loss": 0.5806,
+      "step": 2295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.763313609467456e-06,
+      "loss": 0.6596,
+      "step": 2296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.76669484361792e-06,
+      "loss": 0.6173,
+      "step": 2297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.770076077768387e-06,
+      "loss": 0.6087,
+      "step": 2298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.773457311918851e-06,
+      "loss": 0.5665,
+      "step": 2299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.776838546069316e-06,
+      "loss": 0.5176,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.78021978021978e-06,
+      "loss": 0.6303,
+      "step": 2301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.783601014370245e-06,
+      "loss": 0.6456,
+      "step": 2302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.786982248520711e-06,
+      "loss": 0.5888,
+      "step": 2303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.790363482671176e-06,
+      "loss": 0.6543,
+      "step": 2304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.79374471682164e-06,
+      "loss": 0.6473,
+      "step": 2305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.797125950972105e-06,
+      "loss": 0.6287,
+      "step": 2306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.80050718512257e-06,
+      "loss": 0.6351,
+      "step": 2307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.803888419273036e-06,
+      "loss": 0.6306,
+      "step": 2308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.8072696534235e-06,
+      "loss": 0.5679,
+      "step": 2309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.810650887573965e-06,
+      "loss": 0.6101,
+      "step": 2310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.81403212172443e-06,
+      "loss": 0.6271,
+      "step": 2311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.817413355874895e-06,
+      "loss": 0.6654,
+      "step": 2312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.82079459002536e-06,
+      "loss": 0.5449,
+      "step": 2313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.824175824175825e-06,
+      "loss": 0.6215,
+      "step": 2314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.827557058326289e-06,
+      "loss": 0.5722,
+      "step": 2315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.830938292476755e-06,
+      "loss": 0.6852,
+      "step": 2316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.83431952662722e-06,
+      "loss": 0.6219,
+      "step": 2317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.837700760777684e-06,
+      "loss": 0.712,
+      "step": 2318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.841081994928149e-06,
+      "loss": 0.741,
+      "step": 2319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.844463229078613e-06,
+      "loss": 0.4781,
+      "step": 2320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.84784446322908e-06,
+      "loss": 0.5855,
+      "step": 2321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.851225697379544e-06,
+      "loss": 0.5502,
+      "step": 2322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.854606931530009e-06,
+      "loss": 0.6126,
+      "step": 2323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.857988165680473e-06,
+      "loss": 0.6431,
+      "step": 2324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.86136939983094e-06,
+      "loss": 0.6298,
+      "step": 2325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.864750633981404e-06,
+      "loss": 0.6155,
+      "step": 2326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.868131868131869e-06,
+      "loss": 0.698,
+      "step": 2327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.871513102282333e-06,
+      "loss": 0.664,
+      "step": 2328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.8748943364328e-06,
+      "loss": 0.5319,
+      "step": 2329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.878275570583264e-06,
+      "loss": 0.6349,
+      "step": 2330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.881656804733729e-06,
+      "loss": 0.6416,
+      "step": 2331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.885038038884193e-06,
+      "loss": 0.5551,
+      "step": 2332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.88841927303466e-06,
+      "loss": 0.6895,
+      "step": 2333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.891800507185124e-06,
+      "loss": 0.6029,
+      "step": 2334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.895181741335588e-06,
+      "loss": 0.6333,
+      "step": 2335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.898562975486053e-06,
+      "loss": 0.6531,
+      "step": 2336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.901944209636518e-06,
+      "loss": 0.5652,
+      "step": 2337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.905325443786984e-06,
+      "loss": 0.6081,
+      "step": 2338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.908706677937448e-06,
+      "loss": 0.6013,
+      "step": 2339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.912087912087913e-06,
+      "loss": 0.5616,
+      "step": 2340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.915469146238377e-06,
+      "loss": 0.5564,
+      "step": 2341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.918850380388842e-06,
+      "loss": 0.6536,
+      "step": 2342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.922231614539308e-06,
+      "loss": 0.5952,
+      "step": 2343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.925612848689773e-06,
+      "loss": 0.5904,
+      "step": 2344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.928994082840237e-06,
+      "loss": 0.594,
+      "step": 2345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.932375316990702e-06,
+      "loss": 0.6417,
+      "step": 2346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.935756551141166e-06,
+      "loss": 0.627,
+      "step": 2347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.939137785291631e-06,
+      "loss": 0.6578,
+      "step": 2348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.942519019442097e-06,
+      "loss": 0.6357,
+      "step": 2349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.945900253592562e-06,
+      "loss": 0.6456,
+      "step": 2350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.949281487743026e-06,
+      "loss": 0.5954,
+      "step": 2351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.952662721893491e-06,
+      "loss": 0.5908,
+      "step": 2352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.956043956043955e-06,
+      "loss": 0.6788,
+      "step": 2353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.959425190194422e-06,
+      "loss": 0.5862,
+      "step": 2354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.962806424344886e-06,
+      "loss": 0.631,
+      "step": 2355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.96618765849535e-06,
+      "loss": 0.6197,
+      "step": 2356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.969568892645815e-06,
+      "loss": 0.7195,
+      "step": 2357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.972950126796282e-06,
+      "loss": 0.5498,
+      "step": 2358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.976331360946746e-06,
+      "loss": 0.6533,
+      "step": 2359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.97971259509721e-06,
+      "loss": 0.7691,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.983093829247675e-06,
+      "loss": 0.6731,
+      "step": 2361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.986475063398141e-06,
+      "loss": 0.6833,
+      "step": 2362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.989856297548606e-06,
+      "loss": 0.5255,
+      "step": 2363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.99323753169907e-06,
+      "loss": 0.6597,
+      "step": 2364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.996618765849535e-06,
+      "loss": 0.6134,
+      "step": 2365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7396,
+      "step": 2366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.003381234150466e-06,
+      "loss": 0.6447,
+      "step": 2367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.00676246830093e-06,
+      "loss": 0.646,
+      "step": 2368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.010143702451395e-06,
+      "loss": 0.534,
+      "step": 2369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.013524936601861e-06,
+      "loss": 0.6132,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.016906170752326e-06,
+      "loss": 0.6317,
+      "step": 2371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.02028740490279e-06,
+      "loss": 0.6762,
+      "step": 2372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.023668639053255e-06,
+      "loss": 0.6853,
+      "step": 2373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.027049873203721e-06,
+      "loss": 0.6625,
+      "step": 2374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.030431107354186e-06,
+      "loss": 0.6333,
+      "step": 2375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.03381234150465e-06,
+      "loss": 0.6891,
+      "step": 2376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.037193575655115e-06,
+      "loss": 0.6275,
+      "step": 2377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.04057480980558e-06,
+      "loss": 0.5873,
+      "step": 2378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.043956043956046e-06,
+      "loss": 0.606,
+      "step": 2379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.04733727810651e-06,
+      "loss": 0.5852,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.050718512256975e-06,
+      "loss": 0.6253,
+      "step": 2381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.05409974640744e-06,
+      "loss": 0.5605,
+      "step": 2382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.057480980557905e-06,
+      "loss": 0.6192,
+      "step": 2383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.06086221470837e-06,
+      "loss": 0.5335,
+      "step": 2384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.064243448858835e-06,
+      "loss": 0.6123,
+      "step": 2385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.067624683009299e-06,
+      "loss": 0.6225,
+      "step": 2386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.071005917159764e-06,
+      "loss": 0.7085,
+      "step": 2387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.074387151310228e-06,
+      "loss": 0.5616,
+      "step": 2388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.077768385460694e-06,
+      "loss": 0.5535,
+      "step": 2389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.081149619611159e-06,
+      "loss": 0.6188,
+      "step": 2390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.084530853761624e-06,
+      "loss": 0.6659,
+      "step": 2391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.087912087912088e-06,
+      "loss": 0.6392,
+      "step": 2392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.091293322062553e-06,
+      "loss": 0.6288,
+      "step": 2393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.094674556213019e-06,
+      "loss": 0.631,
+      "step": 2394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.098055790363483e-06,
+      "loss": 0.5801,
+      "step": 2395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.101437024513948e-06,
+      "loss": 0.6369,
+      "step": 2396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.104818258664412e-06,
+      "loss": 0.5967,
+      "step": 2397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.108199492814877e-06,
+      "loss": 0.5774,
+      "step": 2398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.111580726965343e-06,
+      "loss": 0.6298,
+      "step": 2399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.114961961115808e-06,
+      "loss": 0.6499,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.118343195266272e-06,
+      "loss": 0.5928,
+      "step": 2401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.121724429416737e-06,
+      "loss": 0.6911,
+      "step": 2402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.125105663567203e-06,
+      "loss": 0.584,
+      "step": 2403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.128486897717668e-06,
+      "loss": 0.5631,
+      "step": 2404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.131868131868132e-06,
+      "loss": 0.5941,
+      "step": 2405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.135249366018597e-06,
+      "loss": 0.5394,
+      "step": 2406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.138630600169063e-06,
+      "loss": 0.6527,
+      "step": 2407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.142011834319528e-06,
+      "loss": 0.5518,
+      "step": 2408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.145393068469992e-06,
+      "loss": 0.6088,
+      "step": 2409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.148774302620457e-06,
+      "loss": 0.5784,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.152155536770921e-06,
+      "loss": 0.5397,
+      "step": 2411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.155536770921387e-06,
+      "loss": 0.7095,
+      "step": 2412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.158918005071852e-06,
+      "loss": 0.6262,
+      "step": 2413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.162299239222317e-06,
+      "loss": 0.6401,
+      "step": 2414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.165680473372781e-06,
+      "loss": 0.6631,
+      "step": 2415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.169061707523247e-06,
+      "loss": 0.5767,
+      "step": 2416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.172442941673712e-06,
+      "loss": 0.6158,
+      "step": 2417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.175824175824176e-06,
+      "loss": 0.6064,
+      "step": 2418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.179205409974641e-06,
+      "loss": 0.6274,
+      "step": 2419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.182586644125107e-06,
+      "loss": 0.6734,
+      "step": 2420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.185967878275572e-06,
+      "loss": 0.6635,
+      "step": 2421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.189349112426036e-06,
+      "loss": 0.6287,
+      "step": 2422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.192730346576501e-06,
+      "loss": 0.5846,
+      "step": 2423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.196111580726967e-06,
+      "loss": 0.5905,
+      "step": 2424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.199492814877432e-06,
+      "loss": 0.7185,
+      "step": 2425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.202874049027896e-06,
+      "loss": 0.743,
+      "step": 2426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.20625528317836e-06,
+      "loss": 0.5876,
+      "step": 2427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.209636517328825e-06,
+      "loss": 0.5804,
+      "step": 2428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.213017751479292e-06,
+      "loss": 0.6663,
+      "step": 2429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.216398985629756e-06,
+      "loss": 0.5364,
+      "step": 2430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.21978021978022e-06,
+      "loss": 0.5791,
+      "step": 2431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.223161453930685e-06,
+      "loss": 0.6524,
+      "step": 2432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.22654268808115e-06,
+      "loss": 0.6531,
+      "step": 2433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.229923922231616e-06,
+      "loss": 0.6605,
+      "step": 2434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.23330515638208e-06,
+      "loss": 0.6328,
+      "step": 2435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.236686390532545e-06,
+      "loss": 0.5574,
+      "step": 2436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.24006762468301e-06,
+      "loss": 0.551,
+      "step": 2437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.243448858833474e-06,
+      "loss": 0.6093,
+      "step": 2438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.246830092983939e-06,
+      "loss": 0.6047,
+      "step": 2439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.250211327134405e-06,
+      "loss": 0.6229,
+      "step": 2440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.25359256128487e-06,
+      "loss": 0.6278,
+      "step": 2441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.256973795435334e-06,
+      "loss": 0.5943,
+      "step": 2442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.260355029585799e-06,
+      "loss": 0.6694,
+      "step": 2443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.263736263736263e-06,
+      "loss": 0.6837,
+      "step": 2444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.26711749788673e-06,
+      "loss": 0.6954,
+      "step": 2445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.270498732037194e-06,
+      "loss": 0.632,
+      "step": 2446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.273879966187659e-06,
+      "loss": 0.6062,
+      "step": 2447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.277261200338123e-06,
+      "loss": 0.6027,
+      "step": 2448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.28064243448859e-06,
+      "loss": 0.5283,
+      "step": 2449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.284023668639054e-06,
+      "loss": 0.6279,
+      "step": 2450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.287404902789518e-06,
+      "loss": 0.635,
+      "step": 2451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.290786136939983e-06,
+      "loss": 0.6344,
+      "step": 2452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.29416737109045e-06,
+      "loss": 0.6266,
+      "step": 2453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.297548605240914e-06,
+      "loss": 0.6908,
+      "step": 2454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.300929839391378e-06,
+      "loss": 0.5141,
+      "step": 2455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.304311073541843e-06,
+      "loss": 0.6095,
+      "step": 2456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.307692307692309e-06,
+      "loss": 0.6318,
+      "step": 2457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.311073541842774e-06,
+      "loss": 0.6284,
+      "step": 2458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.314454775993238e-06,
+      "loss": 0.5996,
+      "step": 2459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.317836010143703e-06,
+      "loss": 0.5902,
+      "step": 2460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.321217244294169e-06,
+      "loss": 0.7046,
+      "step": 2461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.324598478444634e-06,
+      "loss": 0.5841,
+      "step": 2462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.327979712595098e-06,
+      "loss": 0.6753,
+      "step": 2463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.331360946745563e-06,
+      "loss": 0.6253,
+      "step": 2464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.334742180896029e-06,
+      "loss": 0.569,
+      "step": 2465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.338123415046493e-06,
+      "loss": 0.5834,
+      "step": 2466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.341504649196958e-06,
+      "loss": 0.5741,
+      "step": 2467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.344885883347422e-06,
+      "loss": 0.5848,
+      "step": 2468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.348267117497887e-06,
+      "loss": 0.5802,
+      "step": 2469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.351648351648353e-06,
+      "loss": 0.6,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.355029585798818e-06,
+      "loss": 0.6112,
+      "step": 2471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.358410819949282e-06,
+      "loss": 0.5549,
+      "step": 2472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.361792054099747e-06,
+      "loss": 0.607,
+      "step": 2473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.365173288250213e-06,
+      "loss": 0.5385,
+      "step": 2474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.368554522400678e-06,
+      "loss": 0.6128,
+      "step": 2475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.371935756551142e-06,
+      "loss": 0.4976,
+      "step": 2476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.375316990701607e-06,
+      "loss": 0.5939,
+      "step": 2477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.378698224852071e-06,
+      "loss": 0.6399,
+      "step": 2478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.382079459002536e-06,
+      "loss": 0.5465,
+      "step": 2479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.385460693153002e-06,
+      "loss": 0.564,
+      "step": 2480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.388841927303467e-06,
+      "loss": 0.6065,
+      "step": 2481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.392223161453931e-06,
+      "loss": 0.6183,
+      "step": 2482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.395604395604396e-06,
+      "loss": 0.5965,
+      "step": 2483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.39898562975486e-06,
+      "loss": 0.5725,
+      "step": 2484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.402366863905327e-06,
+      "loss": 0.6564,
+      "step": 2485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.405748098055791e-06,
+      "loss": 0.6895,
+      "step": 2486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.409129332206256e-06,
+      "loss": 0.5967,
+      "step": 2487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.41251056635672e-06,
+      "loss": 0.5793,
+      "step": 2488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.415891800507185e-06,
+      "loss": 0.6614,
+      "step": 2489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.419273034657651e-06,
+      "loss": 0.5487,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.422654268808116e-06,
+      "loss": 0.6734,
+      "step": 2491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.42603550295858e-06,
+      "loss": 0.6114,
+      "step": 2492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.429416737109045e-06,
+      "loss": 0.709,
+      "step": 2493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.432797971259511e-06,
+      "loss": 0.5462,
+      "step": 2494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.436179205409975e-06,
+      "loss": 0.6382,
+      "step": 2495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.43956043956044e-06,
+      "loss": 0.6228,
+      "step": 2496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.442941673710905e-06,
+      "loss": 0.6542,
+      "step": 2497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.446322907861369e-06,
+      "loss": 0.5985,
+      "step": 2498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.449704142011835e-06,
+      "loss": 0.6006,
+      "step": 2499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.4530853761623e-06,
+      "loss": 0.6036,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.456466610312764e-06,
+      "loss": 0.5601,
+      "step": 2501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.459847844463229e-06,
+      "loss": 0.6161,
+      "step": 2502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.463229078613695e-06,
+      "loss": 0.6016,
+      "step": 2503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.46661031276416e-06,
+      "loss": 0.5164,
+      "step": 2504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.469991546914624e-06,
+      "loss": 0.5509,
+      "step": 2505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.473372781065089e-06,
+      "loss": 0.5977,
+      "step": 2506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.476754015215555e-06,
+      "loss": 0.7034,
+      "step": 2507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.48013524936602e-06,
+      "loss": 0.6836,
+      "step": 2508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.483516483516484e-06,
+      "loss": 0.7274,
+      "step": 2509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.486897717666949e-06,
+      "loss": 0.6943,
+      "step": 2510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.490278951817415e-06,
+      "loss": 0.5929,
+      "step": 2511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.49366018596788e-06,
+      "loss": 0.5902,
+      "step": 2512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.497041420118344e-06,
+      "loss": 0.5805,
+      "step": 2513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.500422654268809e-06,
+      "loss": 0.6602,
+      "step": 2514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.503803888419275e-06,
+      "loss": 0.6107,
+      "step": 2515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.50718512256974e-06,
+      "loss": 0.6427,
+      "step": 2516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.510566356720204e-06,
+      "loss": 0.6457,
+      "step": 2517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.513947590870669e-06,
+      "loss": 0.6275,
+      "step": 2518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.517328825021133e-06,
+      "loss": 0.6239,
+      "step": 2519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.5207100591716e-06,
+      "loss": 0.6643,
+      "step": 2520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.524091293322064e-06,
+      "loss": 0.6477,
+      "step": 2521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.527472527472528e-06,
+      "loss": 0.5879,
+      "step": 2522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.530853761622993e-06,
+      "loss": 0.6083,
+      "step": 2523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.534234995773458e-06,
+      "loss": 0.6933,
+      "step": 2524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.537616229923924e-06,
+      "loss": 0.559,
+      "step": 2525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.540997464074388e-06,
+      "loss": 0.571,
+      "step": 2526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.544378698224853e-06,
+      "loss": 0.6221,
+      "step": 2527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.547759932375317e-06,
+      "loss": 0.5828,
+      "step": 2528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.551141166525782e-06,
+      "loss": 0.6675,
+      "step": 2529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.554522400676246e-06,
+      "loss": 0.5884,
+      "step": 2530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.557903634826713e-06,
+      "loss": 0.628,
+      "step": 2531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.561284868977177e-06,
+      "loss": 0.5745,
+      "step": 2532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.564666103127642e-06,
+      "loss": 0.666,
+      "step": 2533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.568047337278106e-06,
+      "loss": 0.6286,
+      "step": 2534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.5777,
+      "step": 2535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.574809805579037e-06,
+      "loss": 0.6212,
+      "step": 2536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.578191039729502e-06,
+      "loss": 0.5586,
+      "step": 2537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.581572273879966e-06,
+      "loss": 0.6118,
+      "step": 2538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.58495350803043e-06,
+      "loss": 0.5076,
+      "step": 2539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.588334742180897e-06,
+      "loss": 0.5843,
+      "step": 2540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.591715976331362e-06,
+      "loss": 0.6256,
+      "step": 2541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.595097210481826e-06,
+      "loss": 0.5185,
+      "step": 2542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.59847844463229e-06,
+      "loss": 0.628,
+      "step": 2543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.601859678782757e-06,
+      "loss": 0.5645,
+      "step": 2544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.605240912933221e-06,
+      "loss": 0.6006,
+      "step": 2545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.608622147083686e-06,
+      "loss": 0.5656,
+      "step": 2546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.61200338123415e-06,
+      "loss": 0.5937,
+      "step": 2547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.615384615384617e-06,
+      "loss": 0.8357,
+      "step": 2548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.618765849535081e-06,
+      "loss": 0.5258,
+      "step": 2549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.622147083685546e-06,
+      "loss": 0.6758,
+      "step": 2550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62552831783601e-06,
+      "loss": 0.6337,
+      "step": 2551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.628909551986477e-06,
+      "loss": 0.6017,
+      "step": 2552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632290786136941e-06,
+      "loss": 0.6159,
+      "step": 2553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.635672020287406e-06,
+      "loss": 0.6458,
+      "step": 2554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63905325443787e-06,
+      "loss": 0.5554,
+      "step": 2555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642434488588337e-06,
+      "loss": 0.6488,
+      "step": 2556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645815722738801e-06,
+      "loss": 0.5638,
+      "step": 2557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649196956889266e-06,
+      "loss": 0.7422,
+      "step": 2558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65257819103973e-06,
+      "loss": 0.5682,
+      "step": 2559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655959425190195e-06,
+      "loss": 0.6588,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659340659340661e-06,
+      "loss": 0.6614,
+      "step": 2561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662721893491126e-06,
+      "loss": 0.5714,
+      "step": 2562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66610312764159e-06,
+      "loss": 0.718,
+      "step": 2563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669484361792055e-06,
+      "loss": 0.6974,
+      "step": 2564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672865595942521e-06,
+      "loss": 0.599,
+      "step": 2565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676246830092985e-06,
+      "loss": 0.6148,
+      "step": 2566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67962806424345e-06,
+      "loss": 0.5302,
+      "step": 2567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683009298393915e-06,
+      "loss": 0.538,
+      "step": 2568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686390532544379e-06,
+      "loss": 0.5523,
+      "step": 2569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.689771766694844e-06,
+      "loss": 0.6988,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69315300084531e-06,
+      "loss": 0.5683,
+      "step": 2571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696534234995774e-06,
+      "loss": 0.6827,
+      "step": 2572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699915469146239e-06,
+      "loss": 0.5722,
+      "step": 2573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703296703296704e-06,
+      "loss": 0.5835,
+      "step": 2574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706677937447168e-06,
+      "loss": 0.6587,
+      "step": 2575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710059171597633e-06,
+      "loss": 0.5876,
+      "step": 2576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.713440405748099e-06,
+      "loss": 0.5648,
+      "step": 2577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.716821639898563e-06,
+      "loss": 0.5529,
+      "step": 2578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720202874049028e-06,
+      "loss": 0.5916,
+      "step": 2579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723584108199493e-06,
+      "loss": 0.6704,
+      "step": 2580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726965342349959e-06,
+      "loss": 0.6183,
+      "step": 2581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.730346576500423e-06,
+      "loss": 0.5833,
+      "step": 2582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.733727810650888e-06,
+      "loss": 0.6695,
+      "step": 2583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.737109044801352e-06,
+      "loss": 0.5385,
+      "step": 2584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.740490278951819e-06,
+      "loss": 0.712,
+      "step": 2585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.743871513102283e-06,
+      "loss": 0.674,
+      "step": 2586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.747252747252748e-06,
+      "loss": 0.5817,
+      "step": 2587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.750633981403212e-06,
+      "loss": 0.6813,
+      "step": 2588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.754015215553677e-06,
+      "loss": 0.5925,
+      "step": 2589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.757396449704143e-06,
+      "loss": 0.6826,
+      "step": 2590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.760777683854608e-06,
+      "loss": 0.6718,
+      "step": 2591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.764158918005072e-06,
+      "loss": 0.6038,
+      "step": 2592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.767540152155537e-06,
+      "loss": 0.5765,
+      "step": 2593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.770921386306003e-06,
+      "loss": 0.5871,
+      "step": 2594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.774302620456468e-06,
+      "loss": 0.6057,
+      "step": 2595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.777683854606932e-06,
+      "loss": 0.615,
+      "step": 2596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.781065088757397e-06,
+      "loss": 0.6192,
+      "step": 2597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.784446322907863e-06,
+      "loss": 0.6176,
+      "step": 2598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.787827557058327e-06,
+      "loss": 0.6514,
+      "step": 2599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.791208791208792e-06,
+      "loss": 0.6386,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.794590025359256e-06,
+      "loss": 0.708,
+      "step": 2601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.797971259509723e-06,
+      "loss": 0.6041,
+      "step": 2602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.801352493660187e-06,
+      "loss": 0.6375,
+      "step": 2603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.804733727810652e-06,
+      "loss": 0.6328,
+      "step": 2604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.808114961961116e-06,
+      "loss": 0.6021,
+      "step": 2605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.811496196111583e-06,
+      "loss": 0.5781,
+      "step": 2606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.814877430262047e-06,
+      "loss": 0.5565,
+      "step": 2607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.818258664412512e-06,
+      "loss": 0.6361,
+      "step": 2608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.821639898562976e-06,
+      "loss": 0.7011,
+      "step": 2609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.82502113271344e-06,
+      "loss": 0.5961,
+      "step": 2610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.828402366863907e-06,
+      "loss": 0.5842,
+      "step": 2611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.831783601014372e-06,
+      "loss": 0.5969,
+      "step": 2612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.835164835164836e-06,
+      "loss": 0.6452,
+      "step": 2613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.8385460693153e-06,
+      "loss": 0.6476,
+      "step": 2614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.841927303465765e-06,
+      "loss": 0.6823,
+      "step": 2615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.845308537616231e-06,
+      "loss": 0.5544,
+      "step": 2616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.848689771766696e-06,
+      "loss": 0.5603,
+      "step": 2617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85207100591716e-06,
+      "loss": 0.5881,
+      "step": 2618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.855452240067625e-06,
+      "loss": 0.6063,
+      "step": 2619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85883347421809e-06,
+      "loss": 0.6164,
+      "step": 2620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.862214708368554e-06,
+      "loss": 0.5315,
+      "step": 2621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.86559594251902e-06,
+      "loss": 0.6191,
+      "step": 2622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.868977176669485e-06,
+      "loss": 0.6794,
+      "step": 2623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.87235841081995e-06,
+      "loss": 0.5627,
+      "step": 2624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.875739644970414e-06,
+      "loss": 0.5694,
+      "step": 2625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.879120879120879e-06,
+      "loss": 0.6133,
+      "step": 2626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.882502113271345e-06,
+      "loss": 0.6509,
+      "step": 2627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.88588334742181e-06,
+      "loss": 0.666,
+      "step": 2628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.889264581572274e-06,
+      "loss": 0.6512,
+      "step": 2629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.892645815722739e-06,
+      "loss": 0.6239,
+      "step": 2630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.896027049873205e-06,
+      "loss": 0.6425,
+      "step": 2631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.89940828402367e-06,
+      "loss": 0.6709,
+      "step": 2632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.902789518174134e-06,
+      "loss": 0.5463,
+      "step": 2633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.906170752324598e-06,
+      "loss": 0.7176,
+      "step": 2634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.909551986475065e-06,
+      "loss": 0.5906,
+      "step": 2635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.91293322062553e-06,
+      "loss": 0.6778,
+      "step": 2636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.916314454775994e-06,
+      "loss": 0.7468,
+      "step": 2637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.919695688926458e-06,
+      "loss": 0.5299,
+      "step": 2638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.923076923076925e-06,
+      "loss": 0.547,
+      "step": 2639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.926458157227389e-06,
+      "loss": 0.5347,
+      "step": 2640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.929839391377854e-06,
+      "loss": 0.6046,
+      "step": 2641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.933220625528318e-06,
+      "loss": 0.6109,
+      "step": 2642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.936601859678784e-06,
+      "loss": 0.6044,
+      "step": 2643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.939983093829249e-06,
+      "loss": 0.549,
+      "step": 2644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.943364327979714e-06,
+      "loss": 0.642,
+      "step": 2645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.946745562130178e-06,
+      "loss": 0.6346,
+      "step": 2646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.950126796280643e-06,
+      "loss": 0.6641,
+      "step": 2647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.953508030431109e-06,
+      "loss": 0.5375,
+      "step": 2648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.956889264581573e-06,
+      "loss": 0.5757,
+      "step": 2649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.960270498732038e-06,
+      "loss": 0.6378,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.963651732882503e-06,
+      "loss": 0.6369,
+      "step": 2651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.967032967032969e-06,
+      "loss": 0.6313,
+      "step": 2652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.970414201183433e-06,
+      "loss": 0.7188,
+      "step": 2653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.973795435333898e-06,
+      "loss": 0.6215,
+      "step": 2654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.977176669484362e-06,
+      "loss": 0.5329,
+      "step": 2655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.980557903634829e-06,
+      "loss": 0.6439,
+      "step": 2656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.983939137785293e-06,
+      "loss": 0.6317,
+      "step": 2657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.987320371935758e-06,
+      "loss": 0.627,
+      "step": 2658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.990701606086222e-06,
+      "loss": 0.6855,
+      "step": 2659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.994082840236687e-06,
+      "loss": 0.5787,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.997464074387151e-06,
+      "loss": 0.6267,
+      "step": 2661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.000845308537618e-06,
+      "loss": 0.5332,
+      "step": 2662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.004226542688082e-06,
+      "loss": 0.6359,
+      "step": 2663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.007607776838547e-06,
+      "loss": 0.6679,
+      "step": 2664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.010989010989011e-06,
+      "loss": 0.648,
+      "step": 2665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.014370245139476e-06,
+      "loss": 0.5096,
+      "step": 2666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.01775147928994e-06,
+      "loss": 0.532,
+      "step": 2667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.021132713440407e-06,
+      "loss": 0.6106,
+      "step": 2668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.024513947590871e-06,
+      "loss": 0.5891,
+      "step": 2669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.027895181741336e-06,
+      "loss": 0.5319,
+      "step": 2670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.0312764158918e-06,
+      "loss": 0.5861,
+      "step": 2671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.034657650042267e-06,
+      "loss": 0.5866,
+      "step": 2672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.038038884192731e-06,
+      "loss": 0.6185,
+      "step": 2673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.041420118343196e-06,
+      "loss": 0.5808,
+      "step": 2674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.04480135249366e-06,
+      "loss": 0.5406,
+      "step": 2675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.048182586644125e-06,
+      "loss": 0.7238,
+      "step": 2676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.051563820794591e-06,
+      "loss": 0.6985,
+      "step": 2677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.054945054945055e-06,
+      "loss": 0.6376,
+      "step": 2678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.05832628909552e-06,
+      "loss": 0.542,
+      "step": 2679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.061707523245985e-06,
+      "loss": 0.6751,
+      "step": 2680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.06508875739645e-06,
+      "loss": 0.6295,
+      "step": 2681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.068469991546915e-06,
+      "loss": 0.5948,
+      "step": 2682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.07185122569738e-06,
+      "loss": 0.6023,
+      "step": 2683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.075232459847844e-06,
+      "loss": 0.737,
+      "step": 2684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.07861369399831e-06,
+      "loss": 0.6327,
+      "step": 2685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.081994928148775e-06,
+      "loss": 0.6122,
+      "step": 2686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.08537616229924e-06,
+      "loss": 0.71,
+      "step": 2687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.088757396449704e-06,
+      "loss": 0.6838,
+      "step": 2688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.09213863060017e-06,
+      "loss": 0.5905,
+      "step": 2689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.095519864750635e-06,
+      "loss": 0.6743,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.0989010989011e-06,
+      "loss": 0.6052,
+      "step": 2691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.102282333051564e-06,
+      "loss": 0.5873,
+      "step": 2692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.10566356720203e-06,
+      "loss": 0.6456,
+      "step": 2693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.109044801352495e-06,
+      "loss": 0.6536,
+      "step": 2694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.11242603550296e-06,
+      "loss": 0.6598,
+      "step": 2695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.115807269653424e-06,
+      "loss": 0.5824,
+      "step": 2696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.11918850380389e-06,
+      "loss": 0.6204,
+      "step": 2697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.122569737954355e-06,
+      "loss": 0.6914,
+      "step": 2698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.12595097210482e-06,
+      "loss": 0.6452,
+      "step": 2699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.129332206255284e-06,
+      "loss": 0.5281,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.132713440405749e-06,
+      "loss": 0.6259,
+      "step": 2701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.136094674556215e-06,
+      "loss": 0.6362,
+      "step": 2702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.13947590870668e-06,
+      "loss": 0.6471,
+      "step": 2703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 0.6335,
+      "step": 2704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.146238377007608e-06,
+      "loss": 0.5477,
+      "step": 2705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.149619611158073e-06,
+      "loss": 0.5447,
+      "step": 2706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.153000845308538e-06,
+      "loss": 0.7051,
+      "step": 2707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.156382079459004e-06,
+      "loss": 0.6012,
+      "step": 2708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.159763313609468e-06,
+      "loss": 0.5935,
+      "step": 2709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.163144547759933e-06,
+      "loss": 0.5388,
+      "step": 2710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.166525781910397e-06,
+      "loss": 0.7839,
+      "step": 2711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.169907016060862e-06,
+      "loss": 0.6735,
+      "step": 2712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.173288250211328e-06,
+      "loss": 0.5513,
+      "step": 2713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.176669484361793e-06,
+      "loss": 0.5726,
+      "step": 2714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.180050718512257e-06,
+      "loss": 0.5732,
+      "step": 2715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.183431952662722e-06,
+      "loss": 0.6713,
+      "step": 2716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.186813186813186e-06,
+      "loss": 0.4929,
+      "step": 2717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.190194420963653e-06,
+      "loss": 0.5254,
+      "step": 2718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.193575655114117e-06,
+      "loss": 0.6959,
+      "step": 2719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.196956889264582e-06,
+      "loss": 0.6601,
+      "step": 2720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.200338123415046e-06,
+      "loss": 0.6151,
+      "step": 2721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.203719357565513e-06,
+      "loss": 0.5444,
+      "step": 2722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.207100591715977e-06,
+      "loss": 0.5757,
+      "step": 2723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.210481825866442e-06,
+      "loss": 0.5865,
+      "step": 2724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.213863060016906e-06,
+      "loss": 0.5931,
+      "step": 2725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.217244294167372e-06,
+      "loss": 0.5999,
+      "step": 2726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.220625528317837e-06,
+      "loss": 0.6098,
+      "step": 2727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.224006762468302e-06,
+      "loss": 0.5256,
+      "step": 2728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.227387996618766e-06,
+      "loss": 0.6515,
+      "step": 2729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.6453,
+      "step": 2730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.234150464919697e-06,
+      "loss": 0.5739,
+      "step": 2731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.237531699070161e-06,
+      "loss": 0.612,
+      "step": 2732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.240912933220626e-06,
+      "loss": 0.5326,
+      "step": 2733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.244294167371092e-06,
+      "loss": 0.5618,
+      "step": 2734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.247675401521557e-06,
+      "loss": 0.7378,
+      "step": 2735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.251056635672021e-06,
+      "loss": 0.6289,
+      "step": 2736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.254437869822486e-06,
+      "loss": 0.6394,
+      "step": 2737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.25781910397295e-06,
+      "loss": 0.5275,
+      "step": 2738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.261200338123417e-06,
+      "loss": 0.5969,
+      "step": 2739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.264581572273881e-06,
+      "loss": 0.6139,
+      "step": 2740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.267962806424346e-06,
+      "loss": 0.62,
+      "step": 2741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.27134404057481e-06,
+      "loss": 0.6928,
+      "step": 2742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.274725274725277e-06,
+      "loss": 0.6149,
+      "step": 2743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.278106508875741e-06,
+      "loss": 0.6233,
+      "step": 2744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.281487743026206e-06,
+      "loss": 0.6379,
+      "step": 2745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.28486897717667e-06,
+      "loss": 0.5971,
+      "step": 2746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.288250211327136e-06,
+      "loss": 0.6146,
+      "step": 2747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.291631445477601e-06,
+      "loss": 0.549,
+      "step": 2748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.295012679628065e-06,
+      "loss": 0.5624,
+      "step": 2749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.29839391377853e-06,
+      "loss": 0.5492,
+      "step": 2750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.301775147928995e-06,
+      "loss": 0.5427,
+      "step": 2751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.305156382079459e-06,
+      "loss": 0.599,
+      "step": 2752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.308537616229925e-06,
+      "loss": 0.5879,
+      "step": 2753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.31191885038039e-06,
+      "loss": 0.5523,
+      "step": 2754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.315300084530854e-06,
+      "loss": 0.6359,
+      "step": 2755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.318681318681319e-06,
+      "loss": 0.5358,
+      "step": 2756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.322062552831784e-06,
+      "loss": 0.6131,
+      "step": 2757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.325443786982248e-06,
+      "loss": 0.5222,
+      "step": 2758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.328825021132714e-06,
+      "loss": 0.66,
+      "step": 2759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.332206255283179e-06,
+      "loss": 0.6737,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.335587489433643e-06,
+      "loss": 0.6211,
+      "step": 2761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.338968723584108e-06,
+      "loss": 0.662,
+      "step": 2762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.342349957734574e-06,
+      "loss": 0.666,
+      "step": 2763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.345731191885039e-06,
+      "loss": 0.5796,
+      "step": 2764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.349112426035503e-06,
+      "loss": 0.5626,
+      "step": 2765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.352493660185968e-06,
+      "loss": 0.5685,
+      "step": 2766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.355874894336432e-06,
+      "loss": 0.5665,
+      "step": 2767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.359256128486899e-06,
+      "loss": 0.6074,
+      "step": 2768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.362637362637363e-06,
+      "loss": 0.6923,
+      "step": 2769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.366018596787828e-06,
+      "loss": 0.5641,
+      "step": 2770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.369399830938292e-06,
+      "loss": 0.6015,
+      "step": 2771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.372781065088759e-06,
+      "loss": 0.6378,
+      "step": 2772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.376162299239223e-06,
+      "loss": 0.6006,
+      "step": 2773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.379543533389688e-06,
+      "loss": 0.5552,
+      "step": 2774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.382924767540152e-06,
+      "loss": 0.6187,
+      "step": 2775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.386306001690618e-06,
+      "loss": 0.6036,
+      "step": 2776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.389687235841083e-06,
+      "loss": 0.6058,
+      "step": 2777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.393068469991548e-06,
+      "loss": 0.4989,
+      "step": 2778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.396449704142012e-06,
+      "loss": 0.6106,
+      "step": 2779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.399830938292478e-06,
+      "loss": 0.6019,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.403212172442943e-06,
+      "loss": 0.6067,
+      "step": 2781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.406593406593407e-06,
+      "loss": 0.6845,
+      "step": 2782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.409974640743872e-06,
+      "loss": 0.6622,
+      "step": 2783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.413355874894338e-06,
+      "loss": 0.6411,
+      "step": 2784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.416737109044803e-06,
+      "loss": 0.5326,
+      "step": 2785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.420118343195267e-06,
+      "loss": 0.6053,
+      "step": 2786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.423499577345732e-06,
+      "loss": 0.6488,
+      "step": 2787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.426880811496198e-06,
+      "loss": 0.6321,
+      "step": 2788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.430262045646663e-06,
+      "loss": 0.5363,
+      "step": 2789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.433643279797127e-06,
+      "loss": 0.5892,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.437024513947592e-06,
+      "loss": 0.6795,
+      "step": 2791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.440405748098056e-06,
+      "loss": 0.6509,
+      "step": 2792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.443786982248523e-06,
+      "loss": 0.6224,
+      "step": 2793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.447168216398987e-06,
+      "loss": 0.5421,
+      "step": 2794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.450549450549452e-06,
+      "loss": 0.7379,
+      "step": 2795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.453930684699916e-06,
+      "loss": 0.6323,
+      "step": 2796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.45731191885038e-06,
+      "loss": 0.6012,
+      "step": 2797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.460693153000845e-06,
+      "loss": 0.6092,
+      "step": 2798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.464074387151312e-06,
+      "loss": 0.7167,
+      "step": 2799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.467455621301776e-06,
+      "loss": 0.6201,
+      "step": 2800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.47083685545224e-06,
+      "loss": 0.6906,
+      "step": 2801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.474218089602705e-06,
+      "loss": 0.5776,
+      "step": 2802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.47759932375317e-06,
+      "loss": 0.5719,
+      "step": 2803
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.480980557903636e-06,
+      "loss": 0.6491,
+      "step": 2804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.4843617920541e-06,
+      "loss": 0.6642,
+      "step": 2805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.487743026204565e-06,
+      "loss": 0.5827,
+      "step": 2806
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.49112426035503e-06,
+      "loss": 0.5956,
+      "step": 2807
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.494505494505494e-06,
+      "loss": 0.5802,
+      "step": 2808
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.49788672865596e-06,
+      "loss": 0.6445,
+      "step": 2809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.501267962806425e-06,
+      "loss": 0.5656,
+      "step": 2810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.50464919695689e-06,
+      "loss": 0.6579,
+      "step": 2811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.508030431107354e-06,
+      "loss": 0.736,
+      "step": 2812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.51141166525782e-06,
+      "loss": 0.5383,
+      "step": 2813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.514792899408285e-06,
+      "loss": 0.5517,
+      "step": 2814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.51817413355875e-06,
+      "loss": 0.715,
+      "step": 2815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.521555367709214e-06,
+      "loss": 0.5374,
+      "step": 2816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.52493660185968e-06,
+      "loss": 0.61,
+      "step": 2817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.528317836010145e-06,
+      "loss": 0.6086,
+      "step": 2818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.53169907016061e-06,
+      "loss": 0.6088,
+      "step": 2819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.535080304311074e-06,
+      "loss": 0.7193,
+      "step": 2820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.53846153846154e-06,
+      "loss": 0.7178,
+      "step": 2821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.541842772612005e-06,
+      "loss": 0.6156,
+      "step": 2822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.545224006762469e-06,
+      "loss": 0.5819,
+      "step": 2823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.548605240912934e-06,
+      "loss": 0.5703,
+      "step": 2824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.551986475063398e-06,
+      "loss": 0.6427,
+      "step": 2825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.555367709213864e-06,
+      "loss": 0.6382,
+      "step": 2826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.558748943364329e-06,
+      "loss": 0.5653,
+      "step": 2827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.562130177514794e-06,
+      "loss": 0.5295,
+      "step": 2828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.565511411665258e-06,
+      "loss": 0.5982,
+      "step": 2829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.568892645815724e-06,
+      "loss": 0.6257,
+      "step": 2830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.572273879966189e-06,
+      "loss": 0.6728,
+      "step": 2831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.575655114116653e-06,
+      "loss": 0.7089,
+      "step": 2832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.579036348267118e-06,
+      "loss": 0.5577,
+      "step": 2833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.582417582417584e-06,
+      "loss": 0.6374,
+      "step": 2834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.585798816568049e-06,
+      "loss": 0.6427,
+      "step": 2835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.589180050718513e-06,
+      "loss": 0.6307,
+      "step": 2836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.592561284868978e-06,
+      "loss": 0.6564,
+      "step": 2837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.595942519019442e-06,
+      "loss": 0.6118,
+      "step": 2838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.599323753169909e-06,
+      "loss": 0.6768,
+      "step": 2839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.602704987320373e-06,
+      "loss": 0.6633,
+      "step": 2840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.606086221470838e-06,
+      "loss": 0.6403,
+      "step": 2841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.609467455621302e-06,
+      "loss": 0.565,
+      "step": 2842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.612848689771767e-06,
+      "loss": 0.5505,
+      "step": 2843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.616229923922233e-06,
+      "loss": 0.6387,
+      "step": 2844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.619611158072698e-06,
+      "loss": 0.5697,
+      "step": 2845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.622992392223162e-06,
+      "loss": 0.5678,
+      "step": 2846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.626373626373627e-06,
+      "loss": 0.6042,
+      "step": 2847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.629754860524091e-06,
+      "loss": 0.595,
+      "step": 2848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.633136094674556e-06,
+      "loss": 0.6459,
+      "step": 2849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.636517328825022e-06,
+      "loss": 0.605,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.639898562975487e-06,
+      "loss": 0.6397,
+      "step": 2851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.643279797125951e-06,
+      "loss": 0.6808,
+      "step": 2852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.646661031276416e-06,
+      "loss": 0.5597,
+      "step": 2853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.65004226542688e-06,
+      "loss": 0.6817,
+      "step": 2854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.653423499577347e-06,
+      "loss": 0.5527,
+      "step": 2855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.656804733727811e-06,
+      "loss": 0.6005,
+      "step": 2856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.660185967878276e-06,
+      "loss": 0.5884,
+      "step": 2857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.66356720202874e-06,
+      "loss": 0.5402,
+      "step": 2858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.666948436179206e-06,
+      "loss": 0.736,
+      "step": 2859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.670329670329671e-06,
+      "loss": 0.5721,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.673710904480136e-06,
+      "loss": 0.6791,
+      "step": 2861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.6770921386306e-06,
+      "loss": 0.5945,
+      "step": 2862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.680473372781066e-06,
+      "loss": 0.5692,
+      "step": 2863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.683854606931531e-06,
+      "loss": 0.6974,
+      "step": 2864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.687235841081995e-06,
+      "loss": 0.5919,
+      "step": 2865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.69061707523246e-06,
+      "loss": 0.541,
+      "step": 2866
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.693998309382926e-06,
+      "loss": 0.6467,
+      "step": 2867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.69737954353339e-06,
+      "loss": 0.5517,
+      "step": 2868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.700760777683855e-06,
+      "loss": 0.6814,
+      "step": 2869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.70414201183432e-06,
+      "loss": 0.7472,
+      "step": 2870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.707523245984786e-06,
+      "loss": 0.6332,
+      "step": 2871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.71090448013525e-06,
+      "loss": 0.5125,
+      "step": 2872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 0.6621,
+      "step": 2873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.71766694843618e-06,
+      "loss": 0.6379,
+      "step": 2874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.721048182586646e-06,
+      "loss": 0.5663,
+      "step": 2875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.72442941673711e-06,
+      "loss": 0.5952,
+      "step": 2876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.727810650887575e-06,
+      "loss": 0.6058,
+      "step": 2877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.73119188503804e-06,
+      "loss": 0.6037,
+      "step": 2878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.734573119188506e-06,
+      "loss": 0.6447,
+      "step": 2879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.73795435333897e-06,
+      "loss": 0.6795,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.741335587489435e-06,
+      "loss": 0.5395,
+      "step": 2881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.7447168216399e-06,
+      "loss": 0.6561,
+      "step": 2882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.748098055790364e-06,
+      "loss": 0.6395,
+      "step": 2883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.75147928994083e-06,
+      "loss": 0.5822,
+      "step": 2884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.754860524091295e-06,
+      "loss": 0.5803,
+      "step": 2885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.75824175824176e-06,
+      "loss": 0.6317,
+      "step": 2886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.761622992392224e-06,
+      "loss": 0.6615,
+      "step": 2887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.765004226542688e-06,
+      "loss": 0.5672,
+      "step": 2888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.768385460693153e-06,
+      "loss": 0.6223,
+      "step": 2889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.77176669484362e-06,
+      "loss": 0.5811,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.775147928994084e-06,
+      "loss": 0.7139,
+      "step": 2891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.778529163144548e-06,
+      "loss": 0.6172,
+      "step": 2892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.781910397295013e-06,
+      "loss": 0.6184,
+      "step": 2893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.785291631445477e-06,
+      "loss": 0.7169,
+      "step": 2894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.788672865595944e-06,
+      "loss": 0.6346,
+      "step": 2895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.792054099746408e-06,
+      "loss": 0.6204,
+      "step": 2896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.795435333896873e-06,
+      "loss": 0.6989,
+      "step": 2897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.798816568047337e-06,
+      "loss": 0.6487,
+      "step": 2898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.802197802197802e-06,
+      "loss": 0.6746,
+      "step": 2899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.805579036348268e-06,
+      "loss": 0.5587,
+      "step": 2900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.808960270498733e-06,
+      "loss": 0.6927,
+      "step": 2901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.812341504649197e-06,
+      "loss": 0.666,
+      "step": 2902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.815722738799662e-06,
+      "loss": 0.5682,
+      "step": 2903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.819103972950128e-06,
+      "loss": 0.6922,
+      "step": 2904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.822485207100593e-06,
+      "loss": 0.5917,
+      "step": 2905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.825866441251057e-06,
+      "loss": 0.6694,
+      "step": 2906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.829247675401522e-06,
+      "loss": 0.5843,
+      "step": 2907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.832628909551988e-06,
+      "loss": 0.6295,
+      "step": 2908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.836010143702452e-06,
+      "loss": 0.5324,
+      "step": 2909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.839391377852917e-06,
+      "loss": 0.601,
+      "step": 2910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.842772612003382e-06,
+      "loss": 0.678,
+      "step": 2911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.846153846153848e-06,
+      "loss": 0.6912,
+      "step": 2912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.849535080304312e-06,
+      "loss": 0.6189,
+      "step": 2913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.852916314454777e-06,
+      "loss": 0.5909,
+      "step": 2914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.856297548605241e-06,
+      "loss": 0.6509,
+      "step": 2915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.859678782755706e-06,
+      "loss": 0.5788,
+      "step": 2916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.863060016906172e-06,
+      "loss": 0.5871,
+      "step": 2917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.866441251056637e-06,
+      "loss": 0.7941,
+      "step": 2918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.869822485207101e-06,
+      "loss": 0.6614,
+      "step": 2919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.873203719357566e-06,
+      "loss": 0.6907,
+      "step": 2920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.876584953508032e-06,
+      "loss": 0.6495,
+      "step": 2921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.879966187658497e-06,
+      "loss": 0.5411,
+      "step": 2922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.883347421808961e-06,
+      "loss": 0.5768,
+      "step": 2923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.886728655959426e-06,
+      "loss": 0.6763,
+      "step": 2924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.890109890109892e-06,
+      "loss": 0.5877,
+      "step": 2925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.893491124260357e-06,
+      "loss": 0.5758,
+      "step": 2926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.896872358410821e-06,
+      "loss": 0.7004,
+      "step": 2927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.900253592561286e-06,
+      "loss": 0.534,
+      "step": 2928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.90363482671175e-06,
+      "loss": 0.6163,
+      "step": 2929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.907016060862216e-06,
+      "loss": 0.6419,
+      "step": 2930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.910397295012681e-06,
+      "loss": 0.6031,
+      "step": 2931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.913778529163146e-06,
+      "loss": 0.598,
+      "step": 2932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.91715976331361e-06,
+      "loss": 0.6343,
+      "step": 2933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.920540997464075e-06,
+      "loss": 0.5596,
+      "step": 2934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.923922231614541e-06,
+      "loss": 0.6167,
+      "step": 2935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.927303465765005e-06,
+      "loss": 0.632,
+      "step": 2936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.93068469991547e-06,
+      "loss": 0.6639,
+      "step": 2937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.934065934065935e-06,
+      "loss": 0.5615,
+      "step": 2938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.937447168216399e-06,
+      "loss": 0.6744,
+      "step": 2939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.940828402366864e-06,
+      "loss": 0.5671,
+      "step": 2940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.94420963651733e-06,
+      "loss": 0.5338,
+      "step": 2941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.947590870667794e-06,
+      "loss": 0.5998,
+      "step": 2942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.950972104818259e-06,
+      "loss": 0.6273,
+      "step": 2943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.954353338968723e-06,
+      "loss": 0.5306,
+      "step": 2944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.957734573119188e-06,
+      "loss": 0.5892,
+      "step": 2945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.961115807269654e-06,
+      "loss": 0.6728,
+      "step": 2946
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.964497041420119e-06,
+      "loss": 0.6746,
+      "step": 2947
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.967878275570583e-06,
+      "loss": 0.5963,
+      "step": 2948
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.971259509721048e-06,
+      "loss": 0.5527,
+      "step": 2949
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.974640743871514e-06,
+      "loss": 0.5901,
+      "step": 2950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.978021978021979e-06,
+      "loss": 0.644,
+      "step": 2951
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.981403212172443e-06,
+      "loss": 0.5997,
+      "step": 2952
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.984784446322908e-06,
+      "loss": 0.6002,
+      "step": 2953
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.988165680473374e-06,
+      "loss": 0.5754,
+      "step": 2954
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.991546914623839e-06,
+      "loss": 0.6646,
+      "step": 2955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.994928148774303e-06,
+      "loss": 0.5584,
+      "step": 2956
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.998309382924768e-06,
+      "loss": 0.6562,
+      "step": 2957
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0001690617075232e-05,
+      "loss": 0.6416,
+      "step": 2958
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0005071851225698e-05,
+      "loss": 0.66,
+      "step": 2959
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0008453085376165e-05,
+      "loss": 0.6564,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0011834319526628e-05,
+      "loss": 0.6442,
+      "step": 2961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0015215553677094e-05,
+      "loss": 0.5616,
+      "step": 2962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0018596787827557e-05,
+      "loss": 0.6403,
+      "step": 2963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0021978021978023e-05,
+      "loss": 0.5487,
+      "step": 2964
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.002535925612849e-05,
+      "loss": 0.6266,
+      "step": 2965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0028740490278952e-05,
+      "loss": 0.5971,
+      "step": 2966
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0032121724429418e-05,
+      "loss": 0.6217,
+      "step": 2967
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0035502958579883e-05,
+      "loss": 0.6325,
+      "step": 2968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0038884192730347e-05,
+      "loss": 0.7349,
+      "step": 2969
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0042265426880814e-05,
+      "loss": 0.5752,
+      "step": 2970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0045646661031276e-05,
+      "loss": 0.5472,
+      "step": 2971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0049027895181743e-05,
+      "loss": 0.6302,
+      "step": 2972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0052409129332207e-05,
+      "loss": 0.66,
+      "step": 2973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0055790363482672e-05,
+      "loss": 0.8495,
+      "step": 2974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0059171597633138e-05,
+      "loss": 0.6155,
+      "step": 2975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0062552831783601e-05,
+      "loss": 0.5274,
+      "step": 2976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0065934065934067e-05,
+      "loss": 0.5852,
+      "step": 2977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0069315300084532e-05,
+      "loss": 0.6324,
+      "step": 2978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0072696534234996e-05,
+      "loss": 0.5734,
+      "step": 2979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.007607776838546e-05,
+      "loss": 0.5834,
+      "step": 2980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0079459002535927e-05,
+      "loss": 0.5165,
+      "step": 2981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0082840236686392e-05,
+      "loss": 0.795,
+      "step": 2982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0086221470836856e-05,
+      "loss": 0.5676,
+      "step": 2983
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.008960270498732e-05,
+      "loss": 0.5589,
+      "step": 2984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0092983939137785e-05,
+      "loss": 0.5944,
+      "step": 2985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0096365173288251e-05,
+      "loss": 0.5909,
+      "step": 2986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0099746407438716e-05,
+      "loss": 0.5558,
+      "step": 2987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.010312764158918e-05,
+      "loss": 0.6013,
+      "step": 2988
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0106508875739647e-05,
+      "loss": 0.5446,
+      "step": 2989
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.010989010989011e-05,
+      "loss": 0.5378,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0113271344040576e-05,
+      "loss": 0.6377,
+      "step": 2991
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.011665257819104e-05,
+      "loss": 0.5554,
+      "step": 2992
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0120033812341505e-05,
+      "loss": 0.6497,
+      "step": 2993
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0123415046491971e-05,
+      "loss": 0.6235,
+      "step": 2994
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0126796280642434e-05,
+      "loss": 0.5205,
+      "step": 2995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.01301775147929e-05,
+      "loss": 0.5709,
+      "step": 2996
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0133558748943367e-05,
+      "loss": 0.543,
+      "step": 2997
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.013693998309383e-05,
+      "loss": 0.6261,
+      "step": 2998
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0140321217244296e-05,
+      "loss": 0.5747,
+      "step": 2999
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0143702451394759e-05,
+      "loss": 0.6508,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0147083685545225e-05,
+      "loss": 0.6471,
+      "step": 3001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0150464919695691e-05,
+      "loss": 0.5547,
+      "step": 3002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0153846153846154e-05,
+      "loss": 0.6525,
+      "step": 3003
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.015722738799662e-05,
+      "loss": 0.5145,
+      "step": 3004
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0160608622147083e-05,
+      "loss": 0.5818,
+      "step": 3005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.016398985629755e-05,
+      "loss": 0.5456,
+      "step": 3006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0167371090448015e-05,
+      "loss": 0.6113,
+      "step": 3007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0170752324598478e-05,
+      "loss": 0.6356,
+      "step": 3008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0174133558748945e-05,
+      "loss": 0.5621,
+      "step": 3009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.017751479289941e-05,
+      "loss": 0.5734,
+      "step": 3010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0180896027049874e-05,
+      "loss": 0.5799,
+      "step": 3011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.018427726120034e-05,
+      "loss": 0.6224,
+      "step": 3012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0187658495350803e-05,
+      "loss": 0.6475,
+      "step": 3013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0191039729501269e-05,
+      "loss": 0.6747,
+      "step": 3014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0194420963651735e-05,
+      "loss": 0.6386,
+      "step": 3015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0197802197802198e-05,
+      "loss": 0.5887,
+      "step": 3016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0201183431952664e-05,
+      "loss": 0.579,
+      "step": 3017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0204564666103129e-05,
+      "loss": 0.7107,
+      "step": 3018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0207945900253593e-05,
+      "loss": 0.6536,
+      "step": 3019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0211327134404058e-05,
+      "loss": 0.6804,
+      "step": 3020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0214708368554522e-05,
+      "loss": 0.5909,
+      "step": 3021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0218089602704989e-05,
+      "loss": 0.5574,
+      "step": 3022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0221470836855453e-05,
+      "loss": 0.5774,
+      "step": 3023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0224852071005918e-05,
+      "loss": 0.5432,
+      "step": 3024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0228233305156382e-05,
+      "loss": 0.5927,
+      "step": 3025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0231614539306849e-05,
+      "loss": 0.6319,
+      "step": 3026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0234995773457313e-05,
+      "loss": 0.557,
+      "step": 3027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0238377007607778e-05,
+      "loss": 0.5688,
+      "step": 3028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0241758241758242e-05,
+      "loss": 0.7162,
+      "step": 3029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0245139475908707e-05,
+      "loss": 0.5623,
+      "step": 3030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0248520710059173e-05,
+      "loss": 0.5792,
+      "step": 3031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0251901944209638e-05,
+      "loss": 0.6052,
+      "step": 3032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0255283178360102e-05,
+      "loss": 0.5594,
+      "step": 3033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0258664412510567e-05,
+      "loss": 0.5326,
+      "step": 3034
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0262045646661031e-05,
+      "loss": 0.5704,
+      "step": 3035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0265426880811497e-05,
+      "loss": 0.6177,
+      "step": 3036
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.026880811496196e-05,
+      "loss": 0.5636,
+      "step": 3037
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0272189349112427e-05,
+      "loss": 0.5722,
+      "step": 3038
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0275570583262893e-05,
+      "loss": 0.5917,
+      "step": 3039
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0278951817413356e-05,
+      "loss": 0.7338,
+      "step": 3040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0282333051563822e-05,
+      "loss": 0.6258,
+      "step": 3041
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0285714285714285e-05,
+      "loss": 0.506,
+      "step": 3042
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0289095519864751e-05,
+      "loss": 0.5351,
+      "step": 3043
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0292476754015217e-05,
+      "loss": 0.6464,
+      "step": 3044
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.029585798816568e-05,
+      "loss": 0.6939,
+      "step": 3045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0299239222316146e-05,
+      "loss": 0.56,
+      "step": 3046
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0302620456466613e-05,
+      "loss": 0.6791,
+      "step": 3047
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0306001690617075e-05,
+      "loss": 0.6323,
+      "step": 3048
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0309382924767542e-05,
+      "loss": 0.6484,
+      "step": 3049
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0312764158918005e-05,
+      "loss": 0.5291,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.031614539306847e-05,
+      "loss": 0.6411,
+      "step": 3051
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0319526627218937e-05,
+      "loss": 0.6126,
+      "step": 3052
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.03229078613694e-05,
+      "loss": 0.5862,
+      "step": 3053
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0326289095519866e-05,
+      "loss": 0.6502,
+      "step": 3054
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0329670329670332e-05,
+      "loss": 0.5778,
+      "step": 3055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0333051563820795e-05,
+      "loss": 0.6542,
+      "step": 3056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0336432797971261e-05,
+      "loss": 0.6194,
+      "step": 3057
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0339814032121724e-05,
+      "loss": 0.6631,
+      "step": 3058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.034319526627219e-05,
+      "loss": 0.6114,
+      "step": 3059
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0346576500422655e-05,
+      "loss": 0.6452,
+      "step": 3060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.034995773457312e-05,
+      "loss": 0.5947,
+      "step": 3061
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0353338968723586e-05,
+      "loss": 0.5799,
+      "step": 3062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0356720202874049e-05,
+      "loss": 0.597,
+      "step": 3063
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0360101437024515e-05,
+      "loss": 0.7367,
+      "step": 3064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.036348267117498e-05,
+      "loss": 0.5473,
+      "step": 3065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0366863905325444e-05,
+      "loss": 0.5937,
+      "step": 3066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.037024513947591e-05,
+      "loss": 0.7135,
+      "step": 3067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0373626373626375e-05,
+      "loss": 0.6151,
+      "step": 3068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.037700760777684e-05,
+      "loss": 0.5977,
+      "step": 3069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0380388841927304e-05,
+      "loss": 0.565,
+      "step": 3070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0383770076077769e-05,
+      "loss": 0.6998,
+      "step": 3071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0387151310228235e-05,
+      "loss": 0.6,
+      "step": 3072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.03905325443787e-05,
+      "loss": 0.5622,
+      "step": 3073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0393913778529164e-05,
+      "loss": 0.6363,
+      "step": 3074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0397295012679628e-05,
+      "loss": 0.6059,
+      "step": 3075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0400676246830095e-05,
+      "loss": 0.5763,
+      "step": 3076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.040405748098056e-05,
+      "loss": 0.548,
+      "step": 3077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0407438715131024e-05,
+      "loss": 0.589,
+      "step": 3078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0410819949281488e-05,
+      "loss": 0.6247,
+      "step": 3079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0414201183431953e-05,
+      "loss": 0.5459,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0417582417582419e-05,
+      "loss": 0.6119,
+      "step": 3081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0420963651732882e-05,
+      "loss": 0.5867,
+      "step": 3082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0424344885883348e-05,
+      "loss": 0.6196,
+      "step": 3083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0427726120033814e-05,
+      "loss": 0.6234,
+      "step": 3084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0431107354184277e-05,
+      "loss": 0.6212,
+      "step": 3085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0434488588334744e-05,
+      "loss": 0.631,
+      "step": 3086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0437869822485206e-05,
+      "loss": 0.5126,
+      "step": 3087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0441251056635673e-05,
+      "loss": 0.6355,
+      "step": 3088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0444632290786139e-05,
+      "loss": 0.5755,
+      "step": 3089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0448013524936602e-05,
+      "loss": 0.5569,
+      "step": 3090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0451394759087068e-05,
+      "loss": 0.6604,
+      "step": 3091
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.045477599323753e-05,
+      "loss": 0.5751,
+      "step": 3092
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0458157227387997e-05,
+      "loss": 0.6643,
+      "step": 3093
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0461538461538463e-05,
+      "loss": 0.5756,
+      "step": 3094
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0464919695688926e-05,
+      "loss": 0.6185,
+      "step": 3095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0468300929839392e-05,
+      "loss": 0.6127,
+      "step": 3096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0471682163989859e-05,
+      "loss": 0.6669,
+      "step": 3097
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0475063398140321e-05,
+      "loss": 0.564,
+      "step": 3098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0478444632290788e-05,
+      "loss": 0.5579,
+      "step": 3099
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.048182586644125e-05,
+      "loss": 0.6477,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0485207100591717e-05,
+      "loss": 0.614,
+      "step": 3101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0488588334742183e-05,
+      "loss": 0.56,
+      "step": 3102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0491969568892646e-05,
+      "loss": 0.6601,
+      "step": 3103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0495350803043112e-05,
+      "loss": 0.7692,
+      "step": 3104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0498732037193577e-05,
+      "loss": 0.6283,
+      "step": 3105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0502113271344041e-05,
+      "loss": 0.5731,
+      "step": 3106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0505494505494507e-05,
+      "loss": 0.6136,
+      "step": 3107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.050887573964497e-05,
+      "loss": 0.7268,
+      "step": 3108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0512256973795437e-05,
+      "loss": 0.5526,
+      "step": 3109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0515638207945901e-05,
+      "loss": 0.5669,
+      "step": 3110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0519019442096366e-05,
+      "loss": 0.6052,
+      "step": 3111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0522400676246832e-05,
+      "loss": 0.5738,
+      "step": 3112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0525781910397296e-05,
+      "loss": 0.6315,
+      "step": 3113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0529163144547761e-05,
+      "loss": 0.6092,
+      "step": 3114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0532544378698226e-05,
+      "loss": 0.6149,
+      "step": 3115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.053592561284869e-05,
+      "loss": 0.5828,
+      "step": 3116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0539306846999156e-05,
+      "loss": 0.6827,
+      "step": 3117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0542688081149621e-05,
+      "loss": 0.6253,
+      "step": 3118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0546069315300085e-05,
+      "loss": 0.7038,
+      "step": 3119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.054945054945055e-05,
+      "loss": 0.6482,
+      "step": 3120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0552831783601015e-05,
+      "loss": 0.646,
+      "step": 3121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0556213017751479e-05,
+      "loss": 0.6808,
+      "step": 3122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0559594251901945e-05,
+      "loss": 0.5047,
+      "step": 3123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.056297548605241e-05,
+      "loss": 0.6153,
+      "step": 3124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0566356720202874e-05,
+      "loss": 0.5876,
+      "step": 3125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.056973795435334e-05,
+      "loss": 0.6591,
+      "step": 3126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0573119188503804e-05,
+      "loss": 0.6931,
+      "step": 3127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.057650042265427e-05,
+      "loss": 0.6607,
+      "step": 3128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0579881656804734e-05,
+      "loss": 0.6067,
+      "step": 3129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0583262890955199e-05,
+      "loss": 0.6015,
+      "step": 3130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0586644125105665e-05,
+      "loss": 0.568,
+      "step": 3131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0590025359256128e-05,
+      "loss": 0.534,
+      "step": 3132
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0593406593406594e-05,
+      "loss": 0.5952,
+      "step": 3133
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.059678782755706e-05,
+      "loss": 0.5938,
+      "step": 3134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0600169061707523e-05,
+      "loss": 0.7326,
+      "step": 3135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.060355029585799e-05,
+      "loss": 0.6455,
+      "step": 3136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0606931530008452e-05,
+      "loss": 0.5733,
+      "step": 3137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0610312764158919e-05,
+      "loss": 0.7367,
+      "step": 3138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0613693998309385e-05,
+      "loss": 0.5961,
+      "step": 3139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0617075232459848e-05,
+      "loss": 0.7577,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0620456466610314e-05,
+      "loss": 0.6592,
+      "step": 3141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.062383770076078e-05,
+      "loss": 0.5895,
+      "step": 3142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0627218934911243e-05,
+      "loss": 0.7594,
+      "step": 3143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.063060016906171e-05,
+      "loss": 0.5115,
+      "step": 3144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0633981403212172e-05,
+      "loss": 0.5634,
+      "step": 3145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0637362637362638e-05,
+      "loss": 0.5332,
+      "step": 3146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0640743871513105e-05,
+      "loss": 0.6277,
+      "step": 3147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0644125105663568e-05,
+      "loss": 0.6137,
+      "step": 3148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0647506339814034e-05,
+      "loss": 0.6987,
+      "step": 3149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0650887573964498e-05,
+      "loss": 0.6157,
+      "step": 3150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0654268808114963e-05,
+      "loss": 0.5609,
+      "step": 3151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0657650042265429e-05,
+      "loss": 0.6226,
+      "step": 3152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0661031276415892e-05,
+      "loss": 0.6014,
+      "step": 3153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0664412510566358e-05,
+      "loss": 0.5543,
+      "step": 3154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0667793744716823e-05,
+      "loss": 0.5527,
+      "step": 3155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0671174978867287e-05,
+      "loss": 0.6032,
+      "step": 3156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0674556213017754e-05,
+      "loss": 0.5506,
+      "step": 3157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0677937447168216e-05,
+      "loss": 0.5694,
+      "step": 3158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0681318681318683e-05,
+      "loss": 0.582,
+      "step": 3159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0684699915469147e-05,
+      "loss": 0.5501,
+      "step": 3160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0688081149619612e-05,
+      "loss": 0.5765,
+      "step": 3161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0691462383770076e-05,
+      "loss": 0.5866,
+      "step": 3162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0694843617920542e-05,
+      "loss": 0.6948,
+      "step": 3163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0698224852071007e-05,
+      "loss": 0.5552,
+      "step": 3164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0701606086221472e-05,
+      "loss": 0.5707,
+      "step": 3165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0704987320371936e-05,
+      "loss": 0.6572,
+      "step": 3166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.07083685545224e-05,
+      "loss": 0.616,
+      "step": 3167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0711749788672867e-05,
+      "loss": 0.5747,
+      "step": 3168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0715131022823331e-05,
+      "loss": 0.5341,
+      "step": 3169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0718512256973796e-05,
+      "loss": 0.6333,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0721893491124262e-05,
+      "loss": 0.6372,
+      "step": 3171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0725274725274725e-05,
+      "loss": 0.607,
+      "step": 3172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0728655959425191e-05,
+      "loss": 0.717,
+      "step": 3173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0732037193575656e-05,
+      "loss": 0.6416,
+      "step": 3174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.073541842772612e-05,
+      "loss": 0.6964,
+      "step": 3175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0738799661876587e-05,
+      "loss": 0.6006,
+      "step": 3176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.074218089602705e-05,
+      "loss": 0.5476,
+      "step": 3177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0745562130177516e-05,
+      "loss": 0.5751,
+      "step": 3178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0748943364327982e-05,
+      "loss": 0.7175,
+      "step": 3179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0752324598478445e-05,
+      "loss": 0.6073,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0755705832628911e-05,
+      "loss": 0.531,
+      "step": 3181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0759087066779374e-05,
+      "loss": 0.5602,
+      "step": 3182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.076246830092984e-05,
+      "loss": 0.5886,
+      "step": 3183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0765849535080306e-05,
+      "loss": 0.5994,
+      "step": 3184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.5889,
+      "step": 3185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0772612003381236e-05,
+      "loss": 0.5893,
+      "step": 3186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0775993237531698e-05,
+      "loss": 0.6635,
+      "step": 3187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0779374471682165e-05,
+      "loss": 0.6142,
+      "step": 3188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0782755705832631e-05,
+      "loss": 0.5295,
+      "step": 3189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0786136939983094e-05,
+      "loss": 0.6086,
+      "step": 3190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.078951817413356e-05,
+      "loss": 0.6265,
+      "step": 3191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0792899408284026e-05,
+      "loss": 0.5848,
+      "step": 3192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0796280642434489e-05,
+      "loss": 0.6869,
+      "step": 3193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0799661876584955e-05,
+      "loss": 0.5848,
+      "step": 3194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0803043110735418e-05,
+      "loss": 0.6157,
+      "step": 3195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0806424344885884e-05,
+      "loss": 0.6086,
+      "step": 3196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.080980557903635e-05,
+      "loss": 0.5699,
+      "step": 3197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0813186813186814e-05,
+      "loss": 0.6269,
+      "step": 3198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.081656804733728e-05,
+      "loss": 0.6452,
+      "step": 3199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0819949281487744e-05,
+      "loss": 0.6402,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0823330515638209e-05,
+      "loss": 0.6484,
+      "step": 3201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0826711749788673e-05,
+      "loss": 0.5684,
+      "step": 3202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0830092983939138e-05,
+      "loss": 0.5943,
+      "step": 3203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0833474218089604e-05,
+      "loss": 0.6149,
+      "step": 3204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0836855452240069e-05,
+      "loss": 0.6194,
+      "step": 3205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0840236686390533e-05,
+      "loss": 0.6629,
+      "step": 3206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0843617920540998e-05,
+      "loss": 0.647,
+      "step": 3207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0846999154691464e-05,
+      "loss": 0.5333,
+      "step": 3208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0850380388841929e-05,
+      "loss": 0.5603,
+      "step": 3209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0853761622992393e-05,
+      "loss": 0.6099,
+      "step": 3210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 0.645,
+      "step": 3211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0860524091293322e-05,
+      "loss": 0.6143,
+      "step": 3212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0863905325443789e-05,
+      "loss": 0.6084,
+      "step": 3213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0867286559594253e-05,
+      "loss": 0.6317,
+      "step": 3214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0870667793744718e-05,
+      "loss": 0.5806,
+      "step": 3215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0874049027895182e-05,
+      "loss": 0.64,
+      "step": 3216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0877430262045647e-05,
+      "loss": 0.6271,
+      "step": 3217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0880811496196113e-05,
+      "loss": 0.6251,
+      "step": 3218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0884192730346576e-05,
+      "loss": 0.6128,
+      "step": 3219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0887573964497042e-05,
+      "loss": 0.6915,
+      "step": 3220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0890955198647508e-05,
+      "loss": 0.5899,
+      "step": 3221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0894336432797971e-05,
+      "loss": 0.6457,
+      "step": 3222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0897717666948437e-05,
+      "loss": 0.5872,
+      "step": 3223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.09010989010989e-05,
+      "loss": 0.5934,
+      "step": 3224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0904480135249366e-05,
+      "loss": 0.6419,
+      "step": 3225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0907861369399833e-05,
+      "loss": 0.4983,
+      "step": 3226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0911242603550296e-05,
+      "loss": 0.6296,
+      "step": 3227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0914623837700762e-05,
+      "loss": 0.6172,
+      "step": 3228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0918005071851228e-05,
+      "loss": 0.7044,
+      "step": 3229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0921386306001691e-05,
+      "loss": 0.618,
+      "step": 3230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0924767540152157e-05,
+      "loss": 0.62,
+      "step": 3231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.092814877430262e-05,
+      "loss": 0.6032,
+      "step": 3232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0931530008453086e-05,
+      "loss": 0.6352,
+      "step": 3233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0934911242603553e-05,
+      "loss": 0.5408,
+      "step": 3234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0938292476754015e-05,
+      "loss": 0.6045,
+      "step": 3235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0941673710904482e-05,
+      "loss": 0.5799,
+      "step": 3236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0945054945054948e-05,
+      "loss": 0.6722,
+      "step": 3237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.094843617920541e-05,
+      "loss": 0.7065,
+      "step": 3238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0951817413355877e-05,
+      "loss": 0.734,
+      "step": 3239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.095519864750634e-05,
+      "loss": 0.6441,
+      "step": 3240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0958579881656806e-05,
+      "loss": 0.6745,
+      "step": 3241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.096196111580727e-05,
+      "loss": 0.5781,
+      "step": 3242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0965342349957735e-05,
+      "loss": 0.5828,
+      "step": 3243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0968723584108201e-05,
+      "loss": 0.5832,
+      "step": 3244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0972104818258664e-05,
+      "loss": 0.7458,
+      "step": 3245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.097548605240913e-05,
+      "loss": 0.5886,
+      "step": 3246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0978867286559595e-05,
+      "loss": 0.5881,
+      "step": 3247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.098224852071006e-05,
+      "loss": 0.6681,
+      "step": 3248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0985629754860526e-05,
+      "loss": 0.5931,
+      "step": 3249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.098901098901099e-05,
+      "loss": 0.6819,
+      "step": 3250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0992392223161455e-05,
+      "loss": 0.5533,
+      "step": 3251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.099577345731192e-05,
+      "loss": 0.5968,
+      "step": 3252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0999154691462384e-05,
+      "loss": 0.5745,
+      "step": 3253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.100253592561285e-05,
+      "loss": 0.5488,
+      "step": 3254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1005917159763315e-05,
+      "loss": 0.7016,
+      "step": 3255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.100929839391378e-05,
+      "loss": 0.6601,
+      "step": 3256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1012679628064244e-05,
+      "loss": 0.5485,
+      "step": 3257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.101606086221471e-05,
+      "loss": 0.6307,
+      "step": 3258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1019442096365173e-05,
+      "loss": 0.5398,
+      "step": 3259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.102282333051564e-05,
+      "loss": 0.6727,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1026204564666104e-05,
+      "loss": 0.5864,
+      "step": 3261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1029585798816568e-05,
+      "loss": 0.5693,
+      "step": 3262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1032967032967035e-05,
+      "loss": 0.7491,
+      "step": 3263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1036348267117497e-05,
+      "loss": 0.6048,
+      "step": 3264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1039729501267964e-05,
+      "loss": 0.5913,
+      "step": 3265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.104311073541843e-05,
+      "loss": 0.6499,
+      "step": 3266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1046491969568893e-05,
+      "loss": 0.5764,
+      "step": 3267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1049873203719359e-05,
+      "loss": 0.5984,
+      "step": 3268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1053254437869822e-05,
+      "loss": 0.5407,
+      "step": 3269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1056635672020288e-05,
+      "loss": 0.5431,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1060016906170754e-05,
+      "loss": 0.5628,
+      "step": 3271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1063398140321217e-05,
+      "loss": 0.6047,
+      "step": 3272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1066779374471683e-05,
+      "loss": 0.5414,
+      "step": 3273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1070160608622146e-05,
+      "loss": 0.7191,
+      "step": 3274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1073541842772613e-05,
+      "loss": 0.6133,
+      "step": 3275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1076923076923079e-05,
+      "loss": 0.526,
+      "step": 3276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1080304311073542e-05,
+      "loss": 0.6851,
+      "step": 3277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1083685545224008e-05,
+      "loss": 0.6659,
+      "step": 3278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1087066779374474e-05,
+      "loss": 0.6449,
+      "step": 3279
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1090448013524937e-05,
+      "loss": 0.5825,
+      "step": 3280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1093829247675403e-05,
+      "loss": 0.5977,
+      "step": 3281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1097210481825866e-05,
+      "loss": 0.6888,
+      "step": 3282
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1100591715976332e-05,
+      "loss": 0.5287,
+      "step": 3283
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1103972950126799e-05,
+      "loss": 0.6269,
+      "step": 3284
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1107354184277261e-05,
+      "loss": 0.5704,
+      "step": 3285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1110735418427728e-05,
+      "loss": 0.6233,
+      "step": 3286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1114116652578192e-05,
+      "loss": 0.5836,
+      "step": 3287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1117497886728657e-05,
+      "loss": 0.5259,
+      "step": 3288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1120879120879123e-05,
+      "loss": 0.5784,
+      "step": 3289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1124260355029586e-05,
+      "loss": 0.6632,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1127641589180052e-05,
+      "loss": 0.5739,
+      "step": 3291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1131022823330517e-05,
+      "loss": 0.5453,
+      "step": 3292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1134404057480981e-05,
+      "loss": 0.521,
+      "step": 3293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1137785291631447e-05,
+      "loss": 0.6224,
+      "step": 3294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1141166525781912e-05,
+      "loss": 0.5826,
+      "step": 3295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1144547759932377e-05,
+      "loss": 0.6055,
+      "step": 3296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1147928994082841e-05,
+      "loss": 0.6708,
+      "step": 3297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1151310228233306e-05,
+      "loss": 0.6267,
+      "step": 3298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.115469146238377e-05,
+      "loss": 0.5593,
+      "step": 3299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1158072696534236e-05,
+      "loss": 0.6007,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1161453930684701e-05,
+      "loss": 0.6057,
+      "step": 3301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1164835164835165e-05,
+      "loss": 0.5895,
+      "step": 3302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.116821639898563e-05,
+      "loss": 0.5055,
+      "step": 3303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1171597633136095e-05,
+      "loss": 0.5371,
+      "step": 3304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.117497886728656e-05,
+      "loss": 0.5402,
+      "step": 3305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1178360101437025e-05,
+      "loss": 0.5369,
+      "step": 3306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.118174133558749e-05,
+      "loss": 0.601,
+      "step": 3307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1185122569737956e-05,
+      "loss": 0.5894,
+      "step": 3308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1188503803888419e-05,
+      "loss": 0.7055,
+      "step": 3309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1191885038038885e-05,
+      "loss": 0.5778,
+      "step": 3310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.119526627218935e-05,
+      "loss": 0.5479,
+      "step": 3311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1198647506339814e-05,
+      "loss": 0.5626,
+      "step": 3312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.120202874049028e-05,
+      "loss": 0.6116,
+      "step": 3313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1205409974640743e-05,
+      "loss": 0.6416,
+      "step": 3314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.120879120879121e-05,
+      "loss": 0.5046,
+      "step": 3315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1212172442941676e-05,
+      "loss": 0.6866,
+      "step": 3316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1215553677092139e-05,
+      "loss": 0.5585,
+      "step": 3317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1218934911242605e-05,
+      "loss": 0.5809,
+      "step": 3318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1222316145393068e-05,
+      "loss": 0.5894,
+      "step": 3319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1225697379543534e-05,
+      "loss": 0.7235,
+      "step": 3320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1229078613694e-05,
+      "loss": 0.6335,
+      "step": 3321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1232459847844463e-05,
+      "loss": 0.7742,
+      "step": 3322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.123584108199493e-05,
+      "loss": 0.594,
+      "step": 3323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1239222316145396e-05,
+      "loss": 0.5872,
+      "step": 3324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1242603550295859e-05,
+      "loss": 0.6626,
+      "step": 3325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1245984784446325e-05,
+      "loss": 0.6611,
+      "step": 3326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1249366018596788e-05,
+      "loss": 0.5832,
+      "step": 3327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1252747252747254e-05,
+      "loss": 0.583,
+      "step": 3328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.125612848689772e-05,
+      "loss": 0.6259,
+      "step": 3329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1259509721048183e-05,
+      "loss": 0.6818,
+      "step": 3330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.126289095519865e-05,
+      "loss": 0.6931,
+      "step": 3331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1266272189349112e-05,
+      "loss": 0.5331,
+      "step": 3332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1269653423499578e-05,
+      "loss": 0.5576,
+      "step": 3333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1273034657650045e-05,
+      "loss": 0.6623,
+      "step": 3334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1276415891800507e-05,
+      "loss": 0.5965,
+      "step": 3335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1279797125950974e-05,
+      "loss": 0.5216,
+      "step": 3336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1283178360101438e-05,
+      "loss": 0.5872,
+      "step": 3337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1286559594251903e-05,
+      "loss": 0.6002,
+      "step": 3338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1289940828402369e-05,
+      "loss": 0.6649,
+      "step": 3339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1293322062552832e-05,
+      "loss": 0.7317,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1296703296703298e-05,
+      "loss": 0.6261,
+      "step": 3341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1300084530853763e-05,
+      "loss": 0.618,
+      "step": 3342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1303465765004227e-05,
+      "loss": 0.7393,
+      "step": 3343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1306846999154692e-05,
+      "loss": 0.6988,
+      "step": 3344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1310228233305158e-05,
+      "loss": 0.6237,
+      "step": 3345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1313609467455623e-05,
+      "loss": 0.5549,
+      "step": 3346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1316990701606087e-05,
+      "loss": 0.6317,
+      "step": 3347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1320371935756552e-05,
+      "loss": 0.6548,
+      "step": 3348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1323753169907016e-05,
+      "loss": 0.5624,
+      "step": 3349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1327134404057482e-05,
+      "loss": 0.5885,
+      "step": 3350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1330515638207947e-05,
+      "loss": 0.6184,
+      "step": 3351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1333896872358412e-05,
+      "loss": 0.6314,
+      "step": 3352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1337278106508878e-05,
+      "loss": 0.6049,
+      "step": 3353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.134065934065934e-05,
+      "loss": 0.6523,
+      "step": 3354
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1344040574809807e-05,
+      "loss": 0.6074,
+      "step": 3355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1347421808960271e-05,
+      "loss": 0.5529,
+      "step": 3356
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1350803043110736e-05,
+      "loss": 0.5822,
+      "step": 3357
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1354184277261202e-05,
+      "loss": 0.5063,
+      "step": 3358
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1357565511411665e-05,
+      "loss": 0.5856,
+      "step": 3359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1360946745562131e-05,
+      "loss": 0.5767,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1364327979712594e-05,
+      "loss": 0.6108,
+      "step": 3361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.136770921386306e-05,
+      "loss": 0.5724,
+      "step": 3362
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1371090448013527e-05,
+      "loss": 0.6905,
+      "step": 3363
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.137447168216399e-05,
+      "loss": 0.5528,
+      "step": 3364
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1377852916314456e-05,
+      "loss": 0.6366,
+      "step": 3365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1381234150464922e-05,
+      "loss": 0.6423,
+      "step": 3366
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1384615384615385e-05,
+      "loss": 0.5617,
+      "step": 3367
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1387996618765851e-05,
+      "loss": 0.5981,
+      "step": 3368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1391377852916314e-05,
+      "loss": 0.6143,
+      "step": 3369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.139475908706678e-05,
+      "loss": 0.5887,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1398140321217246e-05,
+      "loss": 0.5575,
+      "step": 3371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.140152155536771e-05,
+      "loss": 0.6142,
+      "step": 3372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1404902789518175e-05,
+      "loss": 0.6422,
+      "step": 3373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1408284023668642e-05,
+      "loss": 0.5636,
+      "step": 3374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1411665257819105e-05,
+      "loss": 0.5111,
+      "step": 3375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.141504649196957e-05,
+      "loss": 0.5928,
+      "step": 3376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1418427726120034e-05,
+      "loss": 0.5714,
+      "step": 3377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.14218089602705e-05,
+      "loss": 0.6212,
+      "step": 3378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1425190194420966e-05,
+      "loss": 0.5365,
+      "step": 3379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.5687,
+      "step": 3380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1431952662721895e-05,
+      "loss": 0.5602,
+      "step": 3381
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.143533389687236e-05,
+      "loss": 0.5402,
+      "step": 3382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1438715131022824e-05,
+      "loss": 0.5706,
+      "step": 3383
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1442096365173289e-05,
+      "loss": 0.6271,
+      "step": 3384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1445477599323753e-05,
+      "loss": 0.5568,
+      "step": 3385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.144885883347422e-05,
+      "loss": 0.6984,
+      "step": 3386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1452240067624684e-05,
+      "loss": 0.6041,
+      "step": 3387
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1455621301775149e-05,
+      "loss": 0.6145,
+      "step": 3388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1459002535925613e-05,
+      "loss": 0.5738,
+      "step": 3389
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1462383770076078e-05,
+      "loss": 0.6204,
+      "step": 3390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1465765004226544e-05,
+      "loss": 0.7046,
+      "step": 3391
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1469146238377009e-05,
+      "loss": 0.7075,
+      "step": 3392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1472527472527473e-05,
+      "loss": 0.6699,
+      "step": 3393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1475908706677938e-05,
+      "loss": 0.6165,
+      "step": 3394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1479289940828404e-05,
+      "loss": 0.5262,
+      "step": 3395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1482671174978869e-05,
+      "loss": 0.5574,
+      "step": 3396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1486052409129333e-05,
+      "loss": 0.6186,
+      "step": 3397
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1489433643279798e-05,
+      "loss": 0.5929,
+      "step": 3398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1492814877430262e-05,
+      "loss": 0.5625,
+      "step": 3399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1496196111580728e-05,
+      "loss": 0.5632,
+      "step": 3400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1499577345731191e-05,
+      "loss": 0.5281,
+      "step": 3401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1502958579881658e-05,
+      "loss": 0.5513,
+      "step": 3402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1506339814032124e-05,
+      "loss": 0.5761,
+      "step": 3403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1509721048182587e-05,
+      "loss": 0.6008,
+      "step": 3404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1513102282333053e-05,
+      "loss": 0.5456,
+      "step": 3405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1516483516483516e-05,
+      "loss": 0.5342,
+      "step": 3406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1519864750633982e-05,
+      "loss": 0.5923,
+      "step": 3407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1523245984784448e-05,
+      "loss": 0.6608,
+      "step": 3408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1526627218934911e-05,
+      "loss": 0.5123,
+      "step": 3409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1530008453085377e-05,
+      "loss": 0.5597,
+      "step": 3410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1533389687235844e-05,
+      "loss": 0.6316,
+      "step": 3411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1536770921386306e-05,
+      "loss": 0.7462,
+      "step": 3412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1540152155536773e-05,
+      "loss": 0.6025,
+      "step": 3413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1543533389687236e-05,
+      "loss": 0.5829,
+      "step": 3414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1546914623837702e-05,
+      "loss": 0.5614,
+      "step": 3415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1550295857988168e-05,
+      "loss": 0.6075,
+      "step": 3416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1553677092138631e-05,
+      "loss": 0.6013,
+      "step": 3417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1557058326289097e-05,
+      "loss": 0.5526,
+      "step": 3418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.156043956043956e-05,
+      "loss": 0.5793,
+      "step": 3419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1563820794590026e-05,
+      "loss": 0.6035,
+      "step": 3420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1567202028740492e-05,
+      "loss": 0.553,
+      "step": 3421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1570583262890955e-05,
+      "loss": 0.5213,
+      "step": 3422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1573964497041422e-05,
+      "loss": 0.4958,
+      "step": 3423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1577345731191886e-05,
+      "loss": 0.6222,
+      "step": 3424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.158072696534235e-05,
+      "loss": 0.5079,
+      "step": 3425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1584108199492817e-05,
+      "loss": 0.5231,
+      "step": 3426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.158748943364328e-05,
+      "loss": 0.6273,
+      "step": 3427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1590870667793746e-05,
+      "loss": 0.6092,
+      "step": 3428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.159425190194421e-05,
+      "loss": 0.5528,
+      "step": 3429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1597633136094675e-05,
+      "loss": 0.5884,
+      "step": 3430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1601014370245141e-05,
+      "loss": 0.7004,
+      "step": 3431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1604395604395606e-05,
+      "loss": 0.5962,
+      "step": 3432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.160777683854607e-05,
+      "loss": 0.6665,
+      "step": 3433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1611158072696535e-05,
+      "loss": 0.5551,
+      "step": 3434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1614539306847e-05,
+      "loss": 0.6843,
+      "step": 3435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1617920540997466e-05,
+      "loss": 0.5725,
+      "step": 3436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.162130177514793e-05,
+      "loss": 0.6379,
+      "step": 3437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1624683009298395e-05,
+      "loss": 0.5834,
+      "step": 3438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.162806424344886e-05,
+      "loss": 0.5839,
+      "step": 3439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1631445477599326e-05,
+      "loss": 0.6201,
+      "step": 3440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1634826711749788e-05,
+      "loss": 0.5576,
+      "step": 3441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1638207945900255e-05,
+      "loss": 0.5396,
+      "step": 3442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.164158918005072e-05,
+      "loss": 0.6177,
+      "step": 3443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1644970414201184e-05,
+      "loss": 0.6134,
+      "step": 3444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.164835164835165e-05,
+      "loss": 0.5964,
+      "step": 3445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1651732882502113e-05,
+      "loss": 0.5806,
+      "step": 3446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1655114116652579e-05,
+      "loss": 0.6295,
+      "step": 3447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1658495350803044e-05,
+      "loss": 0.6234,
+      "step": 3448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1661876584953508e-05,
+      "loss": 0.6295,
+      "step": 3449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1665257819103974e-05,
+      "loss": 0.5152,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1668639053254437e-05,
+      "loss": 0.6052,
+      "step": 3451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1672020287404904e-05,
+      "loss": 0.6063,
+      "step": 3452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.167540152155537e-05,
+      "loss": 0.596,
+      "step": 3453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1678782755705833e-05,
+      "loss": 0.5739,
+      "step": 3454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1682163989856299e-05,
+      "loss": 0.6062,
+      "step": 3455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1685545224006762e-05,
+      "loss": 0.5428,
+      "step": 3456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1688926458157228e-05,
+      "loss": 0.6058,
+      "step": 3457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1692307692307694e-05,
+      "loss": 0.5405,
+      "step": 3458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1695688926458157e-05,
+      "loss": 0.5882,
+      "step": 3459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1699070160608623e-05,
+      "loss": 0.5606,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.170245139475909e-05,
+      "loss": 0.5303,
+      "step": 3461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1705832628909552e-05,
+      "loss": 0.549,
+      "step": 3462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1709213863060019e-05,
+      "loss": 0.6069,
+      "step": 3463
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1712595097210482e-05,
+      "loss": 0.6608,
+      "step": 3464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1715976331360948e-05,
+      "loss": 0.5635,
+      "step": 3465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1719357565511414e-05,
+      "loss": 0.6169,
+      "step": 3466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1722738799661877e-05,
+      "loss": 0.6977,
+      "step": 3467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1726120033812343e-05,
+      "loss": 0.4815,
+      "step": 3468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1729501267962808e-05,
+      "loss": 0.6076,
+      "step": 3469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1732882502113272e-05,
+      "loss": 0.6687,
+      "step": 3470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1736263736263738e-05,
+      "loss": 0.5621,
+      "step": 3471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1739644970414201e-05,
+      "loss": 0.5927,
+      "step": 3472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1743026204564668e-05,
+      "loss": 0.6667,
+      "step": 3473
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1746407438715132e-05,
+      "loss": 0.7447,
+      "step": 3474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1749788672865597e-05,
+      "loss": 0.5624,
+      "step": 3475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1753169907016063e-05,
+      "loss": 0.6515,
+      "step": 3476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1756551141166527e-05,
+      "loss": 0.578,
+      "step": 3477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1759932375316992e-05,
+      "loss": 0.6418,
+      "step": 3478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1763313609467457e-05,
+      "loss": 0.6179,
+      "step": 3479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1766694843617921e-05,
+      "loss": 0.5376,
+      "step": 3480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1770076077768386e-05,
+      "loss": 0.6233,
+      "step": 3481
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1773457311918852e-05,
+      "loss": 0.6697,
+      "step": 3482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1776838546069316e-05,
+      "loss": 0.6557,
+      "step": 3483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1780219780219781e-05,
+      "loss": 0.6068,
+      "step": 3484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1783601014370246e-05,
+      "loss": 0.6355,
+      "step": 3485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.178698224852071e-05,
+      "loss": 0.6006,
+      "step": 3486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1790363482671176e-05,
+      "loss": 0.6112,
+      "step": 3487
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1793744716821641e-05,
+      "loss": 0.6138,
+      "step": 3488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1797125950972105e-05,
+      "loss": 0.5505,
+      "step": 3489
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1800507185122572e-05,
+      "loss": 0.6092,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1803888419273035e-05,
+      "loss": 0.6722,
+      "step": 3491
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.18072696534235e-05,
+      "loss": 0.556,
+      "step": 3492
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1810650887573965e-05,
+      "loss": 0.5994,
+      "step": 3493
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.181403212172443e-05,
+      "loss": 0.5622,
+      "step": 3494
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1817413355874896e-05,
+      "loss": 0.5569,
+      "step": 3495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1820794590025359e-05,
+      "loss": 0.5879,
+      "step": 3496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1824175824175825e-05,
+      "loss": 0.574,
+      "step": 3497
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1827557058326291e-05,
+      "loss": 0.5475,
+      "step": 3498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1830938292476754e-05,
+      "loss": 0.6411,
+      "step": 3499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.183431952662722e-05,
+      "loss": 0.6479,
+      "step": 3500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1837700760777683e-05,
+      "loss": 0.5089,
+      "step": 3501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.184108199492815e-05,
+      "loss": 0.5831,
+      "step": 3502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1844463229078616e-05,
+      "loss": 0.6338,
+      "step": 3503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1847844463229079e-05,
+      "loss": 0.6795,
+      "step": 3504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1851225697379545e-05,
+      "loss": 0.5427,
+      "step": 3505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1854606931530011e-05,
+      "loss": 0.6795,
+      "step": 3506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1857988165680474e-05,
+      "loss": 0.5765,
+      "step": 3507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.186136939983094e-05,
+      "loss": 0.5627,
+      "step": 3508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1864750633981403e-05,
+      "loss": 0.6433,
+      "step": 3509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.186813186813187e-05,
+      "loss": 0.6486,
+      "step": 3510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1871513102282336e-05,
+      "loss": 0.565,
+      "step": 3511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1874894336432798e-05,
+      "loss": 0.5851,
+      "step": 3512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1878275570583265e-05,
+      "loss": 0.5779,
+      "step": 3513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1881656804733728e-05,
+      "loss": 0.6543,
+      "step": 3514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1885038038884194e-05,
+      "loss": 0.5444,
+      "step": 3515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.188841927303466e-05,
+      "loss": 0.6968,
+      "step": 3516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1891800507185123e-05,
+      "loss": 0.5706,
+      "step": 3517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.189518174133559e-05,
+      "loss": 0.6348,
+      "step": 3518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1898562975486054e-05,
+      "loss": 0.6904,
+      "step": 3519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1901944209636518e-05,
+      "loss": 0.5004,
+      "step": 3520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1905325443786983e-05,
+      "loss": 0.5994,
+      "step": 3521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1908706677937447e-05,
+      "loss": 0.5385,
+      "step": 3522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1912087912087914e-05,
+      "loss": 0.5614,
+      "step": 3523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1915469146238378e-05,
+      "loss": 0.6737,
+      "step": 3524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1918850380388843e-05,
+      "loss": 0.6155,
+      "step": 3525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1922231614539307e-05,
+      "loss": 0.5678,
+      "step": 3526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1925612848689773e-05,
+      "loss": 0.5281,
+      "step": 3527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1928994082840238e-05,
+      "loss": 0.5993,
+      "step": 3528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1932375316990703e-05,
+      "loss": 0.6043,
+      "step": 3529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1935756551141167e-05,
+      "loss": 0.6026,
+      "step": 3530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1939137785291632e-05,
+      "loss": 0.6822,
+      "step": 3531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1942519019442098e-05,
+      "loss": 0.5445,
+      "step": 3532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1945900253592562e-05,
+      "loss": 0.5411,
+      "step": 3533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1949281487743027e-05,
+      "loss": 0.5942,
+      "step": 3534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1952662721893493e-05,
+      "loss": 0.5178,
+      "step": 3535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1956043956043956e-05,
+      "loss": 0.5597,
+      "step": 3536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1959425190194422e-05,
+      "loss": 0.6226,
+      "step": 3537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1962806424344885e-05,
+      "loss": 0.5766,
+      "step": 3538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1966187658495351e-05,
+      "loss": 0.6419,
+      "step": 3539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1969568892645818e-05,
+      "loss": 0.5594,
+      "step": 3540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.197295012679628e-05,
+      "loss": 0.6866,
+      "step": 3541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1976331360946747e-05,
+      "loss": 0.5929,
+      "step": 3542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.197971259509721e-05,
+      "loss": 0.5738,
+      "step": 3543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1983093829247676e-05,
+      "loss": 0.6233,
+      "step": 3544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1986475063398142e-05,
+      "loss": 0.6003,
+      "step": 3545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1989856297548605e-05,
+      "loss": 0.5607,
+      "step": 3546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1993237531699071e-05,
+      "loss": 0.6316,
+      "step": 3547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1996618765849537e-05,
+      "loss": 0.6068,
+      "step": 3548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6016,
+      "step": 3549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2003381234150467e-05,
+      "loss": 0.6046,
+      "step": 3550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.200676246830093e-05,
+      "loss": 0.6787,
+      "step": 3551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2010143702451396e-05,
+      "loss": 0.6198,
+      "step": 3552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2013524936601862e-05,
+      "loss": 0.5831,
+      "step": 3553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2016906170752325e-05,
+      "loss": 0.6706,
+      "step": 3554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2020287404902791e-05,
+      "loss": 0.6277,
+      "step": 3555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2023668639053257e-05,
+      "loss": 0.796,
+      "step": 3556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.202704987320372e-05,
+      "loss": 0.6919,
+      "step": 3557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2030431107354186e-05,
+      "loss": 0.5643,
+      "step": 3558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.203381234150465e-05,
+      "loss": 0.6064,
+      "step": 3559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2037193575655115e-05,
+      "loss": 0.5141,
+      "step": 3560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.204057480980558e-05,
+      "loss": 0.6042,
+      "step": 3561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2043956043956045e-05,
+      "loss": 0.5843,
+      "step": 3562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.204733727810651e-05,
+      "loss": 0.553,
+      "step": 3563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2050718512256975e-05,
+      "loss": 0.6475,
+      "step": 3564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.205409974640744e-05,
+      "loss": 0.5331,
+      "step": 3565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2057480980557904e-05,
+      "loss": 0.5133,
+      "step": 3566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2060862214708369e-05,
+      "loss": 0.569,
+      "step": 3567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2064243448858835e-05,
+      "loss": 0.6499,
+      "step": 3568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.20676246830093e-05,
+      "loss": 0.5694,
+      "step": 3569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2071005917159764e-05,
+      "loss": 0.7089,
+      "step": 3570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2074387151310229e-05,
+      "loss": 0.4979,
+      "step": 3571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2077768385460693e-05,
+      "loss": 0.6228,
+      "step": 3572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.208114961961116e-05,
+      "loss": 0.6004,
+      "step": 3573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2084530853761624e-05,
+      "loss": 0.6373,
+      "step": 3574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2087912087912089e-05,
+      "loss": 0.7315,
+      "step": 3575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2091293322062553e-05,
+      "loss": 0.5805,
+      "step": 3576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.209467455621302e-05,
+      "loss": 0.6063,
+      "step": 3577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2098055790363484e-05,
+      "loss": 0.6277,
+      "step": 3578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2101437024513949e-05,
+      "loss": 0.5282,
+      "step": 3579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2104818258664413e-05,
+      "loss": 0.607,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2108199492814878e-05,
+      "loss": 0.674,
+      "step": 3581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2111580726965344e-05,
+      "loss": 0.5789,
+      "step": 3582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2114961961115807e-05,
+      "loss": 0.6164,
+      "step": 3583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2118343195266273e-05,
+      "loss": 0.6227,
+      "step": 3584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.212172442941674e-05,
+      "loss": 0.4797,
+      "step": 3585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2125105663567202e-05,
+      "loss": 0.5479,
+      "step": 3586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2128486897717668e-05,
+      "loss": 0.6662,
+      "step": 3587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2131868131868131e-05,
+      "loss": 0.5842,
+      "step": 3588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2135249366018597e-05,
+      "loss": 0.6548,
+      "step": 3589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2138630600169064e-05,
+      "loss": 0.5558,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2142011834319527e-05,
+      "loss": 0.5467,
+      "step": 3591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2145393068469993e-05,
+      "loss": 0.5658,
+      "step": 3592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2148774302620459e-05,
+      "loss": 0.6094,
+      "step": 3593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2152155536770922e-05,
+      "loss": 0.6265,
+      "step": 3594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2155536770921388e-05,
+      "loss": 0.5635,
+      "step": 3595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2158918005071851e-05,
+      "loss": 0.6225,
+      "step": 3596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2162299239222317e-05,
+      "loss": 0.5123,
+      "step": 3597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2165680473372783e-05,
+      "loss": 0.5466,
+      "step": 3598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2169061707523246e-05,
+      "loss": 0.6069,
+      "step": 3599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2172442941673713e-05,
+      "loss": 0.6363,
+      "step": 3600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2175824175824175e-05,
+      "loss": 0.5416,
+      "step": 3601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2179205409974642e-05,
+      "loss": 0.5631,
+      "step": 3602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2182586644125108e-05,
+      "loss": 0.6964,
+      "step": 3603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.218596787827557e-05,
+      "loss": 0.6314,
+      "step": 3604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2189349112426037e-05,
+      "loss": 0.5543,
+      "step": 3605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2192730346576502e-05,
+      "loss": 0.5283,
+      "step": 3606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2196111580726966e-05,
+      "loss": 0.5775,
+      "step": 3607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2199492814877432e-05,
+      "loss": 0.5899,
+      "step": 3608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2202874049027895e-05,
+      "loss": 0.6246,
+      "step": 3609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2206255283178361e-05,
+      "loss": 0.6911,
+      "step": 3610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2209636517328826e-05,
+      "loss": 0.6269,
+      "step": 3611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.221301775147929e-05,
+      "loss": 0.5154,
+      "step": 3612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2216398985629757e-05,
+      "loss": 0.6029,
+      "step": 3613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2219780219780221e-05,
+      "loss": 0.67,
+      "step": 3614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2223161453930686e-05,
+      "loss": 0.5971,
+      "step": 3615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.222654268808115e-05,
+      "loss": 0.6325,
+      "step": 3616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2229923922231615e-05,
+      "loss": 0.6239,
+      "step": 3617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2233305156382081e-05,
+      "loss": 0.6254,
+      "step": 3618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2236686390532546e-05,
+      "loss": 0.5778,
+      "step": 3619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.224006762468301e-05,
+      "loss": 0.6494,
+      "step": 3620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2243448858833475e-05,
+      "loss": 0.5513,
+      "step": 3621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2246830092983941e-05,
+      "loss": 0.6254,
+      "step": 3622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2250211327134404e-05,
+      "loss": 0.5912,
+      "step": 3623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.225359256128487e-05,
+      "loss": 0.6769,
+      "step": 3624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2256973795435335e-05,
+      "loss": 0.556,
+      "step": 3625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.22603550295858e-05,
+      "loss": 0.6454,
+      "step": 3626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2263736263736266e-05,
+      "loss": 0.6679,
+      "step": 3627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2267117497886728e-05,
+      "loss": 0.5346,
+      "step": 3628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2270498732037195e-05,
+      "loss": 0.6012,
+      "step": 3629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.227387996618766e-05,
+      "loss": 0.5702,
+      "step": 3630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2277261200338124e-05,
+      "loss": 0.5263,
+      "step": 3631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.228064243448859e-05,
+      "loss": 0.6546,
+      "step": 3632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2284023668639053e-05,
+      "loss": 0.5586,
+      "step": 3633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2287404902789519e-05,
+      "loss": 0.6023,
+      "step": 3634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2290786136939985e-05,
+      "loss": 0.597,
+      "step": 3635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2294167371090448e-05,
+      "loss": 0.8013,
+      "step": 3636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2297548605240914e-05,
+      "loss": 0.5518,
+      "step": 3637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2300929839391377e-05,
+      "loss": 0.6105,
+      "step": 3638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2304311073541843e-05,
+      "loss": 0.6059,
+      "step": 3639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.6194,
+      "step": 3640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2311073541842773e-05,
+      "loss": 0.5936,
+      "step": 3641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2314454775993239e-05,
+      "loss": 0.4962,
+      "step": 3642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2317836010143705e-05,
+      "loss": 0.6858,
+      "step": 3643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2321217244294168e-05,
+      "loss": 0.7037,
+      "step": 3644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2324598478444634e-05,
+      "loss": 0.5773,
+      "step": 3645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2327979712595097e-05,
+      "loss": 0.6542,
+      "step": 3646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2331360946745563e-05,
+      "loss": 0.5877,
+      "step": 3647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.233474218089603e-05,
+      "loss": 0.6083,
+      "step": 3648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2338123415046492e-05,
+      "loss": 0.6789,
+      "step": 3649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2341504649196959e-05,
+      "loss": 0.6535,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2344885883347423e-05,
+      "loss": 0.6127,
+      "step": 3651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2348267117497888e-05,
+      "loss": 0.5441,
+      "step": 3652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2351648351648354e-05,
+      "loss": 0.6049,
+      "step": 3653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2355029585798817e-05,
+      "loss": 0.5205,
+      "step": 3654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2358410819949283e-05,
+      "loss": 0.6097,
+      "step": 3655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2361792054099748e-05,
+      "loss": 0.5376,
+      "step": 3656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2365173288250212e-05,
+      "loss": 0.643,
+      "step": 3657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2368554522400678e-05,
+      "loss": 0.5846,
+      "step": 3658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2371935756551141e-05,
+      "loss": 0.6255,
+      "step": 3659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2375316990701607e-05,
+      "loss": 0.6408,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2378698224852072e-05,
+      "loss": 0.6066,
+      "step": 3661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2382079459002537e-05,
+      "loss": 0.6233,
+      "step": 3662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2385460693153001e-05,
+      "loss": 0.5654,
+      "step": 3663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2388841927303467e-05,
+      "loss": 0.5136,
+      "step": 3664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2392223161453932e-05,
+      "loss": 0.5039,
+      "step": 3665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2395604395604396e-05,
+      "loss": 0.6017,
+      "step": 3666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2398985629754861e-05,
+      "loss": 0.6265,
+      "step": 3667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2402366863905326e-05,
+      "loss": 0.5555,
+      "step": 3668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2405748098055792e-05,
+      "loss": 0.6392,
+      "step": 3669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2409129332206256e-05,
+      "loss": 0.6411,
+      "step": 3670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2412510566356721e-05,
+      "loss": 0.5861,
+      "step": 3671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2415891800507187e-05,
+      "loss": 0.6731,
+      "step": 3672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.241927303465765e-05,
+      "loss": 0.6445,
+      "step": 3673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2422654268808116e-05,
+      "loss": 0.5347,
+      "step": 3674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.242603550295858e-05,
+      "loss": 0.5363,
+      "step": 3675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2429416737109045e-05,
+      "loss": 0.6099,
+      "step": 3676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2432797971259512e-05,
+      "loss": 0.6227,
+      "step": 3677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2436179205409974e-05,
+      "loss": 0.6404,
+      "step": 3678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.243956043956044e-05,
+      "loss": 0.5401,
+      "step": 3679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2442941673710907e-05,
+      "loss": 0.6346,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.244632290786137e-05,
+      "loss": 0.5467,
+      "step": 3681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2449704142011836e-05,
+      "loss": 0.6453,
+      "step": 3682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2453085376162299e-05,
+      "loss": 0.6493,
+      "step": 3683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2456466610312765e-05,
+      "loss": 0.5462,
+      "step": 3684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2459847844463231e-05,
+      "loss": 0.6417,
+      "step": 3685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2463229078613694e-05,
+      "loss": 0.757,
+      "step": 3686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.246661031276416e-05,
+      "loss": 0.5991,
+      "step": 3687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2469991546914623e-05,
+      "loss": 0.6308,
+      "step": 3688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.247337278106509e-05,
+      "loss": 0.6397,
+      "step": 3689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2476754015215556e-05,
+      "loss": 0.6976,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2480135249366019e-05,
+      "loss": 0.6249,
+      "step": 3691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2483516483516485e-05,
+      "loss": 0.5546,
+      "step": 3692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2486897717666951e-05,
+      "loss": 0.5549,
+      "step": 3693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2490278951817414e-05,
+      "loss": 0.6369,
+      "step": 3694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.249366018596788e-05,
+      "loss": 0.6717,
+      "step": 3695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2497041420118343e-05,
+      "loss": 0.6091,
+      "step": 3696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.250042265426881e-05,
+      "loss": 0.5295,
+      "step": 3697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2503803888419276e-05,
+      "loss": 0.6218,
+      "step": 3698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2507185122569738e-05,
+      "loss": 0.6184,
+      "step": 3699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2510566356720205e-05,
+      "loss": 0.6113,
+      "step": 3700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.251394759087067e-05,
+      "loss": 0.5347,
+      "step": 3701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2517328825021134e-05,
+      "loss": 0.6262,
+      "step": 3702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2520710059171598e-05,
+      "loss": 0.6156,
+      "step": 3703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2524091293322063e-05,
+      "loss": 0.6926,
+      "step": 3704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2527472527472529e-05,
+      "loss": 0.5895,
+      "step": 3705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2530853761622994e-05,
+      "loss": 0.6231,
+      "step": 3706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2534234995773458e-05,
+      "loss": 0.6717,
+      "step": 3707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2537616229923923e-05,
+      "loss": 0.5465,
+      "step": 3708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2540997464074389e-05,
+      "loss": 0.6862,
+      "step": 3709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2544378698224854e-05,
+      "loss": 0.6884,
+      "step": 3710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2547759932375318e-05,
+      "loss": 0.6148,
+      "step": 3711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2551141166525783e-05,
+      "loss": 0.5949,
+      "step": 3712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2554522400676247e-05,
+      "loss": 0.6583,
+      "step": 3713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2557903634826713e-05,
+      "loss": 0.5545,
+      "step": 3714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2561284868977178e-05,
+      "loss": 0.6168,
+      "step": 3715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2564666103127642e-05,
+      "loss": 0.639,
+      "step": 3716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2568047337278107e-05,
+      "loss": 0.5373,
+      "step": 3717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2571428571428572e-05,
+      "loss": 0.7067,
+      "step": 3718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2574809805579038e-05,
+      "loss": 0.5665,
+      "step": 3719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.25781910397295e-05,
+      "loss": 0.6257,
+      "step": 3720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2581572273879967e-05,
+      "loss": 0.5515,
+      "step": 3721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2584953508030433e-05,
+      "loss": 0.6638,
+      "step": 3722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2588334742180896e-05,
+      "loss": 0.5227,
+      "step": 3723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2591715976331362e-05,
+      "loss": 0.5455,
+      "step": 3724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2595097210481825e-05,
+      "loss": 0.6271,
+      "step": 3725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2598478444632291e-05,
+      "loss": 0.5432,
+      "step": 3726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2601859678782758e-05,
+      "loss": 0.5865,
+      "step": 3727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.260524091293322e-05,
+      "loss": 0.5857,
+      "step": 3728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2608622147083687e-05,
+      "loss": 0.593,
+      "step": 3729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2612003381234153e-05,
+      "loss": 0.6292,
+      "step": 3730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2615384615384616e-05,
+      "loss": 0.5056,
+      "step": 3731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2618765849535082e-05,
+      "loss": 0.5606,
+      "step": 3732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2622147083685545e-05,
+      "loss": 0.6037,
+      "step": 3733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2625528317836011e-05,
+      "loss": 0.5388,
+      "step": 3734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2628909551986477e-05,
+      "loss": 0.6728,
+      "step": 3735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.263229078613694e-05,
+      "loss": 0.5283,
+      "step": 3736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2635672020287406e-05,
+      "loss": 0.6569,
+      "step": 3737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2639053254437873e-05,
+      "loss": 0.605,
+      "step": 3738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2642434488588336e-05,
+      "loss": 0.5962,
+      "step": 3739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2645815722738802e-05,
+      "loss": 0.5846,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2649196956889265e-05,
+      "loss": 0.6172,
+      "step": 3741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2652578191039731e-05,
+      "loss": 0.6478,
+      "step": 3742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2655959425190195e-05,
+      "loss": 0.6724,
+      "step": 3743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.265934065934066e-05,
+      "loss": 0.5548,
+      "step": 3744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2662721893491126e-05,
+      "loss": 0.5999,
+      "step": 3745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2666103127641589e-05,
+      "loss": 0.5647,
+      "step": 3746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2669484361792055e-05,
+      "loss": 0.5955,
+      "step": 3747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.267286559594252e-05,
+      "loss": 0.5435,
+      "step": 3748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2676246830092984e-05,
+      "loss": 0.6465,
+      "step": 3749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.267962806424345e-05,
+      "loss": 0.6301,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2683009298393915e-05,
+      "loss": 0.686,
+      "step": 3751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.268639053254438e-05,
+      "loss": 0.6566,
+      "step": 3752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2689771766694844e-05,
+      "loss": 0.5799,
+      "step": 3753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2693153000845309e-05,
+      "loss": 0.6626,
+      "step": 3754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2696534234995775e-05,
+      "loss": 0.6511,
+      "step": 3755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.269991546914624e-05,
+      "loss": 0.6246,
+      "step": 3756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2703296703296704e-05,
+      "loss": 0.5272,
+      "step": 3757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2706677937447169e-05,
+      "loss": 0.5522,
+      "step": 3758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2710059171597635e-05,
+      "loss": 0.5748,
+      "step": 3759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2713440405748098e-05,
+      "loss": 0.5474,
+      "step": 3760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2716821639898564e-05,
+      "loss": 0.6129,
+      "step": 3761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2720202874049029e-05,
+      "loss": 0.6126,
+      "step": 3762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2723584108199493e-05,
+      "loss": 0.5494,
+      "step": 3763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.272696534234996e-05,
+      "loss": 0.4948,
+      "step": 3764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2730346576500422e-05,
+      "loss": 0.611,
+      "step": 3765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2733727810650889e-05,
+      "loss": 0.5316,
+      "step": 3766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2737109044801355e-05,
+      "loss": 0.6124,
+      "step": 3767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2740490278951818e-05,
+      "loss": 0.579,
+      "step": 3768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2743871513102284e-05,
+      "loss": 0.5927,
+      "step": 3769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2747252747252747e-05,
+      "loss": 0.6414,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2750633981403213e-05,
+      "loss": 0.5608,
+      "step": 3771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.275401521555368e-05,
+      "loss": 0.5936,
+      "step": 3772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2757396449704142e-05,
+      "loss": 0.6012,
+      "step": 3773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2760777683854608e-05,
+      "loss": 0.6061,
+      "step": 3774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2764158918005071e-05,
+      "loss": 0.6412,
+      "step": 3775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2767540152155537e-05,
+      "loss": 0.5486,
+      "step": 3776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2770921386306004e-05,
+      "loss": 0.6329,
+      "step": 3777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2774302620456466e-05,
+      "loss": 0.5852,
+      "step": 3778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2777683854606933e-05,
+      "loss": 0.6068,
+      "step": 3779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2781065088757399e-05,
+      "loss": 0.6057,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2784446322907862e-05,
+      "loss": 0.545,
+      "step": 3781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2787827557058328e-05,
+      "loss": 0.5627,
+      "step": 3782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2791208791208791e-05,
+      "loss": 0.7261,
+      "step": 3783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2794590025359257e-05,
+      "loss": 0.574,
+      "step": 3784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2797971259509723e-05,
+      "loss": 0.5622,
+      "step": 3785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2801352493660186e-05,
+      "loss": 0.7518,
+      "step": 3786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2804733727810652e-05,
+      "loss": 0.5844,
+      "step": 3787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2808114961961117e-05,
+      "loss": 0.6069,
+      "step": 3788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2811496196111582e-05,
+      "loss": 0.6535,
+      "step": 3789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2814877430262048e-05,
+      "loss": 0.5742,
+      "step": 3790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.281825866441251e-05,
+      "loss": 0.6476,
+      "step": 3791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2821639898562977e-05,
+      "loss": 0.5688,
+      "step": 3792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2825021132713441e-05,
+      "loss": 0.6448,
+      "step": 3793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2828402366863906e-05,
+      "loss": 0.5169,
+      "step": 3794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2831783601014372e-05,
+      "loss": 0.6258,
+      "step": 3795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2835164835164837e-05,
+      "loss": 0.6343,
+      "step": 3796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2838546069315301e-05,
+      "loss": 0.6582,
+      "step": 3797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2841927303465766e-05,
+      "loss": 0.6326,
+      "step": 3798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.284530853761623e-05,
+      "loss": 0.716,
+      "step": 3799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2848689771766695e-05,
+      "loss": 0.6413,
+      "step": 3800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2852071005917161e-05,
+      "loss": 0.6175,
+      "step": 3801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2855452240067626e-05,
+      "loss": 0.6098,
+      "step": 3802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.285883347421809e-05,
+      "loss": 0.6876,
+      "step": 3803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2862214708368557e-05,
+      "loss": 0.5466,
+      "step": 3804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.286559594251902e-05,
+      "loss": 0.6326,
+      "step": 3805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2868977176669486e-05,
+      "loss": 0.6653,
+      "step": 3806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.287235841081995e-05,
+      "loss": 0.6099,
+      "step": 3807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2875739644970415e-05,
+      "loss": 0.576,
+      "step": 3808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2879120879120881e-05,
+      "loss": 0.6157,
+      "step": 3809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2882502113271344e-05,
+      "loss": 0.7046,
+      "step": 3810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.288588334742181e-05,
+      "loss": 0.5663,
+      "step": 3811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2889264581572275e-05,
+      "loss": 0.7535,
+      "step": 3812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.289264581572274e-05,
+      "loss": 0.5604,
+      "step": 3813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2896027049873205e-05,
+      "loss": 0.6075,
+      "step": 3814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2899408284023668e-05,
+      "loss": 0.5854,
+      "step": 3815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2902789518174135e-05,
+      "loss": 0.5559,
+      "step": 3816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.29061707523246e-05,
+      "loss": 0.6264,
+      "step": 3817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2909551986475064e-05,
+      "loss": 0.6519,
+      "step": 3818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.291293322062553e-05,
+      "loss": 0.6397,
+      "step": 3819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2916314454775993e-05,
+      "loss": 0.5906,
+      "step": 3820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2919695688926459e-05,
+      "loss": 0.6364,
+      "step": 3821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2923076923076925e-05,
+      "loss": 0.5179,
+      "step": 3822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2926458157227388e-05,
+      "loss": 0.6645,
+      "step": 3823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2929839391377854e-05,
+      "loss": 0.536,
+      "step": 3824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.293322062552832e-05,
+      "loss": 0.659,
+      "step": 3825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2936601859678783e-05,
+      "loss": 0.5549,
+      "step": 3826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.293998309382925e-05,
+      "loss": 0.5854,
+      "step": 3827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2943364327979713e-05,
+      "loss": 0.509,
+      "step": 3828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2946745562130179e-05,
+      "loss": 0.5308,
+      "step": 3829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2950126796280645e-05,
+      "loss": 0.7247,
+      "step": 3830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2953508030431108e-05,
+      "loss": 0.6356,
+      "step": 3831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2956889264581574e-05,
+      "loss": 0.6768,
+      "step": 3832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2960270498732039e-05,
+      "loss": 0.7349,
+      "step": 3833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2963651732882503e-05,
+      "loss": 0.5929,
+      "step": 3834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.296703296703297e-05,
+      "loss": 0.5057,
+      "step": 3835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2970414201183432e-05,
+      "loss": 0.5895,
+      "step": 3836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2973795435333899e-05,
+      "loss": 0.7234,
+      "step": 3837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2977176669484363e-05,
+      "loss": 0.6493,
+      "step": 3838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2980557903634828e-05,
+      "loss": 0.6233,
+      "step": 3839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2983939137785294e-05,
+      "loss": 0.5491,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2987320371935757e-05,
+      "loss": 0.615,
+      "step": 3841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2990701606086223e-05,
+      "loss": 0.6062,
+      "step": 3842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2994082840236688e-05,
+      "loss": 0.6155,
+      "step": 3843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2997464074387152e-05,
+      "loss": 0.5379,
+      "step": 3844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3000845308537617e-05,
+      "loss": 0.5114,
+      "step": 3845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3004226542688083e-05,
+      "loss": 0.6343,
+      "step": 3846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3007607776838547e-05,
+      "loss": 0.5954,
+      "step": 3847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3010989010989012e-05,
+      "loss": 0.6739,
+      "step": 3848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3014370245139476e-05,
+      "loss": 0.5149,
+      "step": 3849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3017751479289941e-05,
+      "loss": 0.5316,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3021132713440407e-05,
+      "loss": 0.5572,
+      "step": 3851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3024513947590872e-05,
+      "loss": 0.5991,
+      "step": 3852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3027895181741336e-05,
+      "loss": 0.629,
+      "step": 3853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3031276415891803e-05,
+      "loss": 0.62,
+      "step": 3854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3034657650042265e-05,
+      "loss": 0.6344,
+      "step": 3855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3038038884192732e-05,
+      "loss": 0.576,
+      "step": 3856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3041420118343196e-05,
+      "loss": 0.5377,
+      "step": 3857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.304480135249366e-05,
+      "loss": 0.5724,
+      "step": 3858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3048182586644127e-05,
+      "loss": 0.5977,
+      "step": 3859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.305156382079459e-05,
+      "loss": 0.7124,
+      "step": 3860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3054945054945056e-05,
+      "loss": 0.6662,
+      "step": 3861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3058326289095522e-05,
+      "loss": 0.5817,
+      "step": 3862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3061707523245985e-05,
+      "loss": 0.5781,
+      "step": 3863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3065088757396451e-05,
+      "loss": 0.5974,
+      "step": 3864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3068469991546914e-05,
+      "loss": 0.533,
+      "step": 3865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.307185122569738e-05,
+      "loss": 0.6549,
+      "step": 3866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3075232459847847e-05,
+      "loss": 0.5926,
+      "step": 3867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.307861369399831e-05,
+      "loss": 0.5883,
+      "step": 3868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3081994928148776e-05,
+      "loss": 0.5399,
+      "step": 3869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3085376162299239e-05,
+      "loss": 0.6061,
+      "step": 3870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3088757396449705e-05,
+      "loss": 0.6225,
+      "step": 3871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3092138630600171e-05,
+      "loss": 0.6013,
+      "step": 3872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3095519864750634e-05,
+      "loss": 0.5887,
+      "step": 3873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.30989010989011e-05,
+      "loss": 0.511,
+      "step": 3874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3102282333051567e-05,
+      "loss": 0.7085,
+      "step": 3875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.310566356720203e-05,
+      "loss": 0.5975,
+      "step": 3876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3109044801352496e-05,
+      "loss": 0.5234,
+      "step": 3877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3112426035502959e-05,
+      "loss": 0.6196,
+      "step": 3878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3115807269653425e-05,
+      "loss": 0.6,
+      "step": 3879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3119188503803891e-05,
+      "loss": 0.6555,
+      "step": 3880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3122569737954354e-05,
+      "loss": 0.5451,
+      "step": 3881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.312595097210482e-05,
+      "loss": 0.694,
+      "step": 3882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3129332206255285e-05,
+      "loss": 0.6038,
+      "step": 3883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.313271344040575e-05,
+      "loss": 0.6784,
+      "step": 3884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3136094674556214e-05,
+      "loss": 0.5158,
+      "step": 3885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3139475908706678e-05,
+      "loss": 0.5106,
+      "step": 3886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3142857142857145e-05,
+      "loss": 0.5312,
+      "step": 3887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3146238377007609e-05,
+      "loss": 0.6794,
+      "step": 3888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3149619611158074e-05,
+      "loss": 0.5765,
+      "step": 3889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3153000845308538e-05,
+      "loss": 0.5475,
+      "step": 3890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3156382079459004e-05,
+      "loss": 0.6597,
+      "step": 3891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3159763313609469e-05,
+      "loss": 0.5146,
+      "step": 3892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3163144547759934e-05,
+      "loss": 0.611,
+      "step": 3893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3166525781910398e-05,
+      "loss": 0.6232,
+      "step": 3894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3169907016060863e-05,
+      "loss": 0.674,
+      "step": 3895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3173288250211329e-05,
+      "loss": 0.544,
+      "step": 3896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3176669484361793e-05,
+      "loss": 0.6922,
+      "step": 3897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3180050718512258e-05,
+      "loss": 0.617,
+      "step": 3898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3183431952662723e-05,
+      "loss": 0.6046,
+      "step": 3899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3186813186813187e-05,
+      "loss": 0.6061,
+      "step": 3900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3190194420963653e-05,
+      "loss": 0.6912,
+      "step": 3901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3193575655114116e-05,
+      "loss": 0.6502,
+      "step": 3902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3196956889264582e-05,
+      "loss": 0.6496,
+      "step": 3903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3200338123415049e-05,
+      "loss": 0.5859,
+      "step": 3904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3203719357565512e-05,
+      "loss": 0.6829,
+      "step": 3905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3207100591715978e-05,
+      "loss": 0.6114,
+      "step": 3906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.321048182586644e-05,
+      "loss": 0.5182,
+      "step": 3907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3213863060016907e-05,
+      "loss": 0.6802,
+      "step": 3908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3217244294167373e-05,
+      "loss": 0.5233,
+      "step": 3909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3220625528317836e-05,
+      "loss": 0.4745,
+      "step": 3910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3224006762468302e-05,
+      "loss": 0.6475,
+      "step": 3911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3227387996618768e-05,
+      "loss": 0.5973,
+      "step": 3912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3230769230769231e-05,
+      "loss": 0.5683,
+      "step": 3913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3234150464919698e-05,
+      "loss": 0.6826,
+      "step": 3914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.323753169907016e-05,
+      "loss": 0.592,
+      "step": 3915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3240912933220627e-05,
+      "loss": 0.6769,
+      "step": 3916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3244294167371093e-05,
+      "loss": 0.6094,
+      "step": 3917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3247675401521556e-05,
+      "loss": 0.5913,
+      "step": 3918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3251056635672022e-05,
+      "loss": 0.5186,
+      "step": 3919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3254437869822488e-05,
+      "loss": 0.6049,
+      "step": 3920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3257819103972951e-05,
+      "loss": 0.5738,
+      "step": 3921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3261200338123417e-05,
+      "loss": 0.5695,
+      "step": 3922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.326458157227388e-05,
+      "loss": 0.6259,
+      "step": 3923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3267962806424346e-05,
+      "loss": 0.598,
+      "step": 3924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3271344040574811e-05,
+      "loss": 0.5058,
+      "step": 3925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3274725274725275e-05,
+      "loss": 0.6328,
+      "step": 3926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3278106508875742e-05,
+      "loss": 0.5265,
+      "step": 3927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3281487743026205e-05,
+      "loss": 0.5741,
+      "step": 3928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.328486897717667e-05,
+      "loss": 0.7643,
+      "step": 3929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3288250211327135e-05,
+      "loss": 0.5118,
+      "step": 3930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.32916314454776e-05,
+      "loss": 0.55,
+      "step": 3931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3295012679628066e-05,
+      "loss": 0.649,
+      "step": 3932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.329839391377853e-05,
+      "loss": 0.6747,
+      "step": 3933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3301775147928995e-05,
+      "loss": 0.6507,
+      "step": 3934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.330515638207946e-05,
+      "loss": 0.6131,
+      "step": 3935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3308537616229924e-05,
+      "loss": 0.5766,
+      "step": 3936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.331191885038039e-05,
+      "loss": 0.6669,
+      "step": 3937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3315300084530855e-05,
+      "loss": 0.6228,
+      "step": 3938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.331868131868132e-05,
+      "loss": 0.5553,
+      "step": 3939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3322062552831784e-05,
+      "loss": 0.6509,
+      "step": 3940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.332544378698225e-05,
+      "loss": 0.5648,
+      "step": 3941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3328825021132713e-05,
+      "loss": 0.6687,
+      "step": 3942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.333220625528318e-05,
+      "loss": 0.5852,
+      "step": 3943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3335587489433644e-05,
+      "loss": 0.662,
+      "step": 3944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3338968723584109e-05,
+      "loss": 0.5859,
+      "step": 3945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3342349957734575e-05,
+      "loss": 0.6333,
+      "step": 3946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3345731191885038e-05,
+      "loss": 0.6876,
+      "step": 3947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3349112426035504e-05,
+      "loss": 0.5805,
+      "step": 3948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.335249366018597e-05,
+      "loss": 0.6346,
+      "step": 3949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3355874894336433e-05,
+      "loss": 0.5706,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.33592561284869e-05,
+      "loss": 0.5467,
+      "step": 3951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3362637362637362e-05,
+      "loss": 0.6248,
+      "step": 3952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3366018596787828e-05,
+      "loss": 0.5525,
+      "step": 3953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3369399830938295e-05,
+      "loss": 0.6482,
+      "step": 3954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3372781065088758e-05,
+      "loss": 0.5868,
+      "step": 3955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3376162299239224e-05,
+      "loss": 0.5968,
+      "step": 3956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3379543533389687e-05,
+      "loss": 0.6885,
+      "step": 3957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3382924767540153e-05,
+      "loss": 0.6037,
+      "step": 3958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3386306001690619e-05,
+      "loss": 0.556,
+      "step": 3959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3389687235841082e-05,
+      "loss": 0.5413,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3393068469991548e-05,
+      "loss": 0.555,
+      "step": 3961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3396449704142014e-05,
+      "loss": 0.6358,
+      "step": 3962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3399830938292477e-05,
+      "loss": 0.5824,
+      "step": 3963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3403212172442944e-05,
+      "loss": 0.6119,
+      "step": 3964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3406593406593406e-05,
+      "loss": 0.5994,
+      "step": 3965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3409974640743873e-05,
+      "loss": 0.7143,
+      "step": 3966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3413355874894339e-05,
+      "loss": 0.6049,
+      "step": 3967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3416737109044802e-05,
+      "loss": 0.567,
+      "step": 3968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3420118343195268e-05,
+      "loss": 0.6275,
+      "step": 3969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3423499577345733e-05,
+      "loss": 0.7165,
+      "step": 3970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3426880811496197e-05,
+      "loss": 0.6548,
+      "step": 3971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3430262045646663e-05,
+      "loss": 0.5853,
+      "step": 3972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3433643279797126e-05,
+      "loss": 0.6069,
+      "step": 3973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3437024513947592e-05,
+      "loss": 0.5639,
+      "step": 3974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3440405748098057e-05,
+      "loss": 0.6431,
+      "step": 3975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3443786982248522e-05,
+      "loss": 0.6023,
+      "step": 3976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3447168216398988e-05,
+      "loss": 0.6046,
+      "step": 3977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3450549450549452e-05,
+      "loss": 0.5434,
+      "step": 3978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3453930684699917e-05,
+      "loss": 0.6246,
+      "step": 3979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3457311918850381e-05,
+      "loss": 0.6484,
+      "step": 3980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3460693153000846e-05,
+      "loss": 0.6807,
+      "step": 3981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.346407438715131e-05,
+      "loss": 0.5751,
+      "step": 3982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3467455621301777e-05,
+      "loss": 0.6802,
+      "step": 3983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3470836855452241e-05,
+      "loss": 0.6277,
+      "step": 3984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3474218089602706e-05,
+      "loss": 0.617,
+      "step": 3985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.347759932375317e-05,
+      "loss": 0.6468,
+      "step": 3986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3480980557903635e-05,
+      "loss": 0.4991,
+      "step": 3987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3484361792054101e-05,
+      "loss": 0.6408,
+      "step": 3988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3487743026204566e-05,
+      "loss": 0.5588,
+      "step": 3989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349112426035503e-05,
+      "loss": 0.5805,
+      "step": 3990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3494505494505497e-05,
+      "loss": 0.5906,
+      "step": 3991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349788672865596e-05,
+      "loss": 0.6634,
+      "step": 3992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3501267962806426e-05,
+      "loss": 0.6441,
+      "step": 3993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.350464919695689e-05,
+      "loss": 0.6759,
+      "step": 3994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3508030431107355e-05,
+      "loss": 0.6423,
+      "step": 3995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3511411665257821e-05,
+      "loss": 0.5505,
+      "step": 3996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3514792899408284e-05,
+      "loss": 0.5482,
+      "step": 3997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.351817413355875e-05,
+      "loss": 0.5302,
+      "step": 3998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3521555367709216e-05,
+      "loss": 0.6468,
+      "step": 3999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3524936601859679e-05,
+      "loss": 0.6583,
+      "step": 4000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3528317836010145e-05,
+      "loss": 0.5866,
+      "step": 4001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3531699070160608e-05,
+      "loss": 0.6435,
+      "step": 4002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3535080304311074e-05,
+      "loss": 0.5961,
+      "step": 4003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.353846153846154e-05,
+      "loss": 0.5019,
+      "step": 4004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3541842772612004e-05,
+      "loss": 0.5467,
+      "step": 4005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.354522400676247e-05,
+      "loss": 0.5756,
+      "step": 4006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3548605240912936e-05,
+      "loss": 0.5477,
+      "step": 4007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3551986475063399e-05,
+      "loss": 0.5677,
+      "step": 4008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3555367709213865e-05,
+      "loss": 0.6856,
+      "step": 4009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3558748943364328e-05,
+      "loss": 0.6002,
+      "step": 4010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3562130177514794e-05,
+      "loss": 0.5621,
+      "step": 4011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.356551141166526e-05,
+      "loss": 0.6765,
+      "step": 4012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3568892645815723e-05,
+      "loss": 0.6084,
+      "step": 4013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.357227387996619e-05,
+      "loss": 0.6104,
+      "step": 4014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3575655114116652e-05,
+      "loss": 0.5319,
+      "step": 4015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3579036348267119e-05,
+      "loss": 0.6034,
+      "step": 4016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3582417582417585e-05,
+      "loss": 0.6444,
+      "step": 4017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3585798816568048e-05,
+      "loss": 0.5831,
+      "step": 4018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3589180050718514e-05,
+      "loss": 0.56,
+      "step": 4019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3592561284868979e-05,
+      "loss": 0.5853,
+      "step": 4020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3595942519019443e-05,
+      "loss": 0.5228,
+      "step": 4021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3599323753169908e-05,
+      "loss": 0.6493,
+      "step": 4022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3602704987320372e-05,
+      "loss": 0.5813,
+      "step": 4023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3606086221470838e-05,
+      "loss": 0.5941,
+      "step": 4024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3609467455621303e-05,
+      "loss": 0.6295,
+      "step": 4025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3612848689771768e-05,
+      "loss": 0.5191,
+      "step": 4026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3616229923922232e-05,
+      "loss": 0.6647,
+      "step": 4027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3619611158072698e-05,
+      "loss": 0.6167,
+      "step": 4028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3622992392223163e-05,
+      "loss": 0.6814,
+      "step": 4029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3626373626373627e-05,
+      "loss": 0.6127,
+      "step": 4030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3629754860524092e-05,
+      "loss": 0.5617,
+      "step": 4031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3633136094674557e-05,
+      "loss": 0.5549,
+      "step": 4032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3636517328825023e-05,
+      "loss": 0.513,
+      "step": 4033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3639898562975487e-05,
+      "loss": 0.537,
+      "step": 4034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3643279797125952e-05,
+      "loss": 0.6189,
+      "step": 4035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3646661031276418e-05,
+      "loss": 0.5907,
+      "step": 4036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3650042265426881e-05,
+      "loss": 0.5785,
+      "step": 4037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3653423499577347e-05,
+      "loss": 0.6733,
+      "step": 4038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.365680473372781e-05,
+      "loss": 0.6315,
+      "step": 4039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3660185967878276e-05,
+      "loss": 0.5446,
+      "step": 4040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3663567202028743e-05,
+      "loss": 0.6539,
+      "step": 4041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3666948436179205e-05,
+      "loss": 0.5506,
+      "step": 4042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3670329670329672e-05,
+      "loss": 0.5686,
+      "step": 4043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3673710904480134e-05,
+      "loss": 0.534,
+      "step": 4044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.36770921386306e-05,
+      "loss": 0.6499,
+      "step": 4045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3680473372781067e-05,
+      "loss": 0.5836,
+      "step": 4046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.368385460693153e-05,
+      "loss": 0.6474,
+      "step": 4047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3687235841081996e-05,
+      "loss": 0.5167,
+      "step": 4048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3690617075232462e-05,
+      "loss": 0.7022,
+      "step": 4049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3693998309382925e-05,
+      "loss": 0.5029,
+      "step": 4050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3697379543533391e-05,
+      "loss": 0.6782,
+      "step": 4051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3700760777683854e-05,
+      "loss": 0.5737,
+      "step": 4052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.370414201183432e-05,
+      "loss": 0.5561,
+      "step": 4053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3707523245984787e-05,
+      "loss": 0.7174,
+      "step": 4054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.371090448013525e-05,
+      "loss": 0.6243,
+      "step": 4055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 0.5831,
+      "step": 4056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3717666948436182e-05,
+      "loss": 0.6173,
+      "step": 4057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3721048182586645e-05,
+      "loss": 0.6337,
+      "step": 4058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3724429416737111e-05,
+      "loss": 0.6121,
+      "step": 4059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3727810650887574e-05,
+      "loss": 0.5338,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.373119188503804e-05,
+      "loss": 0.5799,
+      "step": 4061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3734573119188505e-05,
+      "loss": 0.5892,
+      "step": 4062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.373795435333897e-05,
+      "loss": 0.6543,
+      "step": 4063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3741335587489436e-05,
+      "loss": 0.6277,
+      "step": 4064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.37447168216399e-05,
+      "loss": 0.5513,
+      "step": 4065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3748098055790365e-05,
+      "loss": 0.536,
+      "step": 4066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.375147928994083e-05,
+      "loss": 0.5923,
+      "step": 4067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3754860524091294e-05,
+      "loss": 0.6531,
+      "step": 4068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.375824175824176e-05,
+      "loss": 0.7084,
+      "step": 4069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3761622992392225e-05,
+      "loss": 0.6263,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3765004226542689e-05,
+      "loss": 0.5972,
+      "step": 4071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3768385460693154e-05,
+      "loss": 0.5669,
+      "step": 4072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3771766694843618e-05,
+      "loss": 0.6183,
+      "step": 4073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3775147928994084e-05,
+      "loss": 0.563,
+      "step": 4074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3778529163144549e-05,
+      "loss": 0.5807,
+      "step": 4075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3781910397295014e-05,
+      "loss": 0.5499,
+      "step": 4076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3785291631445478e-05,
+      "loss": 0.728,
+      "step": 4077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3788672865595944e-05,
+      "loss": 0.6934,
+      "step": 4078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3792054099746409e-05,
+      "loss": 0.5593,
+      "step": 4079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3795435333896873e-05,
+      "loss": 0.6133,
+      "step": 4080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3798816568047338e-05,
+      "loss": 0.5559,
+      "step": 4081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3802197802197803e-05,
+      "loss": 0.6255,
+      "step": 4082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3805579036348269e-05,
+      "loss": 0.5099,
+      "step": 4083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3808960270498732e-05,
+      "loss": 0.531,
+      "step": 4084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3812341504649198e-05,
+      "loss": 0.6109,
+      "step": 4085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3815722738799664e-05,
+      "loss": 0.613,
+      "step": 4086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3819103972950127e-05,
+      "loss": 0.5629,
+      "step": 4087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3822485207100593e-05,
+      "loss": 0.5592,
+      "step": 4088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3825866441251056e-05,
+      "loss": 0.6408,
+      "step": 4089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3829247675401522e-05,
+      "loss": 0.6568,
+      "step": 4090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3832628909551989e-05,
+      "loss": 0.602,
+      "step": 4091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3836010143702451e-05,
+      "loss": 0.692,
+      "step": 4092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3839391377852918e-05,
+      "loss": 0.6482,
+      "step": 4093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3842772612003384e-05,
+      "loss": 0.5728,
+      "step": 4094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.5857,
+      "step": 4095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3849535080304313e-05,
+      "loss": 0.6501,
+      "step": 4096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3852916314454776e-05,
+      "loss": 0.6328,
+      "step": 4097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3856297548605242e-05,
+      "loss": 0.5209,
+      "step": 4098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3859678782755708e-05,
+      "loss": 0.6049,
+      "step": 4099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3863060016906171e-05,
+      "loss": 0.5934,
+      "step": 4100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3866441251056637e-05,
+      "loss": 0.6213,
+      "step": 4101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.38698224852071e-05,
+      "loss": 0.5934,
+      "step": 4102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3873203719357567e-05,
+      "loss": 0.571,
+      "step": 4103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3876584953508033e-05,
+      "loss": 0.819,
+      "step": 4104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3879966187658496e-05,
+      "loss": 0.5784,
+      "step": 4105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3883347421808962e-05,
+      "loss": 0.4931,
+      "step": 4106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3886728655959426e-05,
+      "loss": 0.5878,
+      "step": 4107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3890109890109891e-05,
+      "loss": 0.5719,
+      "step": 4108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3893491124260357e-05,
+      "loss": 0.5151,
+      "step": 4109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.389687235841082e-05,
+      "loss": 0.6697,
+      "step": 4110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3900253592561286e-05,
+      "loss": 0.6101,
+      "step": 4111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3903634826711751e-05,
+      "loss": 0.6478,
+      "step": 4112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3907016060862215e-05,
+      "loss": 0.6406,
+      "step": 4113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3910397295012682e-05,
+      "loss": 0.5836,
+      "step": 4114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3913778529163146e-05,
+      "loss": 0.5796,
+      "step": 4115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.391715976331361e-05,
+      "loss": 0.577,
+      "step": 4116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3920540997464075e-05,
+      "loss": 0.6009,
+      "step": 4117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.392392223161454e-05,
+      "loss": 0.5915,
+      "step": 4118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3927303465765006e-05,
+      "loss": 0.7033,
+      "step": 4119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.393068469991547e-05,
+      "loss": 0.6612,
+      "step": 4120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3934065934065935e-05,
+      "loss": 0.5813,
+      "step": 4121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.39374471682164e-05,
+      "loss": 0.5839,
+      "step": 4122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3940828402366866e-05,
+      "loss": 0.628,
+      "step": 4123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3944209636517329e-05,
+      "loss": 0.6077,
+      "step": 4124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3947590870667795e-05,
+      "loss": 0.6963,
+      "step": 4125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.395097210481826e-05,
+      "loss": 0.5732,
+      "step": 4126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3954353338968724e-05,
+      "loss": 0.5958,
+      "step": 4127
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.395773457311919e-05,
+      "loss": 0.6104,
+      "step": 4128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3961115807269653e-05,
+      "loss": 0.6518,
+      "step": 4129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.396449704142012e-05,
+      "loss": 0.633,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3967878275570586e-05,
+      "loss": 0.5469,
+      "step": 4131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3971259509721049e-05,
+      "loss": 0.6355,
+      "step": 4132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3974640743871515e-05,
+      "loss": 0.5606,
+      "step": 4133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3978021978021978e-05,
+      "loss": 0.6635,
+      "step": 4134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3981403212172444e-05,
+      "loss": 0.6783,
+      "step": 4135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.398478444632291e-05,
+      "loss": 0.5238,
+      "step": 4136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3988165680473373e-05,
+      "loss": 0.6218,
+      "step": 4137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.399154691462384e-05,
+      "loss": 0.6118,
+      "step": 4138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3994928148774302e-05,
+      "loss": 0.5776,
+      "step": 4139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3998309382924768e-05,
+      "loss": 0.5309,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4001690617075235e-05,
+      "loss": 0.5911,
+      "step": 4141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4005071851225697e-05,
+      "loss": 0.5239,
+      "step": 4142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4008453085376164e-05,
+      "loss": 0.6162,
+      "step": 4143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.401183431952663e-05,
+      "loss": 0.5883,
+      "step": 4144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4015215553677093e-05,
+      "loss": 0.5776,
+      "step": 4145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4018596787827559e-05,
+      "loss": 0.6387,
+      "step": 4146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4021978021978022e-05,
+      "loss": 0.58,
+      "step": 4147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4025359256128488e-05,
+      "loss": 0.6778,
+      "step": 4148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4028740490278954e-05,
+      "loss": 0.6355,
+      "step": 4149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4032121724429417e-05,
+      "loss": 0.5668,
+      "step": 4150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4035502958579883e-05,
+      "loss": 0.5636,
+      "step": 4151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4038884192730348e-05,
+      "loss": 0.5862,
+      "step": 4152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4042265426880813e-05,
+      "loss": 0.6298,
+      "step": 4153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4045646661031279e-05,
+      "loss": 0.5766,
+      "step": 4154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4049027895181742e-05,
+      "loss": 0.6773,
+      "step": 4155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4052409129332208e-05,
+      "loss": 0.6324,
+      "step": 4156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4055790363482672e-05,
+      "loss": 0.6517,
+      "step": 4157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4059171597633137e-05,
+      "loss": 0.5241,
+      "step": 4158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4062552831783603e-05,
+      "loss": 0.542,
+      "step": 4159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4065934065934068e-05,
+      "loss": 0.7055,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4069315300084532e-05,
+      "loss": 0.6927,
+      "step": 4161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4072696534234997e-05,
+      "loss": 0.6435,
+      "step": 4162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4076077768385461e-05,
+      "loss": 0.6086,
+      "step": 4163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4079459002535926e-05,
+      "loss": 0.623,
+      "step": 4164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4082840236686392e-05,
+      "loss": 0.6686,
+      "step": 4165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4086221470836857e-05,
+      "loss": 0.6126,
+      "step": 4166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4089602704987321e-05,
+      "loss": 0.6628,
+      "step": 4167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4092983939137786e-05,
+      "loss": 0.6109,
+      "step": 4168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.409636517328825e-05,
+      "loss": 0.562,
+      "step": 4169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4099746407438717e-05,
+      "loss": 0.5931,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4103127641589181e-05,
+      "loss": 0.6254,
+      "step": 4171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4106508875739646e-05,
+      "loss": 0.5954,
+      "step": 4172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4109890109890112e-05,
+      "loss": 0.5895,
+      "step": 4173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4113271344040575e-05,
+      "loss": 0.591,
+      "step": 4174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4116652578191041e-05,
+      "loss": 0.6013,
+      "step": 4175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4120033812341506e-05,
+      "loss": 0.5467,
+      "step": 4176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.412341504649197e-05,
+      "loss": 0.6046,
+      "step": 4177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4126796280642436e-05,
+      "loss": 0.6817,
+      "step": 4178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.41301775147929e-05,
+      "loss": 0.585,
+      "step": 4179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4133558748943366e-05,
+      "loss": 0.6636,
+      "step": 4180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4136939983093832e-05,
+      "loss": 0.6152,
+      "step": 4181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4140321217244295e-05,
+      "loss": 0.5915,
+      "step": 4182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4143702451394761e-05,
+      "loss": 0.6849,
+      "step": 4183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4147083685545224e-05,
+      "loss": 0.61,
+      "step": 4184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.415046491969569e-05,
+      "loss": 0.5538,
+      "step": 4185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4153846153846156e-05,
+      "loss": 0.6398,
+      "step": 4186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4157227387996619e-05,
+      "loss": 0.5667,
+      "step": 4187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4160608622147085e-05,
+      "loss": 0.7113,
+      "step": 4188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4163989856297552e-05,
+      "loss": 0.6218,
+      "step": 4189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4167371090448014e-05,
+      "loss": 0.6062,
+      "step": 4190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.417075232459848e-05,
+      "loss": 0.5771,
+      "step": 4191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4174133558748943e-05,
+      "loss": 0.6202,
+      "step": 4192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.417751479289941e-05,
+      "loss": 0.5418,
+      "step": 4193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4180896027049876e-05,
+      "loss": 0.5337,
+      "step": 4194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4184277261200339e-05,
+      "loss": 0.6007,
+      "step": 4195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4187658495350805e-05,
+      "loss": 0.5585,
+      "step": 4196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4191039729501268e-05,
+      "loss": 0.5825,
+      "step": 4197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4194420963651734e-05,
+      "loss": 0.6443,
+      "step": 4198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.41978021978022e-05,
+      "loss": 0.6496,
+      "step": 4199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4201183431952663e-05,
+      "loss": 0.5948,
+      "step": 4200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.420456466610313e-05,
+      "loss": 0.5518,
+      "step": 4201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4207945900253594e-05,
+      "loss": 0.6217,
+      "step": 4202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4211327134404059e-05,
+      "loss": 0.66,
+      "step": 4203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4214708368554523e-05,
+      "loss": 0.7146,
+      "step": 4204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4218089602704988e-05,
+      "loss": 0.5535,
+      "step": 4205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4221470836855454e-05,
+      "loss": 0.6219,
+      "step": 4206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4224852071005918e-05,
+      "loss": 0.5854,
+      "step": 4207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4228233305156383e-05,
+      "loss": 0.5159,
+      "step": 4208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4231614539306848e-05,
+      "loss": 0.6409,
+      "step": 4209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4234995773457314e-05,
+      "loss": 0.6677,
+      "step": 4210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4238377007607778e-05,
+      "loss": 0.6326,
+      "step": 4211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4241758241758243e-05,
+      "loss": 0.6868,
+      "step": 4212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4245139475908707e-05,
+      "loss": 0.5944,
+      "step": 4213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4248520710059172e-05,
+      "loss": 0.5981,
+      "step": 4214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4251901944209638e-05,
+      "loss": 0.633,
+      "step": 4215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4255283178360103e-05,
+      "loss": 0.5369,
+      "step": 4216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4258664412510567e-05,
+      "loss": 0.6626,
+      "step": 4217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4262045646661034e-05,
+      "loss": 0.6924,
+      "step": 4218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4265426880811496e-05,
+      "loss": 0.5984,
+      "step": 4219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4268808114961963e-05,
+      "loss": 0.5701,
+      "step": 4220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4272189349112426e-05,
+      "loss": 0.5551,
+      "step": 4221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4275570583262892e-05,
+      "loss": 0.5995,
+      "step": 4222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4278951817413358e-05,
+      "loss": 0.6045,
+      "step": 4223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4282333051563821e-05,
+      "loss": 0.6523,
+      "step": 4224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.5654,
+      "step": 4225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.428909551986475e-05,
+      "loss": 0.6511,
+      "step": 4226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4292476754015216e-05,
+      "loss": 0.7954,
+      "step": 4227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4295857988165682e-05,
+      "loss": 0.5361,
+      "step": 4228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4299239222316145e-05,
+      "loss": 0.6687,
+      "step": 4229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4302620456466612e-05,
+      "loss": 0.534,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4306001690617078e-05,
+      "loss": 0.655,
+      "step": 4231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.430938292476754e-05,
+      "loss": 0.5448,
+      "step": 4232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4312764158918007e-05,
+      "loss": 0.4888,
+      "step": 4233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.431614539306847e-05,
+      "loss": 0.6627,
+      "step": 4234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4319526627218936e-05,
+      "loss": 0.4808,
+      "step": 4235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4322907861369402e-05,
+      "loss": 0.6264,
+      "step": 4236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4326289095519865e-05,
+      "loss": 0.6555,
+      "step": 4237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4329670329670331e-05,
+      "loss": 0.555,
+      "step": 4238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4333051563820798e-05,
+      "loss": 0.6515,
+      "step": 4239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.433643279797126e-05,
+      "loss": 0.5686,
+      "step": 4240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4339814032121727e-05,
+      "loss": 0.6032,
+      "step": 4241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.434319526627219e-05,
+      "loss": 0.5993,
+      "step": 4242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4346576500422656e-05,
+      "loss": 0.6561,
+      "step": 4243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.434995773457312e-05,
+      "loss": 0.582,
+      "step": 4244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4353338968723585e-05,
+      "loss": 0.5908,
+      "step": 4245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4356720202874051e-05,
+      "loss": 0.5808,
+      "step": 4246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4360101437024516e-05,
+      "loss": 0.64,
+      "step": 4247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.436348267117498e-05,
+      "loss": 0.5887,
+      "step": 4248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4366863905325445e-05,
+      "loss": 0.5793,
+      "step": 4249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.437024513947591e-05,
+      "loss": 0.5658,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4373626373626376e-05,
+      "loss": 0.6657,
+      "step": 4251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.437700760777684e-05,
+      "loss": 0.6353,
+      "step": 4252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4380388841927305e-05,
+      "loss": 0.6533,
+      "step": 4253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.438377007607777e-05,
+      "loss": 0.5811,
+      "step": 4254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4387151310228234e-05,
+      "loss": 0.6132,
+      "step": 4255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.43905325443787e-05,
+      "loss": 0.6025,
+      "step": 4256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4393913778529165e-05,
+      "loss": 0.5861,
+      "step": 4257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4397295012679629e-05,
+      "loss": 0.6073,
+      "step": 4258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4400676246830094e-05,
+      "loss": 0.6038,
+      "step": 4259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.440405748098056e-05,
+      "loss": 0.5372,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4407438715131023e-05,
+      "loss": 0.5173,
+      "step": 4261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4410819949281489e-05,
+      "loss": 0.5948,
+      "step": 4262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4414201183431953e-05,
+      "loss": 0.6373,
+      "step": 4263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4417582417582418e-05,
+      "loss": 0.6754,
+      "step": 4264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4420963651732884e-05,
+      "loss": 0.5922,
+      "step": 4265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4424344885883347e-05,
+      "loss": 0.5899,
+      "step": 4266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4427726120033813e-05,
+      "loss": 0.525,
+      "step": 4267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.443110735418428e-05,
+      "loss": 0.6806,
+      "step": 4268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4434488588334742e-05,
+      "loss": 0.6486,
+      "step": 4269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4437869822485209e-05,
+      "loss": 0.5075,
+      "step": 4270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4441251056635672e-05,
+      "loss": 0.595,
+      "step": 4271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4444632290786138e-05,
+      "loss": 0.5546,
+      "step": 4272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4448013524936604e-05,
+      "loss": 0.5712,
+      "step": 4273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4451394759087067e-05,
+      "loss": 0.7052,
+      "step": 4274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4454775993237533e-05,
+      "loss": 0.6365,
+      "step": 4275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4458157227388e-05,
+      "loss": 0.5854,
+      "step": 4276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4461538461538462e-05,
+      "loss": 0.6007,
+      "step": 4277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4464919695688928e-05,
+      "loss": 0.5286,
+      "step": 4278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4468300929839391e-05,
+      "loss": 0.5927,
+      "step": 4279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4471682163989858e-05,
+      "loss": 0.608,
+      "step": 4280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4475063398140324e-05,
+      "loss": 0.695,
+      "step": 4281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4478444632290787e-05,
+      "loss": 0.6531,
+      "step": 4282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4481825866441253e-05,
+      "loss": 0.69,
+      "step": 4283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4485207100591716e-05,
+      "loss": 0.5243,
+      "step": 4284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4488588334742182e-05,
+      "loss": 0.6012,
+      "step": 4285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4491969568892648e-05,
+      "loss": 0.6264,
+      "step": 4286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4495350803043111e-05,
+      "loss": 0.6846,
+      "step": 4287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4498732037193577e-05,
+      "loss": 0.5809,
+      "step": 4288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4502113271344042e-05,
+      "loss": 0.5876,
+      "step": 4289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4505494505494506e-05,
+      "loss": 0.7282,
+      "step": 4290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4508875739644973e-05,
+      "loss": 0.7263,
+      "step": 4291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4512256973795436e-05,
+      "loss": 0.6179,
+      "step": 4292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4515638207945902e-05,
+      "loss": 0.6016,
+      "step": 4293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4519019442096366e-05,
+      "loss": 0.6211,
+      "step": 4294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4522400676246831e-05,
+      "loss": 0.6053,
+      "step": 4295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4525781910397297e-05,
+      "loss": 0.552,
+      "step": 4296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4529163144547762e-05,
+      "loss": 0.6906,
+      "step": 4297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4532544378698226e-05,
+      "loss": 0.5723,
+      "step": 4298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.453592561284869e-05,
+      "loss": 0.627,
+      "step": 4299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4539306846999155e-05,
+      "loss": 0.7295,
+      "step": 4300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.454268808114962e-05,
+      "loss": 0.531,
+      "step": 4301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4546069315300086e-05,
+      "loss": 0.4972,
+      "step": 4302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.454945054945055e-05,
+      "loss": 0.5944,
+      "step": 4303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4552831783601015e-05,
+      "loss": 0.5675,
+      "step": 4304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4556213017751481e-05,
+      "loss": 0.711,
+      "step": 4305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4559594251901944e-05,
+      "loss": 0.6655,
+      "step": 4306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.456297548605241e-05,
+      "loss": 0.5384,
+      "step": 4307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4566356720202875e-05,
+      "loss": 0.5642,
+      "step": 4308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.456973795435334e-05,
+      "loss": 0.5486,
+      "step": 4309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4573119188503806e-05,
+      "loss": 0.617,
+      "step": 4310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4576500422654269e-05,
+      "loss": 0.5613,
+      "step": 4311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4579881656804735e-05,
+      "loss": 0.5765,
+      "step": 4312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.45832628909552e-05,
+      "loss": 0.5852,
+      "step": 4313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4586644125105664e-05,
+      "loss": 0.5804,
+      "step": 4314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.459002535925613e-05,
+      "loss": 0.5538,
+      "step": 4315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4593406593406593e-05,
+      "loss": 0.6644,
+      "step": 4316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.459678782755706e-05,
+      "loss": 0.5675,
+      "step": 4317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4600169061707526e-05,
+      "loss": 0.5826,
+      "step": 4318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4603550295857989e-05,
+      "loss": 0.509,
+      "step": 4319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4606931530008455e-05,
+      "loss": 0.6824,
+      "step": 4320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4610312764158918e-05,
+      "loss": 0.583,
+      "step": 4321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4613693998309384e-05,
+      "loss": 0.5942,
+      "step": 4322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.461707523245985e-05,
+      "loss": 0.5062,
+      "step": 4323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4620456466610313e-05,
+      "loss": 0.5514,
+      "step": 4324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.462383770076078e-05,
+      "loss": 0.6208,
+      "step": 4325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4627218934911245e-05,
+      "loss": 0.6234,
+      "step": 4326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4630600169061708e-05,
+      "loss": 0.689,
+      "step": 4327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4633981403212175e-05,
+      "loss": 0.567,
+      "step": 4328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4637362637362637e-05,
+      "loss": 0.6301,
+      "step": 4329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4640743871513104e-05,
+      "loss": 0.5976,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.464412510566357e-05,
+      "loss": 0.6001,
+      "step": 4331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4647506339814033e-05,
+      "loss": 0.5307,
+      "step": 4332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4650887573964499e-05,
+      "loss": 0.5029,
+      "step": 4333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4654268808114964e-05,
+      "loss": 0.5725,
+      "step": 4334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4657650042265428e-05,
+      "loss": 0.6683,
+      "step": 4335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4661031276415894e-05,
+      "loss": 0.554,
+      "step": 4336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4664412510566357e-05,
+      "loss": 0.613,
+      "step": 4337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4667793744716823e-05,
+      "loss": 0.676,
+      "step": 4338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4671174978867288e-05,
+      "loss": 0.5261,
+      "step": 4339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4674556213017752e-05,
+      "loss": 0.5393,
+      "step": 4340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4677937447168219e-05,
+      "loss": 0.5312,
+      "step": 4341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4681318681318682e-05,
+      "loss": 0.575,
+      "step": 4342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4684699915469148e-05,
+      "loss": 0.6317,
+      "step": 4343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4688081149619612e-05,
+      "loss": 0.6271,
+      "step": 4344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4691462383770077e-05,
+      "loss": 0.57,
+      "step": 4345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4694843617920541e-05,
+      "loss": 0.6606,
+      "step": 4346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4698224852071008e-05,
+      "loss": 0.5377,
+      "step": 4347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4701606086221472e-05,
+      "loss": 0.5389,
+      "step": 4348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4704987320371937e-05,
+      "loss": 0.694,
+      "step": 4349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4708368554522401e-05,
+      "loss": 0.535,
+      "step": 4350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4711749788672866e-05,
+      "loss": 0.4849,
+      "step": 4351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4715131022823332e-05,
+      "loss": 0.5585,
+      "step": 4352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4718512256973797e-05,
+      "loss": 0.564,
+      "step": 4353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4721893491124261e-05,
+      "loss": 0.6066,
+      "step": 4354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4725274725274727e-05,
+      "loss": 0.5218,
+      "step": 4355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.472865595942519e-05,
+      "loss": 0.6096,
+      "step": 4356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4732037193575657e-05,
+      "loss": 0.5932,
+      "step": 4357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4735418427726121e-05,
+      "loss": 0.5987,
+      "step": 4358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4738799661876586e-05,
+      "loss": 0.6194,
+      "step": 4359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4742180896027052e-05,
+      "loss": 0.6749,
+      "step": 4360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4745562130177515e-05,
+      "loss": 0.5685,
+      "step": 4361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4748943364327981e-05,
+      "loss": 0.5978,
+      "step": 4362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4752324598478447e-05,
+      "loss": 0.6494,
+      "step": 4363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.475570583262891e-05,
+      "loss": 0.5486,
+      "step": 4364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4759087066779376e-05,
+      "loss": 0.5666,
+      "step": 4365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.476246830092984e-05,
+      "loss": 0.5909,
+      "step": 4366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4765849535080305e-05,
+      "loss": 0.6997,
+      "step": 4367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4769230769230772e-05,
+      "loss": 0.5686,
+      "step": 4368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4772612003381235e-05,
+      "loss": 0.6452,
+      "step": 4369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.47759932375317e-05,
+      "loss": 0.5633,
+      "step": 4370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4779374471682164e-05,
+      "loss": 0.6265,
+      "step": 4371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.478275570583263e-05,
+      "loss": 0.5317,
+      "step": 4372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4786136939983096e-05,
+      "loss": 0.564,
+      "step": 4373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4789518174133559e-05,
+      "loss": 0.5851,
+      "step": 4374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4792899408284025e-05,
+      "loss": 0.57,
+      "step": 4375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4796280642434491e-05,
+      "loss": 0.4957,
+      "step": 4376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4799661876584954e-05,
+      "loss": 0.628,
+      "step": 4377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.480304311073542e-05,
+      "loss": 0.5566,
+      "step": 4378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4806424344885883e-05,
+      "loss": 0.6051,
+      "step": 4379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.480980557903635e-05,
+      "loss": 0.6771,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4813186813186816e-05,
+      "loss": 0.587,
+      "step": 4381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4816568047337279e-05,
+      "loss": 0.5315,
+      "step": 4382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4819949281487745e-05,
+      "loss": 0.6279,
+      "step": 4383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.482333051563821e-05,
+      "loss": 0.5954,
+      "step": 4384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4826711749788674e-05,
+      "loss": 0.5315,
+      "step": 4385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4830092983939139e-05,
+      "loss": 0.6105,
+      "step": 4386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4833474218089603e-05,
+      "loss": 0.6249,
+      "step": 4387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.483685545224007e-05,
+      "loss": 0.5082,
+      "step": 4388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4840236686390534e-05,
+      "loss": 0.638,
+      "step": 4389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4843617920540999e-05,
+      "loss": 0.5408,
+      "step": 4390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4846999154691463e-05,
+      "loss": 0.5766,
+      "step": 4391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.485038038884193e-05,
+      "loss": 0.5939,
+      "step": 4392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4853761622992394e-05,
+      "loss": 0.7336,
+      "step": 4393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 0.613,
+      "step": 4394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4860524091293323e-05,
+      "loss": 0.6161,
+      "step": 4395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4863905325443788e-05,
+      "loss": 0.6011,
+      "step": 4396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4867286559594254e-05,
+      "loss": 0.5876,
+      "step": 4397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4870667793744718e-05,
+      "loss": 0.5429,
+      "step": 4398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4874049027895183e-05,
+      "loss": 0.6495,
+      "step": 4399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4877430262045647e-05,
+      "loss": 0.5885,
+      "step": 4400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4880811496196112e-05,
+      "loss": 0.61,
+      "step": 4401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4884192730346578e-05,
+      "loss": 0.5881,
+      "step": 4402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4887573964497041e-05,
+      "loss": 0.5995,
+      "step": 4403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4890955198647507e-05,
+      "loss": 0.6378,
+      "step": 4404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4894336432797974e-05,
+      "loss": 0.5989,
+      "step": 4405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4897717666948436e-05,
+      "loss": 0.7265,
+      "step": 4406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4901098901098903e-05,
+      "loss": 0.5,
+      "step": 4407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4904480135249365e-05,
+      "loss": 0.5771,
+      "step": 4408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4907861369399832e-05,
+      "loss": 0.5459,
+      "step": 4409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4911242603550298e-05,
+      "loss": 0.597,
+      "step": 4410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.491462383770076e-05,
+      "loss": 0.5487,
+      "step": 4411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4918005071851227e-05,
+      "loss": 0.7183,
+      "step": 4412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4921386306001693e-05,
+      "loss": 0.631,
+      "step": 4413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4924767540152156e-05,
+      "loss": 0.5807,
+      "step": 4414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4928148774302622e-05,
+      "loss": 0.7131,
+      "step": 4415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4931530008453085e-05,
+      "loss": 0.6029,
+      "step": 4416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4934911242603551e-05,
+      "loss": 0.6863,
+      "step": 4417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4938292476754018e-05,
+      "loss": 0.5902,
+      "step": 4418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.494167371090448e-05,
+      "loss": 0.586,
+      "step": 4419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4945054945054947e-05,
+      "loss": 0.563,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4948436179205413e-05,
+      "loss": 0.6906,
+      "step": 4421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4951817413355876e-05,
+      "loss": 0.7272,
+      "step": 4422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4955198647506342e-05,
+      "loss": 0.6347,
+      "step": 4423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4958579881656805e-05,
+      "loss": 0.5278,
+      "step": 4424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4961961115807271e-05,
+      "loss": 0.6546,
+      "step": 4425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4965342349957736e-05,
+      "loss": 0.6222,
+      "step": 4426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.49687235841082e-05,
+      "loss": 0.5403,
+      "step": 4427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4972104818258667e-05,
+      "loss": 0.5508,
+      "step": 4428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.497548605240913e-05,
+      "loss": 0.6999,
+      "step": 4429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4978867286559596e-05,
+      "loss": 0.6635,
+      "step": 4430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.498224852071006e-05,
+      "loss": 0.6291,
+      "step": 4431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4985629754860525e-05,
+      "loss": 0.607,
+      "step": 4432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4989010989010991e-05,
+      "loss": 0.6664,
+      "step": 4433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4992392223161456e-05,
+      "loss": 0.6077,
+      "step": 4434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.499577345731192e-05,
+      "loss": 0.5969,
+      "step": 4435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4999154691462385e-05,
+      "loss": 0.6018,
+      "step": 4436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.500253592561285e-05,
+      "loss": 0.6222,
+      "step": 4437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5005917159763315e-05,
+      "loss": 0.5931,
+      "step": 4438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.500929839391378e-05,
+      "loss": 0.6497,
+      "step": 4439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5012679628064245e-05,
+      "loss": 0.5928,
+      "step": 4440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5016060862214709e-05,
+      "loss": 0.7024,
+      "step": 4441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5019442096365175e-05,
+      "loss": 0.4817,
+      "step": 4442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5022823330515638e-05,
+      "loss": 0.6861,
+      "step": 4443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5026204564666104e-05,
+      "loss": 0.5628,
+      "step": 4444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5029585798816569e-05,
+      "loss": 0.5061,
+      "step": 4445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5032967032967034e-05,
+      "loss": 0.6155,
+      "step": 4446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.50363482671175e-05,
+      "loss": 0.5818,
+      "step": 4447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5039729501267963e-05,
+      "loss": 0.5753,
+      "step": 4448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5043110735418429e-05,
+      "loss": 0.62,
+      "step": 4449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5046491969568895e-05,
+      "loss": 0.5647,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5049873203719358e-05,
+      "loss": 0.6663,
+      "step": 4451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5053254437869824e-05,
+      "loss": 0.6069,
+      "step": 4452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5056635672020287e-05,
+      "loss": 0.5527,
+      "step": 4453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5060016906170753e-05,
+      "loss": 0.6024,
+      "step": 4454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.506339814032122e-05,
+      "loss": 0.6069,
+      "step": 4455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5066779374471682e-05,
+      "loss": 0.703,
+      "step": 4456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5070160608622149e-05,
+      "loss": 0.6223,
+      "step": 4457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5073541842772615e-05,
+      "loss": 0.5619,
+      "step": 4458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5076923076923078e-05,
+      "loss": 0.654,
+      "step": 4459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5080304311073544e-05,
+      "loss": 0.5914,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5083685545224007e-05,
+      "loss": 0.6379,
+      "step": 4461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5087066779374473e-05,
+      "loss": 0.651,
+      "step": 4462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.509044801352494e-05,
+      "loss": 0.5918,
+      "step": 4463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5093829247675402e-05,
+      "loss": 0.5703,
+      "step": 4464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5097210481825868e-05,
+      "loss": 0.617,
+      "step": 4465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5100591715976331e-05,
+      "loss": 0.6631,
+      "step": 4466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5103972950126798e-05,
+      "loss": 0.585,
+      "step": 4467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5107354184277264e-05,
+      "loss": 0.6917,
+      "step": 4468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5110735418427727e-05,
+      "loss": 0.6675,
+      "step": 4469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5114116652578193e-05,
+      "loss": 0.5072,
+      "step": 4470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5117497886728657e-05,
+      "loss": 0.6157,
+      "step": 4471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5120879120879122e-05,
+      "loss": 0.6582,
+      "step": 4472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5124260355029588e-05,
+      "loss": 0.5923,
+      "step": 4473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5127641589180051e-05,
+      "loss": 0.5674,
+      "step": 4474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5131022823330517e-05,
+      "loss": 0.5771,
+      "step": 4475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5134404057480982e-05,
+      "loss": 0.566,
+      "step": 4476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5137785291631446e-05,
+      "loss": 0.6054,
+      "step": 4477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5141166525781913e-05,
+      "loss": 0.6456,
+      "step": 4478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5144547759932377e-05,
+      "loss": 0.5786,
+      "step": 4479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5147928994082842e-05,
+      "loss": 0.5302,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5151310228233306e-05,
+      "loss": 0.5423,
+      "step": 4481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.515469146238377e-05,
+      "loss": 0.541,
+      "step": 4482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5158072696534235e-05,
+      "loss": 0.6167,
+      "step": 4483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5161453930684702e-05,
+      "loss": 0.5772,
+      "step": 4484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5164835164835166e-05,
+      "loss": 0.5924,
+      "step": 4485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.516821639898563e-05,
+      "loss": 0.5743,
+      "step": 4486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5171597633136097e-05,
+      "loss": 0.5935,
+      "step": 4487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.517497886728656e-05,
+      "loss": 0.5326,
+      "step": 4488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5178360101437026e-05,
+      "loss": 0.5915,
+      "step": 4489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.518174133558749e-05,
+      "loss": 0.6031,
+      "step": 4490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5185122569737955e-05,
+      "loss": 0.5359,
+      "step": 4491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5188503803888421e-05,
+      "loss": 0.5713,
+      "step": 4492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5191885038038884e-05,
+      "loss": 0.7075,
+      "step": 4493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.519526627218935e-05,
+      "loss": 0.5829,
+      "step": 4494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5198647506339815e-05,
+      "loss": 0.5791,
+      "step": 4495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.520202874049028e-05,
+      "loss": 0.6721,
+      "step": 4496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5205409974640746e-05,
+      "loss": 0.5051,
+      "step": 4497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5208791208791209e-05,
+      "loss": 0.6068,
+      "step": 4498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5212172442941675e-05,
+      "loss": 0.5322,
+      "step": 4499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5215553677092141e-05,
+      "loss": 0.5761,
+      "step": 4500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5218934911242604e-05,
+      "loss": 0.5963,
+      "step": 4501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.522231614539307e-05,
+      "loss": 0.5991,
+      "step": 4502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5225697379543533e-05,
+      "loss": 0.6181,
+      "step": 4503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5229078613694e-05,
+      "loss": 0.6264,
+      "step": 4504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5232459847844466e-05,
+      "loss": 0.5501,
+      "step": 4505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5235841081994928e-05,
+      "loss": 0.5538,
+      "step": 4506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5239222316145395e-05,
+      "loss": 0.6265,
+      "step": 4507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5242603550295861e-05,
+      "loss": 0.5837,
+      "step": 4508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5245984784446324e-05,
+      "loss": 0.7091,
+      "step": 4509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.524936601859679e-05,
+      "loss": 0.5519,
+      "step": 4510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5252747252747253e-05,
+      "loss": 0.6559,
+      "step": 4511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5256128486897719e-05,
+      "loss": 0.5915,
+      "step": 4512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5259509721048184e-05,
+      "loss": 0.5938,
+      "step": 4513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5262890955198647e-05,
+      "loss": 0.7205,
+      "step": 4514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5266272189349113e-05,
+      "loss": 0.6657,
+      "step": 4515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.526965342349958e-05,
+      "loss": 0.6189,
+      "step": 4516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5273034657650042e-05,
+      "loss": 0.5333,
+      "step": 4517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5276415891800508e-05,
+      "loss": 0.5586,
+      "step": 4518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.527979712595097e-05,
+      "loss": 0.565,
+      "step": 4519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5283178360101437e-05,
+      "loss": 0.6517,
+      "step": 4520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5286559594251903e-05,
+      "loss": 0.6115,
+      "step": 4521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5289940828402366e-05,
+      "loss": 0.6475,
+      "step": 4522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5293322062552833e-05,
+      "loss": 0.7171,
+      "step": 4523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5296703296703295e-05,
+      "loss": 0.6203,
+      "step": 4524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.530008453085376e-05,
+      "loss": 0.5546,
+      "step": 4525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5303465765004228e-05,
+      "loss": 0.6386,
+      "step": 4526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.530684699915469e-05,
+      "loss": 0.5526,
+      "step": 4527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5310228233305157e-05,
+      "loss": 0.6342,
+      "step": 4528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5313609467455623e-05,
+      "loss": 0.5678,
+      "step": 4529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5316990701606086e-05,
+      "loss": 0.5931,
+      "step": 4530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5320371935756552e-05,
+      "loss": 0.6141,
+      "step": 4531
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5323753169907015e-05,
+      "loss": 0.5028,
+      "step": 4532
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.532713440405748e-05,
+      "loss": 0.5976,
+      "step": 4533
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5330515638207948e-05,
+      "loss": 0.6447,
+      "step": 4534
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.533389687235841e-05,
+      "loss": 0.6468,
+      "step": 4535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5337278106508877e-05,
+      "loss": 0.5378,
+      "step": 4536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5340659340659343e-05,
+      "loss": 0.6032,
+      "step": 4537
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5344040574809806e-05,
+      "loss": 0.6243,
+      "step": 4538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5347421808960272e-05,
+      "loss": 0.6538,
+      "step": 4539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5350803043110735e-05,
+      "loss": 0.5966,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.53541842772612e-05,
+      "loss": 0.7002,
+      "step": 4541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5357565511411667e-05,
+      "loss": 0.6393,
+      "step": 4542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.536094674556213e-05,
+      "loss": 0.5436,
+      "step": 4543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5364327979712597e-05,
+      "loss": 0.7789,
+      "step": 4544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5367709213863063e-05,
+      "loss": 0.5626,
+      "step": 4545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5371090448013526e-05,
+      "loss": 0.6811,
+      "step": 4546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5374471682163992e-05,
+      "loss": 0.5212,
+      "step": 4547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5377852916314455e-05,
+      "loss": 0.6157,
+      "step": 4548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.538123415046492e-05,
+      "loss": 0.5518,
+      "step": 4549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.6082,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.538799661876585e-05,
+      "loss": 0.5753,
+      "step": 4551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5391377852916316e-05,
+      "loss": 0.5905,
+      "step": 4552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.539475908706678e-05,
+      "loss": 0.5677,
+      "step": 4553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5398140321217245e-05,
+      "loss": 0.6133,
+      "step": 4554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.540152155536771e-05,
+      "loss": 0.6333,
+      "step": 4555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5404902789518174e-05,
+      "loss": 0.6014,
+      "step": 4556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.540828402366864e-05,
+      "loss": 0.7413,
+      "step": 4557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5411665257819107e-05,
+      "loss": 0.6224,
+      "step": 4558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.541504649196957e-05,
+      "loss": 0.5752,
+      "step": 4559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5418427726120036e-05,
+      "loss": 0.5449,
+      "step": 4560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.54218089602705e-05,
+      "loss": 0.5686,
+      "step": 4561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5425190194420965e-05,
+      "loss": 0.5757,
+      "step": 4562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.542857142857143e-05,
+      "loss": 0.5306,
+      "step": 4563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5431952662721894e-05,
+      "loss": 0.6306,
+      "step": 4564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.543533389687236e-05,
+      "loss": 0.5358,
+      "step": 4565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5438715131022827e-05,
+      "loss": 0.6085,
+      "step": 4566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.544209636517329e-05,
+      "loss": 0.5906,
+      "step": 4567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5445477599323756e-05,
+      "loss": 0.6334,
+      "step": 4568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.544885883347422e-05,
+      "loss": 0.5721,
+      "step": 4569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5452240067624685e-05,
+      "loss": 0.7061,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.545562130177515e-05,
+      "loss": 0.6279,
+      "step": 4571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5459002535925614e-05,
+      "loss": 0.5861,
+      "step": 4572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.546238377007608e-05,
+      "loss": 0.587,
+      "step": 4573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5465765004226546e-05,
+      "loss": 0.5781,
+      "step": 4574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.546914623837701e-05,
+      "loss": 0.6949,
+      "step": 4575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5472527472527476e-05,
+      "loss": 0.5533,
+      "step": 4576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.547590870667794e-05,
+      "loss": 0.5885,
+      "step": 4577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5479289940828405e-05,
+      "loss": 0.5453,
+      "step": 4578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.548267117497887e-05,
+      "loss": 0.661,
+      "step": 4579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5486052409129334e-05,
+      "loss": 0.6907,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.54894336432798e-05,
+      "loss": 0.5813,
+      "step": 4581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5492814877430263e-05,
+      "loss": 0.5375,
+      "step": 4582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.549619611158073e-05,
+      "loss": 0.5922,
+      "step": 4583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5499577345731195e-05,
+      "loss": 0.6119,
+      "step": 4584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5502958579881658e-05,
+      "loss": 0.5549,
+      "step": 4585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5506339814032124e-05,
+      "loss": 0.6256,
+      "step": 4586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5509721048182587e-05,
+      "loss": 0.6711,
+      "step": 4587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5513102282333054e-05,
+      "loss": 0.611,
+      "step": 4588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.551648351648352e-05,
+      "loss": 0.6833,
+      "step": 4589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5519864750633983e-05,
+      "loss": 0.5439,
+      "step": 4590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.552324598478445e-05,
+      "loss": 0.6321,
+      "step": 4591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5526627218934912e-05,
+      "loss": 0.66,
+      "step": 4592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5530008453085378e-05,
+      "loss": 0.5668,
+      "step": 4593
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.553338968723584e-05,
+      "loss": 0.6597,
+      "step": 4594
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5536770921386307e-05,
+      "loss": 0.6225,
+      "step": 4595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5540152155536773e-05,
+      "loss": 0.5614,
+      "step": 4596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5543533389687236e-05,
+      "loss": 0.5374,
+      "step": 4597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5546914623837702e-05,
+      "loss": 0.5302,
+      "step": 4598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5550295857988165e-05,
+      "loss": 0.689,
+      "step": 4599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.555367709213863e-05,
+      "loss": 0.6775,
+      "step": 4600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5557058326289098e-05,
+      "loss": 0.6351,
+      "step": 4601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.556043956043956e-05,
+      "loss": 0.5973,
+      "step": 4602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5563820794590027e-05,
+      "loss": 0.6621,
+      "step": 4603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.556720202874049e-05,
+      "loss": 0.5993,
+      "step": 4604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5570583262890956e-05,
+      "loss": 0.6143,
+      "step": 4605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5573964497041422e-05,
+      "loss": 0.6461,
+      "step": 4606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5577345731191885e-05,
+      "loss": 0.5282,
+      "step": 4607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.558072696534235e-05,
+      "loss": 0.6116,
+      "step": 4608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5584108199492814e-05,
+      "loss": 0.5983,
+      "step": 4609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.558748943364328e-05,
+      "loss": 0.6219,
+      "step": 4610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5590870667793743e-05,
+      "loss": 0.6195,
+      "step": 4611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.559425190194421e-05,
+      "loss": 0.633,
+      "step": 4612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5597633136094676e-05,
+      "loss": 0.6534,
+      "step": 4613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.560101437024514e-05,
+      "loss": 0.6251,
+      "step": 4614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5604395604395605e-05,
+      "loss": 0.6131,
+      "step": 4615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.560777683854607e-05,
+      "loss": 0.6636,
+      "step": 4616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5611158072696534e-05,
+      "loss": 0.7477,
+      "step": 4617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5614539306847e-05,
+      "loss": 0.6981,
+      "step": 4618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5617920540997463e-05,
+      "loss": 0.6546,
+      "step": 4619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.562130177514793e-05,
+      "loss": 0.5362,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5624683009298395e-05,
+      "loss": 0.5659,
+      "step": 4621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.562806424344886e-05,
+      "loss": 0.6381,
+      "step": 4622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5631445477599325e-05,
+      "loss": 0.5683,
+      "step": 4623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.563482671174979e-05,
+      "loss": 0.7511,
+      "step": 4624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5638207945900254e-05,
+      "loss": 0.6329,
+      "step": 4625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.564158918005072e-05,
+      "loss": 0.6049,
+      "step": 4626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5644970414201183e-05,
+      "loss": 0.6775,
+      "step": 4627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.564835164835165e-05,
+      "loss": 0.646,
+      "step": 4628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5651732882502115e-05,
+      "loss": 0.6297,
+      "step": 4629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5655114116652578e-05,
+      "loss": 0.6146,
+      "step": 4630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5658495350803044e-05,
+      "loss": 0.5434,
+      "step": 4631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.566187658495351e-05,
+      "loss": 0.5784,
+      "step": 4632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5665257819103973e-05,
+      "loss": 0.6453,
+      "step": 4633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.566863905325444e-05,
+      "loss": 0.6361,
+      "step": 4634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5672020287404903e-05,
+      "loss": 0.6,
+      "step": 4635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.567540152155537e-05,
+      "loss": 0.6413,
+      "step": 4636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5678782755705835e-05,
+      "loss": 0.6432,
+      "step": 4637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5682163989856298e-05,
+      "loss": 0.4825,
+      "step": 4638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5685545224006764e-05,
+      "loss": 0.6523,
+      "step": 4639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5688926458157227e-05,
+      "loss": 0.5304,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5692307692307693e-05,
+      "loss": 0.6217,
+      "step": 4641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.569568892645816e-05,
+      "loss": 0.6733,
+      "step": 4642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5699070160608622e-05,
+      "loss": 0.6191,
+      "step": 4643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.570245139475909e-05,
+      "loss": 0.57,
+      "step": 4644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5705832628909555e-05,
+      "loss": 0.5724,
+      "step": 4645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5709213863060018e-05,
+      "loss": 0.6244,
+      "step": 4646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5712595097210484e-05,
+      "loss": 0.6086,
+      "step": 4647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5715976331360947e-05,
+      "loss": 0.6343,
+      "step": 4648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5719357565511413e-05,
+      "loss": 0.6263,
+      "step": 4649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.572273879966188e-05,
+      "loss": 0.6026,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5726120033812342e-05,
+      "loss": 0.715,
+      "step": 4651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.572950126796281e-05,
+      "loss": 0.585,
+      "step": 4652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5732882502113275e-05,
+      "loss": 0.6077,
+      "step": 4653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5736263736263737e-05,
+      "loss": 0.6677,
+      "step": 4654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5739644970414204e-05,
+      "loss": 0.6644,
+      "step": 4655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5743026204564667e-05,
+      "loss": 0.6969,
+      "step": 4656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5746407438715133e-05,
+      "loss": 0.651,
+      "step": 4657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.57497886728656e-05,
+      "loss": 0.6558,
+      "step": 4658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5753169907016062e-05,
+      "loss": 0.6153,
+      "step": 4659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5756551141166528e-05,
+      "loss": 0.6072,
+      "step": 4660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5759932375316994e-05,
+      "loss": 0.5987,
+      "step": 4661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5763313609467457e-05,
+      "loss": 0.5851,
+      "step": 4662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5766694843617923e-05,
+      "loss": 0.6008,
+      "step": 4663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5770076077768386e-05,
+      "loss": 0.6785,
+      "step": 4664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5773457311918853e-05,
+      "loss": 0.5144,
+      "step": 4665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.577683854606932e-05,
+      "loss": 0.5475,
+      "step": 4666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.578021978021978e-05,
+      "loss": 0.5846,
+      "step": 4667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5783601014370248e-05,
+      "loss": 0.5359,
+      "step": 4668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.578698224852071e-05,
+      "loss": 0.5791,
+      "step": 4669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5790363482671177e-05,
+      "loss": 0.5518,
+      "step": 4670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5793744716821643e-05,
+      "loss": 0.654,
+      "step": 4671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5797125950972106e-05,
+      "loss": 0.5382,
+      "step": 4672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5800507185122572e-05,
+      "loss": 0.6007,
+      "step": 4673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5803888419273035e-05,
+      "loss": 0.6712,
+      "step": 4674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.58072696534235e-05,
+      "loss": 0.6707,
+      "step": 4675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5810650887573968e-05,
+      "loss": 0.5645,
+      "step": 4676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.581403212172443e-05,
+      "loss": 0.627,
+      "step": 4677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5817413355874897e-05,
+      "loss": 0.5167,
+      "step": 4678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.582079459002536e-05,
+      "loss": 0.5149,
+      "step": 4679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5824175824175826e-05,
+      "loss": 0.5294,
+      "step": 4680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5827557058326292e-05,
+      "loss": 0.6046,
+      "step": 4681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5830938292476755e-05,
+      "loss": 0.6365,
+      "step": 4682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.583431952662722e-05,
+      "loss": 0.5883,
+      "step": 4683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5837700760777684e-05,
+      "loss": 0.6393,
+      "step": 4684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.584108199492815e-05,
+      "loss": 0.5092,
+      "step": 4685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5844463229078617e-05,
+      "loss": 0.6821,
+      "step": 4686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.584784446322908e-05,
+      "loss": 0.6325,
+      "step": 4687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5851225697379546e-05,
+      "loss": 0.5246,
+      "step": 4688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.585460693153001e-05,
+      "loss": 0.6646,
+      "step": 4689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5857988165680475e-05,
+      "loss": 0.5511,
+      "step": 4690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.586136939983094e-05,
+      "loss": 0.6181,
+      "step": 4691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5864750633981404e-05,
+      "loss": 0.6968,
+      "step": 4692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.586813186813187e-05,
+      "loss": 0.5725,
+      "step": 4693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5871513102282333e-05,
+      "loss": 0.6478,
+      "step": 4694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.58748943364328e-05,
+      "loss": 0.5793,
+      "step": 4695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5878275570583262e-05,
+      "loss": 0.5713,
+      "step": 4696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5881656804733728e-05,
+      "loss": 0.5997,
+      "step": 4697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5885038038884194e-05,
+      "loss": 0.5072,
+      "step": 4698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5888419273034657e-05,
+      "loss": 0.623,
+      "step": 4699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5891800507185124e-05,
+      "loss": 0.5729,
+      "step": 4700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5895181741335586e-05,
+      "loss": 0.6749,
+      "step": 4701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5898562975486053e-05,
+      "loss": 0.7253,
+      "step": 4702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.590194420963652e-05,
+      "loss": 0.5697,
+      "step": 4703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5905325443786982e-05,
+      "loss": 0.5087,
+      "step": 4704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5908706677937448e-05,
+      "loss": 0.5484,
+      "step": 4705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.591208791208791e-05,
+      "loss": 0.5647,
+      "step": 4706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5915469146238377e-05,
+      "loss": 0.5753,
+      "step": 4707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5918850380388843e-05,
+      "loss": 0.5908,
+      "step": 4708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5922231614539306e-05,
+      "loss": 0.526,
+      "step": 4709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5925612848689772e-05,
+      "loss": 0.6158,
+      "step": 4710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.592899408284024e-05,
+      "loss": 0.5408,
+      "step": 4711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.59323753169907e-05,
+      "loss": 0.5116,
+      "step": 4712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5935756551141168e-05,
+      "loss": 0.5692,
+      "step": 4713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.593913778529163e-05,
+      "loss": 0.6086,
+      "step": 4714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5942519019442097e-05,
+      "loss": 0.509,
+      "step": 4715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5945900253592563e-05,
+      "loss": 0.5985,
+      "step": 4716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5949281487743026e-05,
+      "loss": 0.5991,
+      "step": 4717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5952662721893492e-05,
+      "loss": 0.6305,
+      "step": 4718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.595604395604396e-05,
+      "loss": 0.6646,
+      "step": 4719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.595942519019442e-05,
+      "loss": 0.5347,
+      "step": 4720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5962806424344888e-05,
+      "loss": 0.6484,
+      "step": 4721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.596618765849535e-05,
+      "loss": 0.5822,
+      "step": 4722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5969568892645817e-05,
+      "loss": 0.5736,
+      "step": 4723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5972950126796283e-05,
+      "loss": 0.5761,
+      "step": 4724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5976331360946746e-05,
+      "loss": 0.5664,
+      "step": 4725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5979712595097212e-05,
+      "loss": 0.6105,
+      "step": 4726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5983093829247675e-05,
+      "loss": 0.6059,
+      "step": 4727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.598647506339814e-05,
+      "loss": 0.5334,
+      "step": 4728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5989856297548607e-05,
+      "loss": 0.546,
+      "step": 4729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.599323753169907e-05,
+      "loss": 0.6067,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5996618765849536e-05,
+      "loss": 0.7258,
+      "step": 4731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6022,
+      "step": 4732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6003381234150466e-05,
+      "loss": 0.5712,
+      "step": 4733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6006762468300932e-05,
+      "loss": 0.625,
+      "step": 4734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6010143702451395e-05,
+      "loss": 0.5755,
+      "step": 4735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.601352493660186e-05,
+      "loss": 0.6584,
+      "step": 4736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6016906170752327e-05,
+      "loss": 0.6852,
+      "step": 4737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.602028740490279e-05,
+      "loss": 0.6307,
+      "step": 4738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6023668639053256e-05,
+      "loss": 0.5629,
+      "step": 4739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6027049873203722e-05,
+      "loss": 0.5948,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6030431107354185e-05,
+      "loss": 0.6743,
+      "step": 4741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.603381234150465e-05,
+      "loss": 0.6027,
+      "step": 4742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6037193575655114e-05,
+      "loss": 0.5852,
+      "step": 4743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.604057480980558e-05,
+      "loss": 0.529,
+      "step": 4744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6043956043956047e-05,
+      "loss": 0.6975,
+      "step": 4745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.604733727810651e-05,
+      "loss": 0.7791,
+      "step": 4746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6050718512256976e-05,
+      "loss": 0.6814,
+      "step": 4747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6054099746407442e-05,
+      "loss": 0.5926,
+      "step": 4748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6057480980557905e-05,
+      "loss": 0.5668,
+      "step": 4749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.606086221470837e-05,
+      "loss": 0.5518,
+      "step": 4750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6064243448858834e-05,
+      "loss": 0.7043,
+      "step": 4751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.60676246830093e-05,
+      "loss": 0.6053,
+      "step": 4752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6071005917159767e-05,
+      "loss": 0.6462,
+      "step": 4753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.607438715131023e-05,
+      "loss": 0.61,
+      "step": 4754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6077768385460696e-05,
+      "loss": 0.6049,
+      "step": 4755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.608114961961116e-05,
+      "loss": 0.5933,
+      "step": 4756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6084530853761625e-05,
+      "loss": 0.6608,
+      "step": 4757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.608791208791209e-05,
+      "loss": 0.6329,
+      "step": 4758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6091293322062554e-05,
+      "loss": 0.7499,
+      "step": 4759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.609467455621302e-05,
+      "loss": 0.5555,
+      "step": 4760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6098055790363486e-05,
+      "loss": 0.7003,
+      "step": 4761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.610143702451395e-05,
+      "loss": 0.6496,
+      "step": 4762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6104818258664416e-05,
+      "loss": 0.6524,
+      "step": 4763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.610819949281488e-05,
+      "loss": 0.6127,
+      "step": 4764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6111580726965345e-05,
+      "loss": 0.6162,
+      "step": 4765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.611496196111581e-05,
+      "loss": 0.5585,
+      "step": 4766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6118343195266274e-05,
+      "loss": 0.537,
+      "step": 4767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.612172442941674e-05,
+      "loss": 0.639,
+      "step": 4768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6125105663567203e-05,
+      "loss": 0.5716,
+      "step": 4769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.612848689771767e-05,
+      "loss": 0.6373,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6131868131868135e-05,
+      "loss": 0.6737,
+      "step": 4771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6135249366018598e-05,
+      "loss": 0.5753,
+      "step": 4772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6138630600169064e-05,
+      "loss": 0.7323,
+      "step": 4773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6142011834319527e-05,
+      "loss": 0.6059,
+      "step": 4774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6145393068469993e-05,
+      "loss": 0.6735,
+      "step": 4775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6148774302620456e-05,
+      "loss": 0.5149,
+      "step": 4776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6152155536770923e-05,
+      "loss": 0.6573,
+      "step": 4777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.615553677092139e-05,
+      "loss": 0.7025,
+      "step": 4778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.615891800507185e-05,
+      "loss": 0.6493,
+      "step": 4779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6162299239222318e-05,
+      "loss": 0.7159,
+      "step": 4780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.616568047337278e-05,
+      "loss": 0.5933,
+      "step": 4781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6169061707523247e-05,
+      "loss": 0.5874,
+      "step": 4782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6172442941673713e-05,
+      "loss": 0.6097,
+      "step": 4783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6175824175824176e-05,
+      "loss": 0.5877,
+      "step": 4784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6179205409974642e-05,
+      "loss": 0.6165,
+      "step": 4785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6182586644125105e-05,
+      "loss": 0.6062,
+      "step": 4786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.618596787827557e-05,
+      "loss": 0.663,
+      "step": 4787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6189349112426038e-05,
+      "loss": 0.718,
+      "step": 4788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.61927303465765e-05,
+      "loss": 0.5421,
+      "step": 4789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6196111580726967e-05,
+      "loss": 0.5144,
+      "step": 4790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.619949281487743e-05,
+      "loss": 0.5309,
+      "step": 4791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6202874049027896e-05,
+      "loss": 0.7341,
+      "step": 4792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.620625528317836e-05,
+      "loss": 0.5653,
+      "step": 4793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6209636517328825e-05,
+      "loss": 0.5243,
+      "step": 4794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.621301775147929e-05,
+      "loss": 0.6104,
+      "step": 4795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6216398985629754e-05,
+      "loss": 0.6838,
+      "step": 4796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.621978021978022e-05,
+      "loss": 0.6341,
+      "step": 4797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6223161453930687e-05,
+      "loss": 0.5938,
+      "step": 4798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.622654268808115e-05,
+      "loss": 0.7008,
+      "step": 4799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6229923922231616e-05,
+      "loss": 0.649,
+      "step": 4800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.623330515638208e-05,
+      "loss": 0.64,
+      "step": 4801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6236686390532545e-05,
+      "loss": 0.5849,
+      "step": 4802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.624006762468301e-05,
+      "loss": 0.6135,
+      "step": 4803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6243448858833474e-05,
+      "loss": 0.5949,
+      "step": 4804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.624683009298394e-05,
+      "loss": 0.6085,
+      "step": 4805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6250211327134406e-05,
+      "loss": 0.6399,
+      "step": 4806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.625359256128487e-05,
+      "loss": 0.5801,
+      "step": 4807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6256973795435335e-05,
+      "loss": 0.6471,
+      "step": 4808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6260355029585798e-05,
+      "loss": 0.6012,
+      "step": 4809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6263736263736265e-05,
+      "loss": 0.511,
+      "step": 4810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.626711749788673e-05,
+      "loss": 0.6779,
+      "step": 4811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6270498732037194e-05,
+      "loss": 0.5676,
+      "step": 4812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.627387996618766e-05,
+      "loss": 0.5946,
+      "step": 4813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6277261200338126e-05,
+      "loss": 0.6364,
+      "step": 4814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.628064243448859e-05,
+      "loss": 0.6343,
+      "step": 4815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6284023668639055e-05,
+      "loss": 0.6382,
+      "step": 4816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6287404902789518e-05,
+      "loss": 0.5826,
+      "step": 4817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6290786136939984e-05,
+      "loss": 0.5973,
+      "step": 4818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.629416737109045e-05,
+      "loss": 0.6288,
+      "step": 4819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6297548605240913e-05,
+      "loss": 0.529,
+      "step": 4820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.630092983939138e-05,
+      "loss": 0.5755,
+      "step": 4821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6304311073541842e-05,
+      "loss": 0.5814,
+      "step": 4822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.630769230769231e-05,
+      "loss": 0.504,
+      "step": 4823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6311073541842775e-05,
+      "loss": 0.5336,
+      "step": 4824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6314454775993238e-05,
+      "loss": 0.6097,
+      "step": 4825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6317836010143704e-05,
+      "loss": 0.5537,
+      "step": 4826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.632121724429417e-05,
+      "loss": 0.5655,
+      "step": 4827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6324598478444633e-05,
+      "loss": 0.5878,
+      "step": 4828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.63279797125951e-05,
+      "loss": 0.6182,
+      "step": 4829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6331360946745562e-05,
+      "loss": 0.7665,
+      "step": 4830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.633474218089603e-05,
+      "loss": 0.5659,
+      "step": 4831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6338123415046495e-05,
+      "loss": 0.611,
+      "step": 4832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6341504649196958e-05,
+      "loss": 0.5671,
+      "step": 4833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6344885883347424e-05,
+      "loss": 0.5675,
+      "step": 4834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.634826711749789e-05,
+      "loss": 0.6329,
+      "step": 4835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6351648351648353e-05,
+      "loss": 0.5057,
+      "step": 4836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.635502958579882e-05,
+      "loss": 0.6578,
+      "step": 4837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6358410819949282e-05,
+      "loss": 0.5878,
+      "step": 4838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6361792054099748e-05,
+      "loss": 0.5558,
+      "step": 4839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6365173288250214e-05,
+      "loss": 0.6079,
+      "step": 4840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6368554522400677e-05,
+      "loss": 0.5994,
+      "step": 4841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6371935756551144e-05,
+      "loss": 0.7188,
+      "step": 4842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.637531699070161e-05,
+      "loss": 0.6079,
+      "step": 4843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6378698224852073e-05,
+      "loss": 0.5021,
+      "step": 4844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.638207945900254e-05,
+      "loss": 0.6454,
+      "step": 4845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6385460693153002e-05,
+      "loss": 0.6416,
+      "step": 4846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6388841927303468e-05,
+      "loss": 0.589,
+      "step": 4847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6392223161453934e-05,
+      "loss": 0.5696,
+      "step": 4848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6395604395604397e-05,
+      "loss": 0.5997,
+      "step": 4849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6398985629754863e-05,
+      "loss": 0.579,
+      "step": 4850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6402366863905326e-05,
+      "loss": 0.603,
+      "step": 4851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6405748098055792e-05,
+      "loss": 0.6006,
+      "step": 4852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.640912933220626e-05,
+      "loss": 0.5703,
+      "step": 4853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.641251056635672e-05,
+      "loss": 0.6093,
+      "step": 4854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6415891800507188e-05,
+      "loss": 0.687,
+      "step": 4855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.641927303465765e-05,
+      "loss": 0.5755,
+      "step": 4856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6422654268808117e-05,
+      "loss": 0.6669,
+      "step": 4857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6426035502958583e-05,
+      "loss": 0.6095,
+      "step": 4858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6429416737109046e-05,
+      "loss": 0.6624,
+      "step": 4859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6432797971259512e-05,
+      "loss": 0.6927,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6436179205409975e-05,
+      "loss": 0.6012,
+      "step": 4861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.643956043956044e-05,
+      "loss": 0.6024,
+      "step": 4862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6442941673710908e-05,
+      "loss": 0.5785,
+      "step": 4863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.644632290786137e-05,
+      "loss": 0.5496,
+      "step": 4864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6449704142011837e-05,
+      "loss": 0.579,
+      "step": 4865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.64530853761623e-05,
+      "loss": 0.6599,
+      "step": 4866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6456466610312766e-05,
+      "loss": 0.6627,
+      "step": 4867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6459847844463232e-05,
+      "loss": 0.7188,
+      "step": 4868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6463229078613695e-05,
+      "loss": 0.7484,
+      "step": 4869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.646661031276416e-05,
+      "loss": 0.6438,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6469991546914624e-05,
+      "loss": 0.5583,
+      "step": 4871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.647337278106509e-05,
+      "loss": 0.7287,
+      "step": 4872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6476754015215553e-05,
+      "loss": 0.6378,
+      "step": 4873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.648013524936602e-05,
+      "loss": 0.5984,
+      "step": 4874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6483516483516486e-05,
+      "loss": 0.5754,
+      "step": 4875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.648689771766695e-05,
+      "loss": 0.6484,
+      "step": 4876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6490278951817415e-05,
+      "loss": 0.5973,
+      "step": 4877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6493660185967877e-05,
+      "loss": 0.5588,
+      "step": 4878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6497041420118344e-05,
+      "loss": 0.5681,
+      "step": 4879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.650042265426881e-05,
+      "loss": 0.709,
+      "step": 4880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6503803888419273e-05,
+      "loss": 0.5206,
+      "step": 4881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.650718512256974e-05,
+      "loss": 0.5335,
+      "step": 4882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6510566356720202e-05,
+      "loss": 0.5121,
+      "step": 4883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6513947590870668e-05,
+      "loss": 0.6878,
+      "step": 4884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6517328825021134e-05,
+      "loss": 0.6579,
+      "step": 4885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6520710059171597e-05,
+      "loss": 0.5679,
+      "step": 4886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6524091293322063e-05,
+      "loss": 0.6109,
+      "step": 4887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6527472527472526e-05,
+      "loss": 0.5693,
+      "step": 4888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6530853761622993e-05,
+      "loss": 0.563,
+      "step": 4889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.653423499577346e-05,
+      "loss": 0.615,
+      "step": 4890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.653761622992392e-05,
+      "loss": 0.5615,
+      "step": 4891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6540997464074388e-05,
+      "loss": 0.5478,
+      "step": 4892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6544378698224854e-05,
+      "loss": 0.6695,
+      "step": 4893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6547759932375317e-05,
+      "loss": 0.5796,
+      "step": 4894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6551141166525783e-05,
+      "loss": 0.6035,
+      "step": 4895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6554522400676246e-05,
+      "loss": 0.5176,
+      "step": 4896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6557903634826712e-05,
+      "loss": 0.6206,
+      "step": 4897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.656128486897718e-05,
+      "loss": 0.5951,
+      "step": 4898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.656466610312764e-05,
+      "loss": 0.5827,
+      "step": 4899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6568047337278108e-05,
+      "loss": 0.6177,
+      "step": 4900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6571428571428574e-05,
+      "loss": 0.6336,
+      "step": 4901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6574809805579037e-05,
+      "loss": 0.7552,
+      "step": 4902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6578191039729503e-05,
+      "loss": 0.5107,
+      "step": 4903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6581572273879966e-05,
+      "loss": 0.5171,
+      "step": 4904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6584953508030432e-05,
+      "loss": 0.6054,
+      "step": 4905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.65883347421809e-05,
+      "loss": 0.7221,
+      "step": 4906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.659171597633136e-05,
+      "loss": 0.6775,
+      "step": 4907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6595097210481827e-05,
+      "loss": 0.7027,
+      "step": 4908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.659847844463229e-05,
+      "loss": 0.6897,
+      "step": 4909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6601859678782757e-05,
+      "loss": 0.6202,
+      "step": 4910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6605240912933223e-05,
+      "loss": 0.5599,
+      "step": 4911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6608622147083686e-05,
+      "loss": 0.6828,
+      "step": 4912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6612003381234152e-05,
+      "loss": 0.6319,
+      "step": 4913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6615384615384618e-05,
+      "loss": 0.6109,
+      "step": 4914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.661876584953508e-05,
+      "loss": 0.6207,
+      "step": 4915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6622147083685547e-05,
+      "loss": 0.6052,
+      "step": 4916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.662552831783601e-05,
+      "loss": 0.5613,
+      "step": 4917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6628909551986476e-05,
+      "loss": 0.5826,
+      "step": 4918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6632290786136943e-05,
+      "loss": 0.6657,
+      "step": 4919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6635672020287405e-05,
+      "loss": 0.5992,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.663905325443787e-05,
+      "loss": 0.6153,
+      "step": 4921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6642434488588338e-05,
+      "loss": 0.5558,
+      "step": 4922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.66458157227388e-05,
+      "loss": 0.6803,
+      "step": 4923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6649196956889267e-05,
+      "loss": 0.5296,
+      "step": 4924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.665257819103973e-05,
+      "loss": 0.6584,
+      "step": 4925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6655959425190196e-05,
+      "loss": 0.5579,
+      "step": 4926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6659340659340662e-05,
+      "loss": 0.582,
+      "step": 4927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6662721893491125e-05,
+      "loss": 0.5657,
+      "step": 4928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.666610312764159e-05,
+      "loss": 0.6312,
+      "step": 4929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6669484361792058e-05,
+      "loss": 0.5161,
+      "step": 4930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.667286559594252e-05,
+      "loss": 0.6659,
+      "step": 4931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6676246830092987e-05,
+      "loss": 0.5641,
+      "step": 4932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.667962806424345e-05,
+      "loss": 0.5663,
+      "step": 4933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6683009298393916e-05,
+      "loss": 0.6324,
+      "step": 4934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6686390532544382e-05,
+      "loss": 0.6824,
+      "step": 4935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6689771766694845e-05,
+      "loss": 0.5767,
+      "step": 4936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.669315300084531e-05,
+      "loss": 0.5854,
+      "step": 4937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6696534234995774e-05,
+      "loss": 0.6422,
+      "step": 4938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.669991546914624e-05,
+      "loss": 0.62,
+      "step": 4939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6703296703296707e-05,
+      "loss": 0.5036,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.670667793744717e-05,
+      "loss": 0.6112,
+      "step": 4941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6710059171597636e-05,
+      "loss": 0.6564,
+      "step": 4942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6713440405748102e-05,
+      "loss": 0.6015,
+      "step": 4943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6716821639898565e-05,
+      "loss": 0.6937,
+      "step": 4944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.672020287404903e-05,
+      "loss": 0.5324,
+      "step": 4945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6723584108199494e-05,
+      "loss": 0.6699,
+      "step": 4946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.672696534234996e-05,
+      "loss": 0.5607,
+      "step": 4947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6730346576500426e-05,
+      "loss": 0.5665,
+      "step": 4948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.673372781065089e-05,
+      "loss": 0.5626,
+      "step": 4949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6737109044801355e-05,
+      "loss": 0.7247,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6740490278951818e-05,
+      "loss": 0.4769,
+      "step": 4951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6743871513102285e-05,
+      "loss": 0.5986,
+      "step": 4952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.674725274725275e-05,
+      "loss": 0.7674,
+      "step": 4953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6750633981403214e-05,
+      "loss": 0.6542,
+      "step": 4954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.675401521555368e-05,
+      "loss": 0.6152,
+      "step": 4955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6757396449704143e-05,
+      "loss": 0.6719,
+      "step": 4956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.676077768385461e-05,
+      "loss": 0.6096,
+      "step": 4957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6764158918005072e-05,
+      "loss": 0.7055,
+      "step": 4958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6767540152155538e-05,
+      "loss": 0.6547,
+      "step": 4959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6770921386306004e-05,
+      "loss": 0.5361,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6774302620456467e-05,
+      "loss": 0.5492,
+      "step": 4961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6777683854606933e-05,
+      "loss": 0.5374,
+      "step": 4962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6781065088757396e-05,
+      "loss": 0.4972,
+      "step": 4963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6784446322907862e-05,
+      "loss": 0.6241,
+      "step": 4964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.678782755705833e-05,
+      "loss": 0.6077,
+      "step": 4965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.679120879120879e-05,
+      "loss": 0.5741,
+      "step": 4966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6794590025359258e-05,
+      "loss": 0.5576,
+      "step": 4967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.679797125950972e-05,
+      "loss": 0.6257,
+      "step": 4968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6801352493660187e-05,
+      "loss": 0.7315,
+      "step": 4969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6804733727810653e-05,
+      "loss": 0.6158,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6808114961961116e-05,
+      "loss": 0.6097,
+      "step": 4971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6811496196111582e-05,
+      "loss": 0.5769,
+      "step": 4972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6814877430262045e-05,
+      "loss": 0.593,
+      "step": 4973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.681825866441251e-05,
+      "loss": 0.5725,
+      "step": 4974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6821639898562974e-05,
+      "loss": 0.6831,
+      "step": 4975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.682502113271344e-05,
+      "loss": 0.7039,
+      "step": 4976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6828402366863907e-05,
+      "loss": 0.5869,
+      "step": 4977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.683178360101437e-05,
+      "loss": 0.5562,
+      "step": 4978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6835164835164836e-05,
+      "loss": 0.5548,
+      "step": 4979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6838546069315302e-05,
+      "loss": 0.5555,
+      "step": 4980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6841927303465765e-05,
+      "loss": 0.6583,
+      "step": 4981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.684530853761623e-05,
+      "loss": 0.6628,
+      "step": 4982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6848689771766694e-05,
+      "loss": 0.593,
+      "step": 4983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.685207100591716e-05,
+      "loss": 0.5293,
+      "step": 4984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6855452240067626e-05,
+      "loss": 0.6231,
+      "step": 4985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.685883347421809e-05,
+      "loss": 0.6035,
+      "step": 4986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6862214708368556e-05,
+      "loss": 0.611,
+      "step": 4987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6865595942519022e-05,
+      "loss": 0.6214,
+      "step": 4988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6868977176669485e-05,
+      "loss": 0.5701,
+      "step": 4989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.687235841081995e-05,
+      "loss": 0.5273,
+      "step": 4990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6875739644970414e-05,
+      "loss": 0.5877,
+      "step": 4991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.687912087912088e-05,
+      "loss": 0.7173,
+      "step": 4992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6882502113271346e-05,
+      "loss": 0.551,
+      "step": 4993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.688588334742181e-05,
+      "loss": 0.5461,
+      "step": 4994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6889264581572275e-05,
+      "loss": 0.6306,
+      "step": 4995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6892645815722738e-05,
+      "loss": 0.6125,
+      "step": 4996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6896027049873204e-05,
+      "loss": 0.6534,
+      "step": 4997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.689940828402367e-05,
+      "loss": 0.6134,
+      "step": 4998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6902789518174134e-05,
+      "loss": 0.5536,
+      "step": 4999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.69061707523246e-05,
+      "loss": 0.7142,
+      "step": 5000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6909551986475066e-05,
+      "loss": 0.6053,
+      "step": 5001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.691293322062553e-05,
+      "loss": 0.5545,
+      "step": 5002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6916314454775995e-05,
+      "loss": 0.6056,
+      "step": 5003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6919695688926458e-05,
+      "loss": 0.5826,
+      "step": 5004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.5661,
+      "step": 5005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.692645815722739e-05,
+      "loss": 0.6467,
+      "step": 5006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6929839391377853e-05,
+      "loss": 0.5741,
+      "step": 5007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.693322062552832e-05,
+      "loss": 0.4842,
+      "step": 5008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6936601859678786e-05,
+      "loss": 0.5236,
+      "step": 5009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.693998309382925e-05,
+      "loss": 0.5846,
+      "step": 5010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6943364327979715e-05,
+      "loss": 0.6657,
+      "step": 5011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6946745562130178e-05,
+      "loss": 0.5889,
+      "step": 5012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6950126796280644e-05,
+      "loss": 0.5956,
+      "step": 5013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.695350803043111e-05,
+      "loss": 0.6196,
+      "step": 5014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6956889264581573e-05,
+      "loss": 0.6327,
+      "step": 5015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.696027049873204e-05,
+      "loss": 0.5402,
+      "step": 5016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6963651732882506e-05,
+      "loss": 0.5694,
+      "step": 5017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.696703296703297e-05,
+      "loss": 0.5935,
+      "step": 5018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6970414201183435e-05,
+      "loss": 0.5778,
+      "step": 5019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6973795435333898e-05,
+      "loss": 0.6742,
+      "step": 5020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6977176669484364e-05,
+      "loss": 0.6804,
+      "step": 5021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.698055790363483e-05,
+      "loss": 0.5946,
+      "step": 5022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6983939137785293e-05,
+      "loss": 0.5701,
+      "step": 5023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.698732037193576e-05,
+      "loss": 0.5707,
+      "step": 5024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6990701606086222e-05,
+      "loss": 0.691,
+      "step": 5025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6994082840236688e-05,
+      "loss": 0.6151,
+      "step": 5026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6997464074387154e-05,
+      "loss": 0.6028,
+      "step": 5027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7000845308537617e-05,
+      "loss": 0.6276,
+      "step": 5028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7004226542688084e-05,
+      "loss": 0.5937,
+      "step": 5029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.700760777683855e-05,
+      "loss": 0.5867,
+      "step": 5030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7010989010989013e-05,
+      "loss": 0.597,
+      "step": 5031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.701437024513948e-05,
+      "loss": 0.6017,
+      "step": 5032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7017751479289942e-05,
+      "loss": 0.5822,
+      "step": 5033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7021132713440408e-05,
+      "loss": 0.5402,
+      "step": 5034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7024513947590874e-05,
+      "loss": 0.6338,
+      "step": 5035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7027895181741337e-05,
+      "loss": 0.5435,
+      "step": 5036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7031276415891803e-05,
+      "loss": 0.6068,
+      "step": 5037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7034657650042266e-05,
+      "loss": 0.55,
+      "step": 5038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7038038884192732e-05,
+      "loss": 0.6091,
+      "step": 5039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.70414201183432e-05,
+      "loss": 0.6461,
+      "step": 5040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.704480135249366e-05,
+      "loss": 0.5975,
+      "step": 5041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7048182586644128e-05,
+      "loss": 0.5302,
+      "step": 5042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.705156382079459e-05,
+      "loss": 0.6277,
+      "step": 5043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7054945054945057e-05,
+      "loss": 0.562,
+      "step": 5044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7058326289095523e-05,
+      "loss": 0.5453,
+      "step": 5045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7061707523245986e-05,
+      "loss": 0.5308,
+      "step": 5046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7065088757396452e-05,
+      "loss": 0.5325,
+      "step": 5047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7068469991546915e-05,
+      "loss": 0.5769,
+      "step": 5048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.707185122569738e-05,
+      "loss": 0.6823,
+      "step": 5049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7075232459847847e-05,
+      "loss": 0.6121,
+      "step": 5050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.707861369399831e-05,
+      "loss": 0.6353,
+      "step": 5051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7081994928148777e-05,
+      "loss": 0.6132,
+      "step": 5052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.708537616229924e-05,
+      "loss": 0.5453,
+      "step": 5053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7088757396449706e-05,
+      "loss": 0.6638,
+      "step": 5054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.709213863060017e-05,
+      "loss": 0.5742,
+      "step": 5055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7095519864750635e-05,
+      "loss": 0.5907,
+      "step": 5056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.70989010989011e-05,
+      "loss": 0.596,
+      "step": 5057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7102282333051564e-05,
+      "loss": 0.6208,
+      "step": 5058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.710566356720203e-05,
+      "loss": 0.5718,
+      "step": 5059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7109044801352493e-05,
+      "loss": 0.6297,
+      "step": 5060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.711242603550296e-05,
+      "loss": 0.6702,
+      "step": 5061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7115807269653425e-05,
+      "loss": 0.5486,
+      "step": 5062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.711918850380389e-05,
+      "loss": 0.6432,
+      "step": 5063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7122569737954355e-05,
+      "loss": 0.619,
+      "step": 5064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7125950972104817e-05,
+      "loss": 0.5497,
+      "step": 5065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7129332206255284e-05,
+      "loss": 0.6655,
+      "step": 5066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.713271344040575e-05,
+      "loss": 0.5403,
+      "step": 5067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7136094674556213e-05,
+      "loss": 0.5843,
+      "step": 5068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.713947590870668e-05,
+      "loss": 0.7066,
+      "step": 5069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.6186,
+      "step": 5070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7146238377007608e-05,
+      "loss": 0.5992,
+      "step": 5071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7149619611158074e-05,
+      "loss": 0.6837,
+      "step": 5072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7153000845308537e-05,
+      "loss": 0.5523,
+      "step": 5073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7156382079459003e-05,
+      "loss": 0.6616,
+      "step": 5074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.715976331360947e-05,
+      "loss": 0.6688,
+      "step": 5075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7163144547759933e-05,
+      "loss": 0.6219,
+      "step": 5076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.71665257819104e-05,
+      "loss": 0.6163,
+      "step": 5077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.716990701606086e-05,
+      "loss": 0.612,
+      "step": 5078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7173288250211328e-05,
+      "loss": 0.6105,
+      "step": 5079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7176669484361794e-05,
+      "loss": 0.5951,
+      "step": 5080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7180050718512257e-05,
+      "loss": 0.5672,
+      "step": 5081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7183431952662723e-05,
+      "loss": 0.5486,
+      "step": 5082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7186813186813186e-05,
+      "loss": 0.6093,
+      "step": 5083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7190194420963652e-05,
+      "loss": 0.5426,
+      "step": 5084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.719357565511412e-05,
+      "loss": 0.6788,
+      "step": 5085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.719695688926458e-05,
+      "loss": 0.6405,
+      "step": 5086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7200338123415048e-05,
+      "loss": 0.4928,
+      "step": 5087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7203719357565514e-05,
+      "loss": 0.5764,
+      "step": 5088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7207100591715977e-05,
+      "loss": 0.6413,
+      "step": 5089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7210481825866443e-05,
+      "loss": 0.576,
+      "step": 5090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7213863060016906e-05,
+      "loss": 0.6131,
+      "step": 5091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7217244294167372e-05,
+      "loss": 0.515,
+      "step": 5092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7220625528317838e-05,
+      "loss": 0.6155,
+      "step": 5093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.72240067624683e-05,
+      "loss": 0.6057,
+      "step": 5094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7227387996618767e-05,
+      "loss": 0.634,
+      "step": 5095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7230769230769234e-05,
+      "loss": 0.6371,
+      "step": 5096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7234150464919696e-05,
+      "loss": 0.5642,
+      "step": 5097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7237531699070163e-05,
+      "loss": 0.6223,
+      "step": 5098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7240912933220626e-05,
+      "loss": 0.6712,
+      "step": 5099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7244294167371092e-05,
+      "loss": 0.6155,
+      "step": 5100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7247675401521558e-05,
+      "loss": 0.6345,
+      "step": 5101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.725105663567202e-05,
+      "loss": 0.705,
+      "step": 5102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7254437869822487e-05,
+      "loss": 0.5764,
+      "step": 5103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7257819103972953e-05,
+      "loss": 0.645,
+      "step": 5104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7261200338123416e-05,
+      "loss": 0.6354,
+      "step": 5105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7264581572273883e-05,
+      "loss": 0.6825,
+      "step": 5106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7267962806424345e-05,
+      "loss": 0.6187,
+      "step": 5107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.727134404057481e-05,
+      "loss": 0.5911,
+      "step": 5108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7274725274725278e-05,
+      "loss": 0.596,
+      "step": 5109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.727810650887574e-05,
+      "loss": 0.5826,
+      "step": 5110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7281487743026207e-05,
+      "loss": 0.544,
+      "step": 5111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7284868977176673e-05,
+      "loss": 0.6818,
+      "step": 5112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7288250211327136e-05,
+      "loss": 0.6134,
+      "step": 5113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7291631445477602e-05,
+      "loss": 0.58,
+      "step": 5114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7295012679628065e-05,
+      "loss": 0.6999,
+      "step": 5115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.729839391377853e-05,
+      "loss": 0.4987,
+      "step": 5116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7301775147928998e-05,
+      "loss": 0.6125,
+      "step": 5117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.730515638207946e-05,
+      "loss": 0.5867,
+      "step": 5118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7308537616229927e-05,
+      "loss": 0.5839,
+      "step": 5119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.731191885038039e-05,
+      "loss": 0.5734,
+      "step": 5120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7315300084530856e-05,
+      "loss": 0.6009,
+      "step": 5121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7318681318681322e-05,
+      "loss": 0.6024,
+      "step": 5122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7322062552831785e-05,
+      "loss": 0.5317,
+      "step": 5123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.732544378698225e-05,
+      "loss": 0.5301,
+      "step": 5124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7328825021132717e-05,
+      "loss": 0.5653,
+      "step": 5125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733220625528318e-05,
+      "loss": 0.5503,
+      "step": 5126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7335587489433646e-05,
+      "loss": 0.656,
+      "step": 5127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733896872358411e-05,
+      "loss": 0.6828,
+      "step": 5128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7342349957734576e-05,
+      "loss": 0.4539,
+      "step": 5129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7345731191885042e-05,
+      "loss": 0.5769,
+      "step": 5130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7349112426035505e-05,
+      "loss": 0.6671,
+      "step": 5131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.735249366018597e-05,
+      "loss": 0.6474,
+      "step": 5132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7355874894336434e-05,
+      "loss": 0.8343,
+      "step": 5133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.73592561284869e-05,
+      "loss": 0.7378,
+      "step": 5134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7362637362637363e-05,
+      "loss": 0.5478,
+      "step": 5135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.736601859678783e-05,
+      "loss": 0.5853,
+      "step": 5136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7369399830938295e-05,
+      "loss": 0.4899,
+      "step": 5137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7372781065088758e-05,
+      "loss": 0.6296,
+      "step": 5138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7376162299239224e-05,
+      "loss": 0.6119,
+      "step": 5139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7379543533389687e-05,
+      "loss": 0.6706,
+      "step": 5140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7382924767540154e-05,
+      "loss": 0.5964,
+      "step": 5141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.738630600169062e-05,
+      "loss": 0.6852,
+      "step": 5142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7389687235841083e-05,
+      "loss": 0.7234,
+      "step": 5143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.739306846999155e-05,
+      "loss": 0.6126,
+      "step": 5144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7396449704142012e-05,
+      "loss": 0.5651,
+      "step": 5145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7399830938292478e-05,
+      "loss": 0.542,
+      "step": 5146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7403212172442944e-05,
+      "loss": 0.6289,
+      "step": 5147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7406593406593407e-05,
+      "loss": 0.6249,
+      "step": 5148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7409974640743873e-05,
+      "loss": 0.5899,
+      "step": 5149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7413355874894336e-05,
+      "loss": 0.6543,
+      "step": 5150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7416737109044802e-05,
+      "loss": 0.6134,
+      "step": 5151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7420118343195265e-05,
+      "loss": 0.5923,
+      "step": 5152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.742349957734573e-05,
+      "loss": 0.5859,
+      "step": 5153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7426880811496198e-05,
+      "loss": 0.6089,
+      "step": 5154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.743026204564666e-05,
+      "loss": 0.5589,
+      "step": 5155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7433643279797127e-05,
+      "loss": 0.6338,
+      "step": 5156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.743702451394759e-05,
+      "loss": 0.593,
+      "step": 5157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7440405748098056e-05,
+      "loss": 0.5394,
+      "step": 5158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7443786982248522e-05,
+      "loss": 0.6252,
+      "step": 5159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7447168216398985e-05,
+      "loss": 0.5552,
+      "step": 5160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.745054945054945e-05,
+      "loss": 0.6909,
+      "step": 5161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7453930684699918e-05,
+      "loss": 0.5619,
+      "step": 5162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.745731191885038e-05,
+      "loss": 0.5531,
+      "step": 5163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7460693153000847e-05,
+      "loss": 0.6203,
+      "step": 5164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.746407438715131e-05,
+      "loss": 0.5505,
+      "step": 5165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7467455621301776e-05,
+      "loss": 0.4913,
+      "step": 5166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7470836855452242e-05,
+      "loss": 0.6267,
+      "step": 5167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7474218089602705e-05,
+      "loss": 0.6281,
+      "step": 5168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.747759932375317e-05,
+      "loss": 0.5377,
+      "step": 5169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7480980557903637e-05,
+      "loss": 0.5607,
+      "step": 5170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.74843617920541e-05,
+      "loss": 0.6694,
+      "step": 5171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7487743026204566e-05,
+      "loss": 0.6883,
+      "step": 5172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.749112426035503e-05,
+      "loss": 0.5298,
+      "step": 5173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7494505494505495e-05,
+      "loss": 0.514,
+      "step": 5174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7497886728655962e-05,
+      "loss": 0.5268,
+      "step": 5175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7501267962806425e-05,
+      "loss": 0.6209,
+      "step": 5176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.750464919695689e-05,
+      "loss": 0.6026,
+      "step": 5177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7508030431107354e-05,
+      "loss": 0.6609,
+      "step": 5178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.751141166525782e-05,
+      "loss": 0.5704,
+      "step": 5179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7514792899408286e-05,
+      "loss": 0.6188,
+      "step": 5180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.751817413355875e-05,
+      "loss": 0.6757,
+      "step": 5181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7521555367709215e-05,
+      "loss": 0.6065,
+      "step": 5182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.752493660185968e-05,
+      "loss": 0.5796,
+      "step": 5183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7528317836010144e-05,
+      "loss": 0.5903,
+      "step": 5184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.753169907016061e-05,
+      "loss": 0.619,
+      "step": 5185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7535080304311073e-05,
+      "loss": 0.5516,
+      "step": 5186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.753846153846154e-05,
+      "loss": 0.6717,
+      "step": 5187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7541842772612006e-05,
+      "loss": 0.6606,
+      "step": 5188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.754522400676247e-05,
+      "loss": 0.7189,
+      "step": 5189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7548605240912935e-05,
+      "loss": 0.5279,
+      "step": 5190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.75519864750634e-05,
+      "loss": 0.6218,
+      "step": 5191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7555367709213864e-05,
+      "loss": 0.5791,
+      "step": 5192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.755874894336433e-05,
+      "loss": 0.5583,
+      "step": 5193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7562130177514793e-05,
+      "loss": 0.5484,
+      "step": 5194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.756551141166526e-05,
+      "loss": 0.664,
+      "step": 5195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7568892645815726e-05,
+      "loss": 0.7019,
+      "step": 5196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.757227387996619e-05,
+      "loss": 0.5572,
+      "step": 5197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7575655114116655e-05,
+      "loss": 0.6823,
+      "step": 5198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.757903634826712e-05,
+      "loss": 0.5815,
+      "step": 5199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7582417582417584e-05,
+      "loss": 0.663,
+      "step": 5200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.758579881656805e-05,
+      "loss": 0.6107,
+      "step": 5201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7589180050718513e-05,
+      "loss": 0.601,
+      "step": 5202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.759256128486898e-05,
+      "loss": 0.5667,
+      "step": 5203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7595942519019445e-05,
+      "loss": 0.6059,
+      "step": 5204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.759932375316991e-05,
+      "loss": 0.6152,
+      "step": 5205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7602704987320375e-05,
+      "loss": 0.5777,
+      "step": 5206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7606086221470837e-05,
+      "loss": 0.6496,
+      "step": 5207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7609467455621304e-05,
+      "loss": 0.6753,
+      "step": 5208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.761284868977177e-05,
+      "loss": 0.65,
+      "step": 5209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7616229923922233e-05,
+      "loss": 0.548,
+      "step": 5210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.76196111580727e-05,
+      "loss": 0.6032,
+      "step": 5211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7622992392223165e-05,
+      "loss": 0.6168,
+      "step": 5212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7626373626373628e-05,
+      "loss": 0.6707,
+      "step": 5213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7629754860524094e-05,
+      "loss": 0.6704,
+      "step": 5214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7633136094674557e-05,
+      "loss": 0.7272,
+      "step": 5215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7636517328825023e-05,
+      "loss": 0.6451,
+      "step": 5216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.763989856297549e-05,
+      "loss": 0.5257,
+      "step": 5217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7643279797125953e-05,
+      "loss": 0.5729,
+      "step": 5218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.764666103127642e-05,
+      "loss": 0.6671,
+      "step": 5219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.765004226542688e-05,
+      "loss": 0.5932,
+      "step": 5220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7653423499577348e-05,
+      "loss": 0.649,
+      "step": 5221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7656804733727814e-05,
+      "loss": 0.676,
+      "step": 5222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7660185967878277e-05,
+      "loss": 0.6959,
+      "step": 5223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7663567202028743e-05,
+      "loss": 0.4865,
+      "step": 5224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7666948436179206e-05,
+      "loss": 0.5649,
+      "step": 5225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7670329670329672e-05,
+      "loss": 0.6979,
+      "step": 5226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.767371090448014e-05,
+      "loss": 0.619,
+      "step": 5227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.76770921386306e-05,
+      "loss": 0.6339,
+      "step": 5228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7680473372781068e-05,
+      "loss": 0.5999,
+      "step": 5229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.768385460693153e-05,
+      "loss": 0.5835,
+      "step": 5230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7687235841081997e-05,
+      "loss": 0.6608,
+      "step": 5231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7690617075232463e-05,
+      "loss": 0.6068,
+      "step": 5232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7693998309382926e-05,
+      "loss": 0.5984,
+      "step": 5233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7697379543533392e-05,
+      "loss": 0.5967,
+      "step": 5234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7700760777683855e-05,
+      "loss": 0.5801,
+      "step": 5235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.770414201183432e-05,
+      "loss": 0.6217,
+      "step": 5236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7707523245984784e-05,
+      "loss": 0.6109,
+      "step": 5237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.771090448013525e-05,
+      "loss": 0.6834,
+      "step": 5238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7714285714285717e-05,
+      "loss": 0.5642,
+      "step": 5239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.771766694843618e-05,
+      "loss": 0.5075,
+      "step": 5240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7721048182586646e-05,
+      "loss": 0.7188,
+      "step": 5241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.772442941673711e-05,
+      "loss": 0.5537,
+      "step": 5242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7727810650887575e-05,
+      "loss": 0.6589,
+      "step": 5243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.773119188503804e-05,
+      "loss": 0.5358,
+      "step": 5244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7734573119188504e-05,
+      "loss": 0.5789,
+      "step": 5245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.773795435333897e-05,
+      "loss": 0.545,
+      "step": 5246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7741335587489433e-05,
+      "loss": 0.6635,
+      "step": 5247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.77447168216399e-05,
+      "loss": 0.5413,
+      "step": 5248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7748098055790365e-05,
+      "loss": 0.5354,
+      "step": 5249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7751479289940828e-05,
+      "loss": 0.5998,
+      "step": 5250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7754860524091294e-05,
+      "loss": 0.6171,
+      "step": 5251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7758241758241757e-05,
+      "loss": 0.5759,
+      "step": 5252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7761622992392224e-05,
+      "loss": 0.631,
+      "step": 5253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.776500422654269e-05,
+      "loss": 0.5059,
+      "step": 5254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7768385460693153e-05,
+      "loss": 0.575,
+      "step": 5255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.777176669484362e-05,
+      "loss": 0.5465,
+      "step": 5256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7775147928994085e-05,
+      "loss": 0.5928,
+      "step": 5257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7778529163144548e-05,
+      "loss": 0.5849,
+      "step": 5258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7781910397295014e-05,
+      "loss": 0.6228,
+      "step": 5259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7785291631445477e-05,
+      "loss": 0.6192,
+      "step": 5260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7788672865595943e-05,
+      "loss": 0.5787,
+      "step": 5261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.779205409974641e-05,
+      "loss": 0.6134,
+      "step": 5262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7795435333896872e-05,
+      "loss": 0.5625,
+      "step": 5263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.779881656804734e-05,
+      "loss": 0.6784,
+      "step": 5264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.78021978021978e-05,
+      "loss": 0.6327,
+      "step": 5265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7805579036348268e-05,
+      "loss": 0.5874,
+      "step": 5266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7808960270498734e-05,
+      "loss": 0.6277,
+      "step": 5267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7812341504649197e-05,
+      "loss": 0.5851,
+      "step": 5268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7815722738799663e-05,
+      "loss": 0.6029,
+      "step": 5269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.781910397295013e-05,
+      "loss": 0.5516,
+      "step": 5270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7822485207100592e-05,
+      "loss": 0.5848,
+      "step": 5271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.782586644125106e-05,
+      "loss": 0.59,
+      "step": 5272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.782924767540152e-05,
+      "loss": 0.6492,
+      "step": 5273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7832628909551988e-05,
+      "loss": 0.5158,
+      "step": 5274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7836010143702454e-05,
+      "loss": 0.5803,
+      "step": 5275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7839391377852917e-05,
+      "loss": 0.5218,
+      "step": 5276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7842772612003383e-05,
+      "loss": 0.6122,
+      "step": 5277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.784615384615385e-05,
+      "loss": 0.6345,
+      "step": 5278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7849535080304312e-05,
+      "loss": 0.6067,
+      "step": 5279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7852916314454778e-05,
+      "loss": 0.6367,
+      "step": 5280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.785629754860524e-05,
+      "loss": 0.5311,
+      "step": 5281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7859678782755707e-05,
+      "loss": 0.6084,
+      "step": 5282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7863060016906174e-05,
+      "loss": 0.5562,
+      "step": 5283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7866441251056636e-05,
+      "loss": 0.5474,
+      "step": 5284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7869822485207103e-05,
+      "loss": 0.5518,
+      "step": 5285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.787320371935757e-05,
+      "loss": 0.5952,
+      "step": 5286
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7876584953508032e-05,
+      "loss": 0.6619,
+      "step": 5287
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7879966187658498e-05,
+      "loss": 0.5129,
+      "step": 5288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.788334742180896e-05,
+      "loss": 0.5872,
+      "step": 5289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7886728655959427e-05,
+      "loss": 0.571,
+      "step": 5290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7890109890109893e-05,
+      "loss": 0.571,
+      "step": 5291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7893491124260356e-05,
+      "loss": 0.5816,
+      "step": 5292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7896872358410822e-05,
+      "loss": 0.6334,
+      "step": 5293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7900253592561285e-05,
+      "loss": 0.556,
+      "step": 5294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.790363482671175e-05,
+      "loss": 0.6162,
+      "step": 5295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7907016060862218e-05,
+      "loss": 0.6375,
+      "step": 5296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.791039729501268e-05,
+      "loss": 0.5284,
+      "step": 5297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7913778529163147e-05,
+      "loss": 0.6285,
+      "step": 5298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7917159763313613e-05,
+      "loss": 0.5761,
+      "step": 5299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7920540997464076e-05,
+      "loss": 0.4891,
+      "step": 5300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7923922231614542e-05,
+      "loss": 0.5762,
+      "step": 5301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7927303465765005e-05,
+      "loss": 0.5548,
+      "step": 5302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.793068469991547e-05,
+      "loss": 0.6245,
+      "step": 5303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7934065934065938e-05,
+      "loss": 0.6361,
+      "step": 5304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.79374471682164e-05,
+      "loss": 0.6309,
+      "step": 5305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7940828402366867e-05,
+      "loss": 0.5805,
+      "step": 5306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7944209636517333e-05,
+      "loss": 0.6087,
+      "step": 5307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7947590870667796e-05,
+      "loss": 0.6157,
+      "step": 5308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7950972104818262e-05,
+      "loss": 0.5037,
+      "step": 5309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7954353338968725e-05,
+      "loss": 0.5497,
+      "step": 5310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.795773457311919e-05,
+      "loss": 0.5897,
+      "step": 5311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7961115807269657e-05,
+      "loss": 0.5637,
+      "step": 5312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.796449704142012e-05,
+      "loss": 0.5648,
+      "step": 5313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7967878275570586e-05,
+      "loss": 0.5539,
+      "step": 5314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.797125950972105e-05,
+      "loss": 0.5809,
+      "step": 5315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7974640743871515e-05,
+      "loss": 0.6585,
+      "step": 5316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.797802197802198e-05,
+      "loss": 0.5,
+      "step": 5317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7981403212172445e-05,
+      "loss": 0.5101,
+      "step": 5318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.798478444632291e-05,
+      "loss": 0.6199,
+      "step": 5319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7988165680473374e-05,
+      "loss": 0.5753,
+      "step": 5320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.799154691462384e-05,
+      "loss": 0.6785,
+      "step": 5321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7994928148774303e-05,
+      "loss": 0.6458,
+      "step": 5322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.799830938292477e-05,
+      "loss": 0.6129,
+      "step": 5323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8001690617075235e-05,
+      "loss": 0.544,
+      "step": 5324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8005071851225698e-05,
+      "loss": 0.5934,
+      "step": 5325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8008453085376164e-05,
+      "loss": 0.5376,
+      "step": 5326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8011834319526627e-05,
+      "loss": 0.6383,
+      "step": 5327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8015215553677093e-05,
+      "loss": 0.6343,
+      "step": 5328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.801859678782756e-05,
+      "loss": 0.599,
+      "step": 5329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8021978021978023e-05,
+      "loss": 0.5453,
+      "step": 5330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.802535925612849e-05,
+      "loss": 0.6058,
+      "step": 5331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.802874049027895e-05,
+      "loss": 0.6065,
+      "step": 5332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8032121724429418e-05,
+      "loss": 0.5765,
+      "step": 5333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.803550295857988e-05,
+      "loss": 0.6569,
+      "step": 5334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8038884192730347e-05,
+      "loss": 0.6221,
+      "step": 5335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8042265426880813e-05,
+      "loss": 0.6218,
+      "step": 5336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8045646661031276e-05,
+      "loss": 0.613,
+      "step": 5337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8049027895181742e-05,
+      "loss": 0.6697,
+      "step": 5338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8052409129332205e-05,
+      "loss": 0.6511,
+      "step": 5339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.805579036348267e-05,
+      "loss": 0.6126,
+      "step": 5340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8059171597633138e-05,
+      "loss": 0.5207,
+      "step": 5341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.80625528317836e-05,
+      "loss": 0.6448,
+      "step": 5342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8065934065934067e-05,
+      "loss": 0.6882,
+      "step": 5343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8069315300084533e-05,
+      "loss": 0.6621,
+      "step": 5344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8072696534234996e-05,
+      "loss": 0.5209,
+      "step": 5345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8076077768385462e-05,
+      "loss": 0.6129,
+      "step": 5346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8079459002535925e-05,
+      "loss": 0.5522,
+      "step": 5347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.808284023668639e-05,
+      "loss": 0.613,
+      "step": 5348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8086221470836857e-05,
+      "loss": 0.6064,
+      "step": 5349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.808960270498732e-05,
+      "loss": 0.5694,
+      "step": 5350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8092983939137787e-05,
+      "loss": 0.5977,
+      "step": 5351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.809636517328825e-05,
+      "loss": 0.7373,
+      "step": 5352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8099746407438716e-05,
+      "loss": 0.5559,
+      "step": 5353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8103127641589182e-05,
+      "loss": 0.6217,
+      "step": 5354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8106508875739645e-05,
+      "loss": 0.5384,
+      "step": 5355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.810989010989011e-05,
+      "loss": 0.6852,
+      "step": 5356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8113271344040577e-05,
+      "loss": 0.6337,
+      "step": 5357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.811665257819104e-05,
+      "loss": 0.6172,
+      "step": 5358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8120033812341506e-05,
+      "loss": 0.5687,
+      "step": 5359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.812341504649197e-05,
+      "loss": 0.6691,
+      "step": 5360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8126796280642435e-05,
+      "loss": 0.6054,
+      "step": 5361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.81301775147929e-05,
+      "loss": 0.5734,
+      "step": 5362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8133558748943364e-05,
+      "loss": 0.5279,
+      "step": 5363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.813693998309383e-05,
+      "loss": 0.5581,
+      "step": 5364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8140321217244297e-05,
+      "loss": 0.5217,
+      "step": 5365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.814370245139476e-05,
+      "loss": 0.5964,
+      "step": 5366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8147083685545226e-05,
+      "loss": 0.599,
+      "step": 5367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.815046491969569e-05,
+      "loss": 0.5236,
+      "step": 5368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8153846153846155e-05,
+      "loss": 0.6053,
+      "step": 5369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.815722738799662e-05,
+      "loss": 0.6207,
+      "step": 5370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8160608622147084e-05,
+      "loss": 0.6473,
+      "step": 5371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.816398985629755e-05,
+      "loss": 0.5188,
+      "step": 5372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8167371090448017e-05,
+      "loss": 0.5941,
+      "step": 5373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.817075232459848e-05,
+      "loss": 0.5768,
+      "step": 5374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8174133558748946e-05,
+      "loss": 0.5766,
+      "step": 5375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.817751479289941e-05,
+      "loss": 0.6654,
+      "step": 5376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8180896027049875e-05,
+      "loss": 0.5648,
+      "step": 5377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.818427726120034e-05,
+      "loss": 0.719,
+      "step": 5378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8187658495350804e-05,
+      "loss": 0.624,
+      "step": 5379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.819103972950127e-05,
+      "loss": 0.4943,
+      "step": 5380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8194420963651733e-05,
+      "loss": 0.5731,
+      "step": 5381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.81978021978022e-05,
+      "loss": 0.5369,
+      "step": 5382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8201183431952666e-05,
+      "loss": 0.6067,
+      "step": 5383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.820456466610313e-05,
+      "loss": 0.6266,
+      "step": 5384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8207945900253595e-05,
+      "loss": 0.5409,
+      "step": 5385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.821132713440406e-05,
+      "loss": 0.5802,
+      "step": 5386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8214708368554524e-05,
+      "loss": 0.5779,
+      "step": 5387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.821808960270499e-05,
+      "loss": 0.6499,
+      "step": 5388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8221470836855453e-05,
+      "loss": 0.6596,
+      "step": 5389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.822485207100592e-05,
+      "loss": 0.5244,
+      "step": 5390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8228233305156385e-05,
+      "loss": 0.5586,
+      "step": 5391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8231614539306848e-05,
+      "loss": 0.5939,
+      "step": 5392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8234995773457314e-05,
+      "loss": 0.6456,
+      "step": 5393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.823837700760778e-05,
+      "loss": 0.6766,
+      "step": 5394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8241758241758244e-05,
+      "loss": 0.6082,
+      "step": 5395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.824513947590871e-05,
+      "loss": 0.6877,
+      "step": 5396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8248520710059173e-05,
+      "loss": 0.6142,
+      "step": 5397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.825190194420964e-05,
+      "loss": 0.6052,
+      "step": 5398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8255283178360105e-05,
+      "loss": 0.6074,
+      "step": 5399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8258664412510568e-05,
+      "loss": 0.6257,
+      "step": 5400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8262045646661034e-05,
+      "loss": 0.6059,
+      "step": 5401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8265426880811497e-05,
+      "loss": 0.583,
+      "step": 5402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8268808114961963e-05,
+      "loss": 0.5805,
+      "step": 5403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.827218934911243e-05,
+      "loss": 0.6277,
+      "step": 5404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8275570583262892e-05,
+      "loss": 0.6723,
+      "step": 5405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.827895181741336e-05,
+      "loss": 0.5763,
+      "step": 5406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.828233305156382e-05,
+      "loss": 0.681,
+      "step": 5407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 0.559,
+      "step": 5408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8289095519864754e-05,
+      "loss": 0.7339,
+      "step": 5409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8292476754015217e-05,
+      "loss": 0.5445,
+      "step": 5410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8295857988165683e-05,
+      "loss": 0.6248,
+      "step": 5411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8299239222316146e-05,
+      "loss": 0.6529,
+      "step": 5412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8302620456466612e-05,
+      "loss": 0.6591,
+      "step": 5413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8306001690617075e-05,
+      "loss": 0.6428,
+      "step": 5414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.830938292476754e-05,
+      "loss": 0.6156,
+      "step": 5415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8312764158918008e-05,
+      "loss": 0.5524,
+      "step": 5416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.831614539306847e-05,
+      "loss": 0.4546,
+      "step": 5417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8319526627218937e-05,
+      "loss": 0.5819,
+      "step": 5418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.83229078613694e-05,
+      "loss": 0.6732,
+      "step": 5419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8326289095519866e-05,
+      "loss": 0.5991,
+      "step": 5420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8329670329670332e-05,
+      "loss": 0.627,
+      "step": 5421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8333051563820795e-05,
+      "loss": 0.7297,
+      "step": 5422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.833643279797126e-05,
+      "loss": 0.558,
+      "step": 5423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8339814032121724e-05,
+      "loss": 0.5441,
+      "step": 5424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.834319526627219e-05,
+      "loss": 0.5991,
+      "step": 5425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8346576500422656e-05,
+      "loss": 0.5707,
+      "step": 5426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.834995773457312e-05,
+      "loss": 0.5569,
+      "step": 5427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8353338968723586e-05,
+      "loss": 0.6846,
+      "step": 5428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.835672020287405e-05,
+      "loss": 0.5717,
+      "step": 5429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8360101437024515e-05,
+      "loss": 0.5461,
+      "step": 5430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.836348267117498e-05,
+      "loss": 0.6774,
+      "step": 5431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8366863905325444e-05,
+      "loss": 0.5437,
+      "step": 5432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.837024513947591e-05,
+      "loss": 0.6816,
+      "step": 5433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8373626373626373e-05,
+      "loss": 0.6011,
+      "step": 5434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.837700760777684e-05,
+      "loss": 0.5752,
+      "step": 5435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8380388841927305e-05,
+      "loss": 0.5477,
+      "step": 5436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8383770076077768e-05,
+      "loss": 0.5758,
+      "step": 5437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8387151310228234e-05,
+      "loss": 0.6294,
+      "step": 5438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.83905325443787e-05,
+      "loss": 0.5836,
+      "step": 5439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8393913778529163e-05,
+      "loss": 0.6565,
+      "step": 5440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.839729501267963e-05,
+      "loss": 0.73,
+      "step": 5441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8400676246830093e-05,
+      "loss": 0.563,
+      "step": 5442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.840405748098056e-05,
+      "loss": 0.6257,
+      "step": 5443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8407438715131025e-05,
+      "loss": 0.5552,
+      "step": 5444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8410819949281488e-05,
+      "loss": 0.5956,
+      "step": 5445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8414201183431954e-05,
+      "loss": 0.5706,
+      "step": 5446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8417582417582417e-05,
+      "loss": 0.5978,
+      "step": 5447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8420963651732883e-05,
+      "loss": 0.6873,
+      "step": 5448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.842434488588335e-05,
+      "loss": 0.5315,
+      "step": 5449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8427726120033812e-05,
+      "loss": 0.5817,
+      "step": 5450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.843110735418428e-05,
+      "loss": 0.5717,
+      "step": 5451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8434488588334745e-05,
+      "loss": 0.5854,
+      "step": 5452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8437869822485208e-05,
+      "loss": 0.6187,
+      "step": 5453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8441251056635674e-05,
+      "loss": 0.5991,
+      "step": 5454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8444632290786137e-05,
+      "loss": 0.574,
+      "step": 5455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8448013524936603e-05,
+      "loss": 0.5241,
+      "step": 5456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.845139475908707e-05,
+      "loss": 0.5634,
+      "step": 5457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8454775993237532e-05,
+      "loss": 0.5869,
+      "step": 5458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8458157227388e-05,
+      "loss": 0.5967,
+      "step": 5459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.6017,
+      "step": 5460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8464919695688927e-05,
+      "loss": 0.5992,
+      "step": 5461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8468300929839394e-05,
+      "loss": 0.5008,
+      "step": 5462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8471682163989857e-05,
+      "loss": 0.5891,
+      "step": 5463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8475063398140323e-05,
+      "loss": 0.5708,
+      "step": 5464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.847844463229079e-05,
+      "loss": 0.5125,
+      "step": 5465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8481825866441252e-05,
+      "loss": 0.6752,
+      "step": 5466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8485207100591718e-05,
+      "loss": 0.6385,
+      "step": 5467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8488588334742184e-05,
+      "loss": 0.5132,
+      "step": 5468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8491969568892647e-05,
+      "loss": 0.5099,
+      "step": 5469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8495350803043113e-05,
+      "loss": 0.5711,
+      "step": 5470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8498732037193576e-05,
+      "loss": 0.599,
+      "step": 5471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8502113271344043e-05,
+      "loss": 0.5492,
+      "step": 5472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.850549450549451e-05,
+      "loss": 0.5264,
+      "step": 5473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.850887573964497e-05,
+      "loss": 0.6042,
+      "step": 5474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8512256973795438e-05,
+      "loss": 0.5554,
+      "step": 5475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.85156382079459e-05,
+      "loss": 0.5838,
+      "step": 5476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8519019442096367e-05,
+      "loss": 0.6757,
+      "step": 5477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8522400676246833e-05,
+      "loss": 0.5349,
+      "step": 5478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8525781910397296e-05,
+      "loss": 0.6957,
+      "step": 5479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8529163144547762e-05,
+      "loss": 0.6557,
+      "step": 5480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.853254437869823e-05,
+      "loss": 0.5419,
+      "step": 5481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.853592561284869e-05,
+      "loss": 0.5874,
+      "step": 5482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8539306846999158e-05,
+      "loss": 0.7254,
+      "step": 5483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.854268808114962e-05,
+      "loss": 0.7016,
+      "step": 5484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8546069315300087e-05,
+      "loss": 0.5168,
+      "step": 5485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8549450549450553e-05,
+      "loss": 0.5492,
+      "step": 5486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8552831783601016e-05,
+      "loss": 0.7514,
+      "step": 5487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8556213017751482e-05,
+      "loss": 0.5716,
+      "step": 5488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.855959425190195e-05,
+      "loss": 0.6271,
+      "step": 5489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.856297548605241e-05,
+      "loss": 0.6064,
+      "step": 5490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8566356720202877e-05,
+      "loss": 0.5451,
+      "step": 5491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.856973795435334e-05,
+      "loss": 0.62,
+      "step": 5492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8573119188503807e-05,
+      "loss": 0.6453,
+      "step": 5493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8576500422654273e-05,
+      "loss": 0.6581,
+      "step": 5494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8579881656804736e-05,
+      "loss": 0.611,
+      "step": 5495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8583262890955202e-05,
+      "loss": 0.5766,
+      "step": 5496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8586644125105665e-05,
+      "loss": 0.5861,
+      "step": 5497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.859002535925613e-05,
+      "loss": 0.5519,
+      "step": 5498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8593406593406594e-05,
+      "loss": 0.6155,
+      "step": 5499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.859678782755706e-05,
+      "loss": 0.667,
+      "step": 5500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8600169061707526e-05,
+      "loss": 0.5755,
+      "step": 5501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.860355029585799e-05,
+      "loss": 0.6293,
+      "step": 5502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8606931530008455e-05,
+      "loss": 0.6582,
+      "step": 5503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8610312764158918e-05,
+      "loss": 0.5631,
+      "step": 5504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8613693998309385e-05,
+      "loss": 0.6545,
+      "step": 5505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.861707523245985e-05,
+      "loss": 0.622,
+      "step": 5506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8620456466610314e-05,
+      "loss": 0.6597,
+      "step": 5507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.862383770076078e-05,
+      "loss": 0.5174,
+      "step": 5508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8627218934911243e-05,
+      "loss": 0.5218,
+      "step": 5509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.863060016906171e-05,
+      "loss": 0.6035,
+      "step": 5510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8633981403212175e-05,
+      "loss": 0.5245,
+      "step": 5511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8637362637362638e-05,
+      "loss": 0.6648,
+      "step": 5512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8640743871513104e-05,
+      "loss": 0.4922,
+      "step": 5513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8644125105663567e-05,
+      "loss": 0.6398,
+      "step": 5514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8647506339814033e-05,
+      "loss": 0.5226,
+      "step": 5515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8650887573964496e-05,
+      "loss": 0.5464,
+      "step": 5516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8654268808114962e-05,
+      "loss": 0.5888,
+      "step": 5517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.865765004226543e-05,
+      "loss": 0.597,
+      "step": 5518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.866103127641589e-05,
+      "loss": 0.6417,
+      "step": 5519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8664412510566358e-05,
+      "loss": 0.5853,
+      "step": 5520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.866779374471682e-05,
+      "loss": 0.578,
+      "step": 5521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8671174978867287e-05,
+      "loss": 0.562,
+      "step": 5522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8674556213017753e-05,
+      "loss": 0.5821,
+      "step": 5523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8677937447168216e-05,
+      "loss": 0.5342,
+      "step": 5524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8681318681318682e-05,
+      "loss": 0.4854,
+      "step": 5525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.868469991546915e-05,
+      "loss": 0.6403,
+      "step": 5526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.868808114961961e-05,
+      "loss": 0.5642,
+      "step": 5527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8691462383770078e-05,
+      "loss": 0.6461,
+      "step": 5528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.869484361792054e-05,
+      "loss": 0.5767,
+      "step": 5529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8698224852071007e-05,
+      "loss": 0.5781,
+      "step": 5530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8701606086221473e-05,
+      "loss": 0.6062,
+      "step": 5531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8704987320371936e-05,
+      "loss": 0.6038,
+      "step": 5532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8708368554522402e-05,
+      "loss": 0.6312,
+      "step": 5533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8711749788672865e-05,
+      "loss": 0.64,
+      "step": 5534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871513102282333e-05,
+      "loss": 0.4846,
+      "step": 5535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8718512256973797e-05,
+      "loss": 0.6422,
+      "step": 5536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872189349112426e-05,
+      "loss": 0.5899,
+      "step": 5537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725274725274726e-05,
+      "loss": 0.5187,
+      "step": 5538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728655959425193e-05,
+      "loss": 0.7371,
+      "step": 5539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8732037193575656e-05,
+      "loss": 0.5633,
+      "step": 5540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8735418427726122e-05,
+      "loss": 0.5923,
+      "step": 5541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738799661876585e-05,
+      "loss": 0.5647,
+      "step": 5542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874218089602705e-05,
+      "loss": 0.5983,
+      "step": 5543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8745562130177517e-05,
+      "loss": 0.6187,
+      "step": 5544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874894336432798e-05,
+      "loss": 0.6443,
+      "step": 5545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8752324598478446e-05,
+      "loss": 0.6963,
+      "step": 5546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8755705832628912e-05,
+      "loss": 0.5575,
+      "step": 5547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8759087066779375e-05,
+      "loss": 0.6001,
+      "step": 5548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876246830092984e-05,
+      "loss": 0.6291,
+      "step": 5549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8765849535080304e-05,
+      "loss": 0.5404,
+      "step": 5550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876923076923077e-05,
+      "loss": 0.5401,
+      "step": 5551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8772612003381237e-05,
+      "loss": 0.6885,
+      "step": 5552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87759932375317e-05,
+      "loss": 0.5906,
+      "step": 5553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8779374471682166e-05,
+      "loss": 0.5903,
+      "step": 5554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8782755705832632e-05,
+      "loss": 0.6462,
+      "step": 5555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8786136939983095e-05,
+      "loss": 0.6029,
+      "step": 5556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.878951817413356e-05,
+      "loss": 0.5511,
+      "step": 5557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8792899408284024e-05,
+      "loss": 0.6409,
+      "step": 5558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.879628064243449e-05,
+      "loss": 0.5669,
+      "step": 5559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8799661876584957e-05,
+      "loss": 0.5388,
+      "step": 5560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.880304311073542e-05,
+      "loss": 0.5183,
+      "step": 5561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8806424344885886e-05,
+      "loss": 0.6351,
+      "step": 5562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.880980557903635e-05,
+      "loss": 0.6051,
+      "step": 5563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8813186813186815e-05,
+      "loss": 0.6356,
+      "step": 5564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.881656804733728e-05,
+      "loss": 0.5727,
+      "step": 5565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8819949281487744e-05,
+      "loss": 0.615,
+      "step": 5566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.882333051563821e-05,
+      "loss": 0.5386,
+      "step": 5567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8826711749788676e-05,
+      "loss": 0.6309,
+      "step": 5568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.883009298393914e-05,
+      "loss": 0.6592,
+      "step": 5569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8833474218089606e-05,
+      "loss": 0.541,
+      "step": 5570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.883685545224007e-05,
+      "loss": 0.602,
+      "step": 5571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8840236686390535e-05,
+      "loss": 0.599,
+      "step": 5572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8843617920541e-05,
+      "loss": 0.5977,
+      "step": 5573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8846999154691464e-05,
+      "loss": 0.6384,
+      "step": 5574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.885038038884193e-05,
+      "loss": 0.6255,
+      "step": 5575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8853761622992396e-05,
+      "loss": 0.5584,
+      "step": 5576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 0.5839,
+      "step": 5577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8860524091293325e-05,
+      "loss": 0.5671,
+      "step": 5578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8863905325443788e-05,
+      "loss": 0.7268,
+      "step": 5579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8867286559594254e-05,
+      "loss": 0.6188,
+      "step": 5580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.887066779374472e-05,
+      "loss": 0.5449,
+      "step": 5581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8874049027895184e-05,
+      "loss": 0.6298,
+      "step": 5582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.887743026204565e-05,
+      "loss": 0.4906,
+      "step": 5583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8880811496196113e-05,
+      "loss": 0.5972,
+      "step": 5584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.888419273034658e-05,
+      "loss": 0.5912,
+      "step": 5585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8887573964497045e-05,
+      "loss": 0.664,
+      "step": 5586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8890955198647508e-05,
+      "loss": 0.6073,
+      "step": 5587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8894336432797974e-05,
+      "loss": 0.5471,
+      "step": 5588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8897717666948437e-05,
+      "loss": 0.5573,
+      "step": 5589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8901098901098903e-05,
+      "loss": 0.6753,
+      "step": 5590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.890448013524937e-05,
+      "loss": 0.5857,
+      "step": 5591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8907861369399832e-05,
+      "loss": 0.5955,
+      "step": 5592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.89112426035503e-05,
+      "loss": 0.5493,
+      "step": 5593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.891462383770076e-05,
+      "loss": 0.6148,
+      "step": 5594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8918005071851228e-05,
+      "loss": 0.6066,
+      "step": 5595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.892138630600169e-05,
+      "loss": 0.6096,
+      "step": 5596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8924767540152157e-05,
+      "loss": 0.5262,
+      "step": 5597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8928148774302623e-05,
+      "loss": 0.629,
+      "step": 5598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8931530008453086e-05,
+      "loss": 0.5282,
+      "step": 5599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8934911242603552e-05,
+      "loss": 0.6518,
+      "step": 5600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8938292476754015e-05,
+      "loss": 0.6579,
+      "step": 5601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894167371090448e-05,
+      "loss": 0.581,
+      "step": 5602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8945054945054947e-05,
+      "loss": 0.6649,
+      "step": 5603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894843617920541e-05,
+      "loss": 0.616,
+      "step": 5604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8951817413355877e-05,
+      "loss": 0.6049,
+      "step": 5605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895519864750634e-05,
+      "loss": 0.5305,
+      "step": 5606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8958579881656806e-05,
+      "loss": 0.5812,
+      "step": 5607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8961961115807272e-05,
+      "loss": 0.5778,
+      "step": 5608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8965342349957735e-05,
+      "loss": 0.6136,
+      "step": 5609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.89687235841082e-05,
+      "loss": 0.682,
+      "step": 5610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8972104818258664e-05,
+      "loss": 0.5958,
+      "step": 5611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.897548605240913e-05,
+      "loss": 0.6162,
+      "step": 5612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8978867286559596e-05,
+      "loss": 0.6007,
+      "step": 5613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.898224852071006e-05,
+      "loss": 0.545,
+      "step": 5614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8985629754860525e-05,
+      "loss": 0.5147,
+      "step": 5615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8989010989010988e-05,
+      "loss": 0.6817,
+      "step": 5616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8992392223161455e-05,
+      "loss": 0.6399,
+      "step": 5617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.899577345731192e-05,
+      "loss": 0.6247,
+      "step": 5618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8999154691462384e-05,
+      "loss": 0.7004,
+      "step": 5619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.900253592561285e-05,
+      "loss": 0.6164,
+      "step": 5620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9005917159763313e-05,
+      "loss": 0.5657,
+      "step": 5621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.900929839391378e-05,
+      "loss": 0.7144,
+      "step": 5622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9012679628064245e-05,
+      "loss": 0.5865,
+      "step": 5623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9016060862214708e-05,
+      "loss": 0.5767,
+      "step": 5624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9019442096365174e-05,
+      "loss": 0.5428,
+      "step": 5625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.902282333051564e-05,
+      "loss": 0.5723,
+      "step": 5626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9026204564666103e-05,
+      "loss": 0.674,
+      "step": 5627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.902958579881657e-05,
+      "loss": 0.6008,
+      "step": 5628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9032967032967033e-05,
+      "loss": 0.5548,
+      "step": 5629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.90363482671175e-05,
+      "loss": 0.6114,
+      "step": 5630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9039729501267965e-05,
+      "loss": 0.6062,
+      "step": 5631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9043110735418428e-05,
+      "loss": 0.6395,
+      "step": 5632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9046491969568894e-05,
+      "loss": 0.5275,
+      "step": 5633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.904987320371936e-05,
+      "loss": 0.5465,
+      "step": 5634
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9053254437869823e-05,
+      "loss": 0.6257,
+      "step": 5635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.905663567202029e-05,
+      "loss": 0.5963,
+      "step": 5636
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9060016906170752e-05,
+      "loss": 0.5619,
+      "step": 5637
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.906339814032122e-05,
+      "loss": 0.504,
+      "step": 5638
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9066779374471685e-05,
+      "loss": 0.6331,
+      "step": 5639
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9070160608622148e-05,
+      "loss": 0.6469,
+      "step": 5640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9073541842772614e-05,
+      "loss": 0.5558,
+      "step": 5641
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.907692307692308e-05,
+      "loss": 0.5755,
+      "step": 5642
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9080304311073543e-05,
+      "loss": 0.5464,
+      "step": 5643
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.908368554522401e-05,
+      "loss": 0.6051,
+      "step": 5644
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9087066779374472e-05,
+      "loss": 0.612,
+      "step": 5645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9090448013524938e-05,
+      "loss": 0.578,
+      "step": 5646
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9093829247675405e-05,
+      "loss": 0.5508,
+      "step": 5647
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9097210481825867e-05,
+      "loss": 0.6328,
+      "step": 5648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9100591715976334e-05,
+      "loss": 0.5501,
+      "step": 5649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9103972950126796e-05,
+      "loss": 0.6073,
+      "step": 5650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9107354184277263e-05,
+      "loss": 0.6539,
+      "step": 5651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.911073541842773e-05,
+      "loss": 0.5553,
+      "step": 5652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9114116652578192e-05,
+      "loss": 0.6671,
+      "step": 5653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9117497886728658e-05,
+      "loss": 0.6532,
+      "step": 5654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9120879120879124e-05,
+      "loss": 0.6073,
+      "step": 5655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9124260355029587e-05,
+      "loss": 0.6105,
+      "step": 5656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9127641589180053e-05,
+      "loss": 0.5537,
+      "step": 5657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9131022823330516e-05,
+      "loss": 0.5721,
+      "step": 5658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9134404057480982e-05,
+      "loss": 0.5338,
+      "step": 5659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.913778529163145e-05,
+      "loss": 0.5046,
+      "step": 5660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.914116652578191e-05,
+      "loss": 0.6112,
+      "step": 5661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9144547759932378e-05,
+      "loss": 0.5541,
+      "step": 5662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9147928994082844e-05,
+      "loss": 0.6555,
+      "step": 5663
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9151310228233307e-05,
+      "loss": 0.6653,
+      "step": 5664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9154691462383773e-05,
+      "loss": 0.5684,
+      "step": 5665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9158072696534236e-05,
+      "loss": 0.6458,
+      "step": 5666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9161453930684702e-05,
+      "loss": 0.6648,
+      "step": 5667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.916483516483517e-05,
+      "loss": 0.6322,
+      "step": 5668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.916821639898563e-05,
+      "loss": 0.6748,
+      "step": 5669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9171597633136098e-05,
+      "loss": 0.5558,
+      "step": 5670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9174978867286564e-05,
+      "loss": 0.5708,
+      "step": 5671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9178360101437027e-05,
+      "loss": 0.6304,
+      "step": 5672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9181741335587493e-05,
+      "loss": 0.6614,
+      "step": 5673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9185122569737956e-05,
+      "loss": 0.6816,
+      "step": 5674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9188503803888422e-05,
+      "loss": 0.6058,
+      "step": 5675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9191885038038885e-05,
+      "loss": 0.616,
+      "step": 5676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.919526627218935e-05,
+      "loss": 0.6963,
+      "step": 5677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9198647506339817e-05,
+      "loss": 0.5945,
+      "step": 5678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.920202874049028e-05,
+      "loss": 0.5839,
+      "step": 5679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9205409974640746e-05,
+      "loss": 0.5471,
+      "step": 5680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.920879120879121e-05,
+      "loss": 0.7826,
+      "step": 5681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9212172442941676e-05,
+      "loss": 0.5645,
+      "step": 5682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9215553677092142e-05,
+      "loss": 0.65,
+      "step": 5683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9218934911242605e-05,
+      "loss": 0.6663,
+      "step": 5684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.922231614539307e-05,
+      "loss": 0.5403,
+      "step": 5685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9225697379543534e-05,
+      "loss": 0.5431,
+      "step": 5686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9229078613694e-05,
+      "loss": 0.6377,
+      "step": 5687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9232459847844466e-05,
+      "loss": 0.5834,
+      "step": 5688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.923584108199493e-05,
+      "loss": 0.5487,
+      "step": 5689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9239222316145395e-05,
+      "loss": 0.722,
+      "step": 5690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9242603550295858e-05,
+      "loss": 0.6795,
+      "step": 5691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9245984784446324e-05,
+      "loss": 0.6598,
+      "step": 5692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.924936601859679e-05,
+      "loss": 0.6212,
+      "step": 5693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9252747252747254e-05,
+      "loss": 0.618,
+      "step": 5694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.925612848689772e-05,
+      "loss": 0.6794,
+      "step": 5695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9259509721048183e-05,
+      "loss": 0.5634,
+      "step": 5696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.926289095519865e-05,
+      "loss": 0.6414,
+      "step": 5697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9266272189349112e-05,
+      "loss": 0.5065,
+      "step": 5698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9269653423499578e-05,
+      "loss": 0.5843,
+      "step": 5699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9273034657650044e-05,
+      "loss": 0.6131,
+      "step": 5700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9276415891800507e-05,
+      "loss": 0.6351,
+      "step": 5701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9279797125950973e-05,
+      "loss": 0.5343,
+      "step": 5702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9283178360101436e-05,
+      "loss": 0.6415,
+      "step": 5703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9286559594251902e-05,
+      "loss": 0.5222,
+      "step": 5704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.928994082840237e-05,
+      "loss": 0.5747,
+      "step": 5705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.929332206255283e-05,
+      "loss": 0.5526,
+      "step": 5706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9296703296703298e-05,
+      "loss": 0.5722,
+      "step": 5707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.930008453085376e-05,
+      "loss": 0.6255,
+      "step": 5708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9303465765004227e-05,
+      "loss": 0.5913,
+      "step": 5709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9306846999154693e-05,
+      "loss": 0.5857,
+      "step": 5710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9310228233305156e-05,
+      "loss": 0.5888,
+      "step": 5711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9313609467455622e-05,
+      "loss": 0.6191,
+      "step": 5712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.931699070160609e-05,
+      "loss": 0.5706,
+      "step": 5713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.932037193575655e-05,
+      "loss": 0.6304,
+      "step": 5714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9323753169907018e-05,
+      "loss": 0.6379,
+      "step": 5715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.932713440405748e-05,
+      "loss": 0.5269,
+      "step": 5716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9330515638207947e-05,
+      "loss": 0.5579,
+      "step": 5717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9333896872358413e-05,
+      "loss": 0.5674,
+      "step": 5718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9337278106508876e-05,
+      "loss": 0.615,
+      "step": 5719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9340659340659342e-05,
+      "loss": 0.6448,
+      "step": 5720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9344040574809808e-05,
+      "loss": 0.6289,
+      "step": 5721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.934742180896027e-05,
+      "loss": 0.5809,
+      "step": 5722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9350803043110737e-05,
+      "loss": 0.6636,
+      "step": 5723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.93541842772612e-05,
+      "loss": 0.5347,
+      "step": 5724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9357565511411666e-05,
+      "loss": 0.7216,
+      "step": 5725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9360946745562133e-05,
+      "loss": 0.5804,
+      "step": 5726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9364327979712595e-05,
+      "loss": 0.5498,
+      "step": 5727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9367709213863062e-05,
+      "loss": 0.5594,
+      "step": 5728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9371090448013528e-05,
+      "loss": 0.5602,
+      "step": 5729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.937447168216399e-05,
+      "loss": 0.6511,
+      "step": 5730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9377852916314457e-05,
+      "loss": 0.5864,
+      "step": 5731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.938123415046492e-05,
+      "loss": 0.6292,
+      "step": 5732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9384615384615386e-05,
+      "loss": 0.6421,
+      "step": 5733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9387996618765852e-05,
+      "loss": 0.593,
+      "step": 5734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9391377852916315e-05,
+      "loss": 0.6609,
+      "step": 5735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.939475908706678e-05,
+      "loss": 0.6653,
+      "step": 5736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9398140321217244e-05,
+      "loss": 0.6404,
+      "step": 5737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.940152155536771e-05,
+      "loss": 0.5955,
+      "step": 5738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9404902789518177e-05,
+      "loss": 0.5682,
+      "step": 5739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.940828402366864e-05,
+      "loss": 0.724,
+      "step": 5740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9411665257819106e-05,
+      "loss": 0.7058,
+      "step": 5741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9415046491969572e-05,
+      "loss": 0.6044,
+      "step": 5742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9418427726120035e-05,
+      "loss": 0.6603,
+      "step": 5743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.94218089602705e-05,
+      "loss": 0.6245,
+      "step": 5744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9425190194420964e-05,
+      "loss": 0.6837,
+      "step": 5745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 0.5414,
+      "step": 5746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9431952662721897e-05,
+      "loss": 0.6506,
+      "step": 5747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.943533389687236e-05,
+      "loss": 0.5567,
+      "step": 5748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9438715131022826e-05,
+      "loss": 0.6612,
+      "step": 5749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9442096365173292e-05,
+      "loss": 0.5212,
+      "step": 5750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9445477599323755e-05,
+      "loss": 0.67,
+      "step": 5751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.944885883347422e-05,
+      "loss": 0.5437,
+      "step": 5752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9452240067624684e-05,
+      "loss": 0.6376,
+      "step": 5753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.945562130177515e-05,
+      "loss": 0.5868,
+      "step": 5754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9459002535925616e-05,
+      "loss": 0.6222,
+      "step": 5755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.946238377007608e-05,
+      "loss": 0.5865,
+      "step": 5756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9465765004226545e-05,
+      "loss": 0.6003,
+      "step": 5757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9469146238377012e-05,
+      "loss": 0.5816,
+      "step": 5758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9472527472527475e-05,
+      "loss": 0.6321,
+      "step": 5759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.947590870667794e-05,
+      "loss": 0.5193,
+      "step": 5760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9479289940828404e-05,
+      "loss": 0.5623,
+      "step": 5761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.948267117497887e-05,
+      "loss": 0.6471,
+      "step": 5762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9486052409129336e-05,
+      "loss": 0.6057,
+      "step": 5763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.94894336432798e-05,
+      "loss": 0.6111,
+      "step": 5764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9492814877430265e-05,
+      "loss": 0.4649,
+      "step": 5765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9496196111580728e-05,
+      "loss": 0.685,
+      "step": 5766
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9499577345731194e-05,
+      "loss": 0.5386,
+      "step": 5767
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.950295857988166e-05,
+      "loss": 0.6833,
+      "step": 5768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9506339814032123e-05,
+      "loss": 0.6417,
+      "step": 5769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.950972104818259e-05,
+      "loss": 0.6844,
+      "step": 5770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9513102282333053e-05,
+      "loss": 0.5776,
+      "step": 5771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.951648351648352e-05,
+      "loss": 0.6676,
+      "step": 5772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9519864750633985e-05,
+      "loss": 0.5802,
+      "step": 5773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9523245984784448e-05,
+      "loss": 0.5938,
+      "step": 5774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9526627218934914e-05,
+      "loss": 0.6273,
+      "step": 5775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9530008453085377e-05,
+      "loss": 0.6081,
+      "step": 5776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9533389687235843e-05,
+      "loss": 0.6131,
+      "step": 5777
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9536770921386306e-05,
+      "loss": 0.6494,
+      "step": 5778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9540152155536772e-05,
+      "loss": 0.5666,
+      "step": 5779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.954353338968724e-05,
+      "loss": 0.5483,
+      "step": 5780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.95469146238377e-05,
+      "loss": 0.6279,
+      "step": 5781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9550295857988168e-05,
+      "loss": 0.5405,
+      "step": 5782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.955367709213863e-05,
+      "loss": 0.6062,
+      "step": 5783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9557058326289097e-05,
+      "loss": 0.6126,
+      "step": 5784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9560439560439563e-05,
+      "loss": 0.6313,
+      "step": 5785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9563820794590026e-05,
+      "loss": 0.5365,
+      "step": 5786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9567202028740492e-05,
+      "loss": 0.6129,
+      "step": 5787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9570583262890955e-05,
+      "loss": 0.5146,
+      "step": 5788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.957396449704142e-05,
+      "loss": 0.5725,
+      "step": 5789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9577345731191887e-05,
+      "loss": 0.5699,
+      "step": 5790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958072696534235e-05,
+      "loss": 0.5603,
+      "step": 5791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9584108199492817e-05,
+      "loss": 0.6271,
+      "step": 5792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958748943364328e-05,
+      "loss": 0.6264,
+      "step": 5793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590870667793746e-05,
+      "loss": 0.4938,
+      "step": 5794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9594251901944212e-05,
+      "loss": 0.521,
+      "step": 5795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597633136094675e-05,
+      "loss": 0.659,
+      "step": 5796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960101437024514e-05,
+      "loss": 0.6032,
+      "step": 5797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604395604395604e-05,
+      "loss": 0.5666,
+      "step": 5798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960777683854607e-05,
+      "loss": 0.6311,
+      "step": 5799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9611158072696536e-05,
+      "loss": 0.643,
+      "step": 5800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9614539306847e-05,
+      "loss": 0.6014,
+      "step": 5801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9617920540997465e-05,
+      "loss": 0.5362,
+      "step": 5802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9621301775147928e-05,
+      "loss": 0.7588,
+      "step": 5803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9624683009298394e-05,
+      "loss": 0.7051,
+      "step": 5804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.962806424344886e-05,
+      "loss": 0.597,
+      "step": 5805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9631445477599324e-05,
+      "loss": 0.5367,
+      "step": 5806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.963482671174979e-05,
+      "loss": 0.567,
+      "step": 5807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9638207945900256e-05,
+      "loss": 0.6115,
+      "step": 5808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.964158918005072e-05,
+      "loss": 0.5842,
+      "step": 5809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9644970414201185e-05,
+      "loss": 0.666,
+      "step": 5810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9648351648351648e-05,
+      "loss": 0.6048,
+      "step": 5811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9651732882502114e-05,
+      "loss": 0.5581,
+      "step": 5812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.965511411665258e-05,
+      "loss": 0.5731,
+      "step": 5813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9658495350803043e-05,
+      "loss": 0.5038,
+      "step": 5814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.966187658495351e-05,
+      "loss": 0.5856,
+      "step": 5815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9665257819103976e-05,
+      "loss": 0.5658,
+      "step": 5816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.966863905325444e-05,
+      "loss": 0.5276,
+      "step": 5817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9672020287404905e-05,
+      "loss": 0.6404,
+      "step": 5818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9675401521555368e-05,
+      "loss": 0.5122,
+      "step": 5819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9678782755705834e-05,
+      "loss": 0.6398,
+      "step": 5820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.96821639898563e-05,
+      "loss": 0.6377,
+      "step": 5821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9685545224006763e-05,
+      "loss": 0.5696,
+      "step": 5822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.968892645815723e-05,
+      "loss": 0.7093,
+      "step": 5823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9692307692307696e-05,
+      "loss": 0.5983,
+      "step": 5824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.969568892645816e-05,
+      "loss": 0.6334,
+      "step": 5825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9699070160608625e-05,
+      "loss": 0.6487,
+      "step": 5826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9702451394759088e-05,
+      "loss": 0.5361,
+      "step": 5827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9705832628909554e-05,
+      "loss": 0.6636,
+      "step": 5828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.970921386306002e-05,
+      "loss": 0.5511,
+      "step": 5829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9712595097210483e-05,
+      "loss": 0.5552,
+      "step": 5830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.971597633136095e-05,
+      "loss": 0.6058,
+      "step": 5831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9719357565511412e-05,
+      "loss": 0.529,
+      "step": 5832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9722738799661878e-05,
+      "loss": 0.582,
+      "step": 5833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9726120033812344e-05,
+      "loss": 0.6081,
+      "step": 5834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9729501267962807e-05,
+      "loss": 0.598,
+      "step": 5835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9732882502113274e-05,
+      "loss": 0.5821,
+      "step": 5836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.973626373626374e-05,
+      "loss": 0.623,
+      "step": 5837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9739644970414203e-05,
+      "loss": 0.5493,
+      "step": 5838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.974302620456467e-05,
+      "loss": 0.5746,
+      "step": 5839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9746407438715132e-05,
+      "loss": 0.5778,
+      "step": 5840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9749788672865598e-05,
+      "loss": 0.6146,
+      "step": 5841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9753169907016064e-05,
+      "loss": 0.6301,
+      "step": 5842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9756551141166527e-05,
+      "loss": 0.7042,
+      "step": 5843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9759932375316993e-05,
+      "loss": 0.6149,
+      "step": 5844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.976331360946746e-05,
+      "loss": 0.5473,
+      "step": 5845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9766694843617922e-05,
+      "loss": 0.5079,
+      "step": 5846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.977007607776839e-05,
+      "loss": 0.5317,
+      "step": 5847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.977345731191885e-05,
+      "loss": 0.5249,
+      "step": 5848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9776838546069318e-05,
+      "loss": 0.6514,
+      "step": 5849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9780219780219784e-05,
+      "loss": 0.6294,
+      "step": 5850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9783601014370247e-05,
+      "loss": 0.6959,
+      "step": 5851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9786982248520713e-05,
+      "loss": 0.7066,
+      "step": 5852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.979036348267118e-05,
+      "loss": 0.5828,
+      "step": 5853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9793744716821642e-05,
+      "loss": 0.6898,
+      "step": 5854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.979712595097211e-05,
+      "loss": 0.6847,
+      "step": 5855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980050718512257e-05,
+      "loss": 0.6351,
+      "step": 5856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803888419273038e-05,
+      "loss": 0.6152,
+      "step": 5857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.98072696534235e-05,
+      "loss": 0.5929,
+      "step": 5858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9810650887573967e-05,
+      "loss": 0.7009,
+      "step": 5859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9814032121724433e-05,
+      "loss": 0.5555,
+      "step": 5860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9817413355874896e-05,
+      "loss": 0.6325,
+      "step": 5861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9820794590025362e-05,
+      "loss": 0.5925,
+      "step": 5862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9824175824175825e-05,
+      "loss": 0.6658,
+      "step": 5863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.982755705832629e-05,
+      "loss": 0.7091,
+      "step": 5864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9830938292476757e-05,
+      "loss": 0.5992,
+      "step": 5865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.983431952662722e-05,
+      "loss": 0.5531,
+      "step": 5866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9837700760777686e-05,
+      "loss": 0.5846,
+      "step": 5867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.984108199492815e-05,
+      "loss": 0.5849,
+      "step": 5868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9844463229078615e-05,
+      "loss": 0.569,
+      "step": 5869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9847844463229082e-05,
+      "loss": 0.596,
+      "step": 5870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9851225697379545e-05,
+      "loss": 0.5165,
+      "step": 5871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.985460693153001e-05,
+      "loss": 0.6267,
+      "step": 5872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9857988165680474e-05,
+      "loss": 0.7352,
+      "step": 5873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.986136939983094e-05,
+      "loss": 0.5599,
+      "step": 5874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9864750633981403e-05,
+      "loss": 0.642,
+      "step": 5875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.986813186813187e-05,
+      "loss": 0.5537,
+      "step": 5876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9871513102282335e-05,
+      "loss": 0.6997,
+      "step": 5877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9874894336432798e-05,
+      "loss": 0.5464,
+      "step": 5878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9878275570583264e-05,
+      "loss": 0.5788,
+      "step": 5879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9881656804733727e-05,
+      "loss": 0.5334,
+      "step": 5880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9885038038884193e-05,
+      "loss": 0.681,
+      "step": 5881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.988841927303466e-05,
+      "loss": 0.6033,
+      "step": 5882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9891800507185123e-05,
+      "loss": 0.6231,
+      "step": 5883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.989518174133559e-05,
+      "loss": 0.6502,
+      "step": 5884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.989856297548605e-05,
+      "loss": 0.5868,
+      "step": 5885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9901944209636518e-05,
+      "loss": 0.6866,
+      "step": 5886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9905325443786984e-05,
+      "loss": 0.5765,
+      "step": 5887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9908706677937447e-05,
+      "loss": 0.5693,
+      "step": 5888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9912087912087913e-05,
+      "loss": 0.629,
+      "step": 5889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9915469146238376e-05,
+      "loss": 0.6597,
+      "step": 5890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9918850380388842e-05,
+      "loss": 0.6194,
+      "step": 5891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.992223161453931e-05,
+      "loss": 0.6587,
+      "step": 5892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.992561284868977e-05,
+      "loss": 0.5848,
+      "step": 5893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9928994082840238e-05,
+      "loss": 0.6215,
+      "step": 5894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9932375316990704e-05,
+      "loss": 0.594,
+      "step": 5895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9935756551141167e-05,
+      "loss": 0.6896,
+      "step": 5896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9939137785291633e-05,
+      "loss": 0.7111,
+      "step": 5897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9942519019442096e-05,
+      "loss": 0.5526,
+      "step": 5898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9945900253592562e-05,
+      "loss": 0.6327,
+      "step": 5899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.994928148774303e-05,
+      "loss": 0.6452,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.995266272189349e-05,
+      "loss": 0.5759,
+      "step": 5901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9956043956043957e-05,
+      "loss": 0.658,
+      "step": 5902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9959425190194424e-05,
+      "loss": 0.6217,
+      "step": 5903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9962806424344887e-05,
+      "loss": 0.5576,
+      "step": 5904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9966187658495353e-05,
+      "loss": 0.6233,
+      "step": 5905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9969568892645816e-05,
+      "loss": 0.5364,
+      "step": 5906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9972950126796282e-05,
+      "loss": 0.5555,
+      "step": 5907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9976331360946748e-05,
+      "loss": 0.6102,
+      "step": 5908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997971259509721e-05,
+      "loss": 0.6549,
+      "step": 5909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9983093829247677e-05,
+      "loss": 0.5519,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9986475063398143e-05,
+      "loss": 0.5454,
+      "step": 5911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9989856297548606e-05,
+      "loss": 0.5907,
+      "step": 5912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9993237531699073e-05,
+      "loss": 0.7273,
+      "step": 5913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9996618765849535e-05,
+      "loss": 0.6346,
+      "step": 5914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.5838,
+      "step": 5915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999998650478e-05,
+      "loss": 0.5954,
+      "step": 5916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999994601908e-05,
+      "loss": 0.6416,
+      "step": 5917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999999998785429e-05,
+      "loss": 0.6202,
+      "step": 5918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999999997840763e-05,
+      "loss": 0.6563,
+      "step": 5919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999966261917e-05,
+      "loss": 0.6596,
+      "step": 5920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999951417163e-05,
+      "loss": 0.6002,
+      "step": 5921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999933873356e-05,
+      "loss": 0.5981,
+      "step": 5922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999913630505e-05,
+      "loss": 0.5382,
+      "step": 5923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999890688607e-05,
+      "loss": 0.5957,
+      "step": 5924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999865047663e-05,
+      "loss": 0.5962,
+      "step": 5925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999836707675e-05,
+      "loss": 0.6507,
+      "step": 5926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999805668638e-05,
+      "loss": 0.606,
+      "step": 5927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999771930553e-05,
+      "loss": 0.6218,
+      "step": 5928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999999973549342e-05,
+      "loss": 0.6253,
+      "step": 5929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999696357243e-05,
+      "loss": 0.6594,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999999965452202e-05,
+      "loss": 0.5593,
+      "step": 5931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999609987747e-05,
+      "loss": 0.6142,
+      "step": 5932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999562754432e-05,
+      "loss": 0.6532,
+      "step": 5933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999512822067e-05,
+      "loss": 0.6252,
+      "step": 5934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999460190658e-05,
+      "loss": 0.5314,
+      "step": 5935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.99999994048602e-05,
+      "loss": 0.608,
+      "step": 5936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999346830697e-05,
+      "loss": 0.6736,
+      "step": 5937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999286102145e-05,
+      "loss": 0.6666,
+      "step": 5938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999999922267455e-05,
+      "loss": 0.5395,
+      "step": 5939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999156547906e-05,
+      "loss": 0.7583,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999087722214e-05,
+      "loss": 0.5384,
+      "step": 5941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999999016197478e-05,
+      "loss": 0.5633,
+      "step": 5942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999998941973695e-05,
+      "loss": 0.7455,
+      "step": 5943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999998865050866e-05,
+      "loss": 0.6955,
+      "step": 5944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999998785428993e-05,
+      "loss": 0.6049,
+      "step": 5945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999999870310807e-05,
+      "loss": 0.5656,
+      "step": 5946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.99999986180881e-05,
+      "loss": 0.6355,
+      "step": 5947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999998530369083e-05,
+      "loss": 0.6604,
+      "step": 5948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999998439951023e-05,
+      "loss": 0.6421,
+      "step": 5949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999998346833916e-05,
+      "loss": 0.4999,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999998251017763e-05,
+      "loss": 0.7225,
+      "step": 5951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999998152502562e-05,
+      "loss": 0.5884,
+      "step": 5952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999998051288315e-05,
+      "loss": 0.7103,
+      "step": 5953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999999794737502e-05,
+      "loss": 0.5571,
+      "step": 5954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999997840762684e-05,
+      "loss": 0.6313,
+      "step": 5955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999997731451297e-05,
+      "loss": 0.6095,
+      "step": 5956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999997619440866e-05,
+      "loss": 0.7126,
+      "step": 5957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999999750473139e-05,
+      "loss": 0.5544,
+      "step": 5958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999997387322864e-05,
+      "loss": 0.6648,
+      "step": 5959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999997267215297e-05,
+      "loss": 0.6725,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999997144408682e-05,
+      "loss": 0.5548,
+      "step": 5961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999997018903018e-05,
+      "loss": 0.6724,
+      "step": 5962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9999996890698313e-05,
+      "loss": 0.5757,
+      "step": 5963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999999675979456e-05,
+      "loss": 0.5625,
+      "step": 5964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999999662619176e-05,
+      "loss": 0.5767,
+      "step": 5965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999996489889915e-05,
+      "loss": 0.633,
+      "step": 5966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999996350889023e-05,
+      "loss": 0.6183,
+      "step": 5967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999999620918909e-05,
+      "loss": 0.5793,
+      "step": 5968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999996064790106e-05,
+      "loss": 0.6498,
+      "step": 5969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999995917692078e-05,
+      "loss": 0.5464,
+      "step": 5970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999995767895002e-05,
+      "loss": 0.5797,
+      "step": 5971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999995615398884e-05,
+      "loss": 0.5521,
+      "step": 5972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999999546020372e-05,
+      "loss": 0.625,
+      "step": 5973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999999530230951e-05,
+      "loss": 0.5173,
+      "step": 5974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999995141716254e-05,
+      "loss": 0.5671,
+      "step": 5975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999994978423955e-05,
+      "loss": 0.663,
+      "step": 5976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999994812432605e-05,
+      "loss": 0.5621,
+      "step": 5977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999994643742213e-05,
+      "loss": 0.5881,
+      "step": 5978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999994472352777e-05,
+      "loss": 0.6287,
+      "step": 5979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999994298264297e-05,
+      "loss": 0.5669,
+      "step": 5980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999994121476768e-05,
+      "loss": 0.5474,
+      "step": 5981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999993941990195e-05,
+      "loss": 0.5862,
+      "step": 5982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999993759804575e-05,
+      "loss": 0.6117,
+      "step": 5983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999993574919912e-05,
+      "loss": 0.6585,
+      "step": 5984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999993387336206e-05,
+      "loss": 0.6149,
+      "step": 5985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999993197053453e-05,
+      "loss": 0.5135,
+      "step": 5986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999993004071653e-05,
+      "loss": 0.6535,
+      "step": 5987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999992808390813e-05,
+      "loss": 0.5303,
+      "step": 5988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999992610010923e-05,
+      "loss": 0.6187,
+      "step": 5989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999992408931992e-05,
+      "loss": 0.6394,
+      "step": 5990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999992205154016e-05,
+      "loss": 0.5625,
+      "step": 5991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999991998676992e-05,
+      "loss": 0.6165,
+      "step": 5992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999991789500925e-05,
+      "loss": 0.6633,
+      "step": 5993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999991577625818e-05,
+      "loss": 0.5592,
+      "step": 5994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999999136305166e-05,
+      "loss": 0.7032,
+      "step": 5995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999999114577846e-05,
+      "loss": 0.5865,
+      "step": 5996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999999092580622e-05,
+      "loss": 0.5747,
+      "step": 5997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999999070313493e-05,
+      "loss": 0.5442,
+      "step": 5998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999990477764598e-05,
+      "loss": 0.5905,
+      "step": 5999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999999024969522e-05,
+      "loss": 0.61,
+      "step": 6000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.99999900189268e-05,
+      "loss": 0.6034,
+      "step": 6001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999989785459335e-05,
+      "loss": 0.625,
+      "step": 6002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999989549292828e-05,
+      "loss": 0.5862,
+      "step": 6003
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999989310427273e-05,
+      "loss": 0.5108,
+      "step": 6004
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998906886268e-05,
+      "loss": 0.6061,
+      "step": 6005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999988824599038e-05,
+      "loss": 0.6181,
+      "step": 6006
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998857763635e-05,
+      "loss": 0.6024,
+      "step": 6007
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999988327974625e-05,
+      "loss": 0.536,
+      "step": 6008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998807561385e-05,
+      "loss": 0.5813,
+      "step": 6009
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998782055404e-05,
+      "loss": 0.5755,
+      "step": 6010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998756279518e-05,
+      "loss": 0.6455,
+      "step": 6011
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999987302337276e-05,
+      "loss": 0.6673,
+      "step": 6012
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998703918033e-05,
+      "loss": 0.6411,
+      "step": 6013
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999986773324343e-05,
+      "loss": 0.5724,
+      "step": 6014
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999986504769312e-05,
+      "loss": 0.7288,
+      "step": 6015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999986233515237e-05,
+      "loss": 0.5902,
+      "step": 6016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998595956212e-05,
+      "loss": 0.6236,
+      "step": 6017
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998568290996e-05,
+      "loss": 0.5528,
+      "step": 6018
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999985403558755e-05,
+      "loss": 0.5931,
+      "step": 6019
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998512150851e-05,
+      "loss": 0.6457,
+      "step": 6020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998483675922e-05,
+      "loss": 0.5908,
+      "step": 6021
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998454931089e-05,
+      "loss": 0.5603,
+      "step": 6022
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999984259163514e-05,
+      "loss": 0.569,
+      "step": 6023
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.99999839663171e-05,
+      "loss": 0.5892,
+      "step": 6024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998367077164e-05,
+      "loss": 0.5328,
+      "step": 6025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999983372527135e-05,
+      "loss": 0.6032,
+      "step": 6026
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999983071583593e-05,
+      "loss": 0.5317,
+      "step": 6027
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999982767941008e-05,
+      "loss": 0.4956,
+      "step": 6028
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999982461599382e-05,
+      "loss": 0.5162,
+      "step": 6029
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999982152558707e-05,
+      "loss": 0.689,
+      "step": 6030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999981840818998e-05,
+      "loss": 0.5798,
+      "step": 6031
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999981526380242e-05,
+      "loss": 0.5671,
+      "step": 6032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999981209242447e-05,
+      "loss": 0.5658,
+      "step": 6033
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998088940561e-05,
+      "loss": 0.5735,
+      "step": 6034
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999980566869733e-05,
+      "loss": 0.5627,
+      "step": 6035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999998024163481e-05,
+      "loss": 0.5544,
+      "step": 6036
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999979913700848e-05,
+      "loss": 0.5698,
+      "step": 6037
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999997958306785e-05,
+      "loss": 0.5687,
+      "step": 6038
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.99999792497358e-05,
+      "loss": 0.6496,
+      "step": 6039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999978913704718e-05,
+      "loss": 0.5315,
+      "step": 6040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999997857497459e-05,
+      "loss": 0.5213,
+      "step": 6041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999978233545423e-05,
+      "loss": 0.6118,
+      "step": 6042
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999977889417214e-05,
+      "loss": 0.5988,
+      "step": 6043
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999977542589965e-05,
+      "loss": 0.6382,
+      "step": 6044
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999977193063676e-05,
+      "loss": 0.5962,
+      "step": 6045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999976840838347e-05,
+      "loss": 0.5236,
+      "step": 6046
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999976485913975e-05,
+      "loss": 0.6124,
+      "step": 6047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999976128290563e-05,
+      "loss": 0.5638,
+      "step": 6048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999997576796811e-05,
+      "loss": 0.569,
+      "step": 6049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999997540494662e-05,
+      "loss": 0.5725,
+      "step": 6050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999975039226086e-05,
+      "loss": 0.6386,
+      "step": 6051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999974670806517e-05,
+      "loss": 0.6256,
+      "step": 6052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999974299687904e-05,
+      "loss": 0.5986,
+      "step": 6053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999973925870252e-05,
+      "loss": 0.6427,
+      "step": 6054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999973549353563e-05,
+      "loss": 0.6134,
+      "step": 6055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999997317013783e-05,
+      "loss": 0.5943,
+      "step": 6056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999997278822306e-05,
+      "loss": 0.5796,
+      "step": 6057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999972403609253e-05,
+      "loss": 0.5756,
+      "step": 6058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.99999720162964e-05,
+      "loss": 0.605,
+      "step": 6059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999971626284512e-05,
+      "loss": 0.6104,
+      "step": 6060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9999971233573586e-05,
+      "loss": 0.7779,
+      "step": 6061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999997083816362e-05,
+      "loss": 0.6632,
+      "step": 6062
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999970440054612e-05,
+      "loss": 0.5847,
+      "step": 6063
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999997003924657e-05,
+      "loss": 0.6041,
+      "step": 6064
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999969635739488e-05,
+      "loss": 0.6232,
+      "step": 6065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999969229533366e-05,
+      "loss": 0.6918,
+      "step": 6066
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999968820628207e-05,
+      "loss": 0.5294,
+      "step": 6067
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999968409024005e-05,
+      "loss": 0.6763,
+      "step": 6068
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999996799472077e-05,
+      "loss": 0.6638,
+      "step": 6069
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999967577718498e-05,
+      "loss": 0.5078,
+      "step": 6070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999967158017183e-05,
+      "loss": 0.6942,
+      "step": 6071
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999966735616834e-05,
+      "loss": 0.6093,
+      "step": 6072
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999966310517446e-05,
+      "loss": 0.6388,
+      "step": 6073
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999996588271902e-05,
+      "loss": 0.6871,
+      "step": 6074
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999996545222156e-05,
+      "loss": 0.6538,
+      "step": 6075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999965019025057e-05,
+      "loss": 0.6581,
+      "step": 6076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999964583129522e-05,
+      "loss": 0.5613,
+      "step": 6077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999964144534944e-05,
+      "loss": 0.6186,
+      "step": 6078
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999963703241332e-05,
+      "loss": 0.5339,
+      "step": 6079
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999963259248684e-05,
+      "loss": 0.5397,
+      "step": 6080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999962812556996e-05,
+      "loss": 0.7201,
+      "step": 6081
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999962363166275e-05,
+      "loss": 0.5475,
+      "step": 6082
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999961911076514e-05,
+      "loss": 0.6565,
+      "step": 6083
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999996145628772e-05,
+      "loss": 0.5526,
+      "step": 6084
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999960998799888e-05,
+      "loss": 0.6527,
+      "step": 6085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999996053861302e-05,
+      "loss": 0.5174,
+      "step": 6086
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999960075727113e-05,
+      "loss": 0.6741,
+      "step": 6087
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999959610142175e-05,
+      "loss": 0.6765,
+      "step": 6088
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999959141858198e-05,
+      "loss": 0.5042,
+      "step": 6089
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999958670875187e-05,
+      "loss": 0.6038,
+      "step": 6090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999958197193143e-05,
+      "loss": 0.5711,
+      "step": 6091
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999957720812055e-05,
+      "loss": 0.7432,
+      "step": 6092
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999957241731938e-05,
+      "loss": 0.5733,
+      "step": 6093
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999956759952784e-05,
+      "loss": 0.687,
+      "step": 6094
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999956275474597e-05,
+      "loss": 0.5029,
+      "step": 6095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999955788297374e-05,
+      "loss": 0.6336,
+      "step": 6096
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999955298421114e-05,
+      "loss": 0.6028,
+      "step": 6097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999954805845824e-05,
+      "loss": 0.5215,
+      "step": 6098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999954310571494e-05,
+      "loss": 0.6059,
+      "step": 6099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999953812598134e-05,
+      "loss": 0.5009,
+      "step": 6100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999953311925737e-05,
+      "loss": 0.5447,
+      "step": 6101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999952808554308e-05,
+      "loss": 0.6683,
+      "step": 6102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999995230248384e-05,
+      "loss": 0.6136,
+      "step": 6103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999951793714345e-05,
+      "loss": 0.74,
+      "step": 6104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999951282245812e-05,
+      "loss": 0.5611,
+      "step": 6105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999950768078243e-05,
+      "loss": 0.58,
+      "step": 6106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999950251211647e-05,
+      "loss": 0.5837,
+      "step": 6107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999949731646015e-05,
+      "loss": 0.558,
+      "step": 6108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999994920938135e-05,
+      "loss": 0.6585,
+      "step": 6109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999994868441765e-05,
+      "loss": 0.5855,
+      "step": 6110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999994815675492e-05,
+      "loss": 0.5623,
+      "step": 6111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999947626393153e-05,
+      "loss": 0.5136,
+      "step": 6112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999947093332358e-05,
+      "loss": 0.5518,
+      "step": 6113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999994655757253e-05,
+      "loss": 0.631,
+      "step": 6114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999994601911367e-05,
+      "loss": 0.5985,
+      "step": 6115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999945477955774e-05,
+      "loss": 0.6036,
+      "step": 6116
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999944934098846e-05,
+      "loss": 0.5641,
+      "step": 6117
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999994438754289e-05,
+      "loss": 0.6393,
+      "step": 6118
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.99999438382879e-05,
+      "loss": 0.5762,
+      "step": 6119
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999994328633388e-05,
+      "loss": 0.6067,
+      "step": 6120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999994273168083e-05,
+      "loss": 0.5796,
+      "step": 6121
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999942174328745e-05,
+      "loss": 0.5586,
+      "step": 6122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999941614277628e-05,
+      "loss": 0.6151,
+      "step": 6123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999941051527484e-05,
+      "loss": 0.5243,
+      "step": 6124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999940486078308e-05,
+      "loss": 0.6194,
+      "step": 6125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.99999399179301e-05,
+      "loss": 0.569,
+      "step": 6126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999993934708286e-05,
+      "loss": 0.5664,
+      "step": 6127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999938773536595e-05,
+      "loss": 0.5981,
+      "step": 6128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999938197291296e-05,
+      "loss": 0.5291,
+      "step": 6129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999937618346966e-05,
+      "loss": 0.5672,
+      "step": 6130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999937036703607e-05,
+      "loss": 0.6411,
+      "step": 6131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999936452361218e-05,
+      "loss": 0.5861,
+      "step": 6132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999935865319803e-05,
+      "loss": 0.7035,
+      "step": 6133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999935275579354e-05,
+      "loss": 0.6268,
+      "step": 6134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999934683139876e-05,
+      "loss": 0.5198,
+      "step": 6135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999934088001374e-05,
+      "loss": 0.6068,
+      "step": 6136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999993349016384e-05,
+      "loss": 0.6306,
+      "step": 6137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999932889627275e-05,
+      "loss": 0.5412,
+      "step": 6138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999993228639168e-05,
+      "loss": 0.6511,
+      "step": 6139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999931680457064e-05,
+      "loss": 0.555,
+      "step": 6140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999931071823413e-05,
+      "loss": 0.6112,
+      "step": 6141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999930460490736e-05,
+      "loss": 0.5708,
+      "step": 6142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999929846459032e-05,
+      "loss": 0.5419,
+      "step": 6143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.99999292297283e-05,
+      "loss": 0.687,
+      "step": 6144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999992861029854e-05,
+      "loss": 0.5943,
+      "step": 6145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999927988169753e-05,
+      "loss": 0.5088,
+      "step": 6146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999927363341937e-05,
+      "loss": 0.6173,
+      "step": 6147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999992673581509e-05,
+      "loss": 0.5936,
+      "step": 6148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999926105589226e-05,
+      "loss": 0.569,
+      "step": 6149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999925472664327e-05,
+      "loss": 0.5467,
+      "step": 6150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999924837040405e-05,
+      "loss": 0.6586,
+      "step": 6151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999924198717457e-05,
+      "loss": 0.5818,
+      "step": 6152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999992355769548e-05,
+      "loss": 0.5596,
+      "step": 6153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999922913974478e-05,
+      "loss": 0.565,
+      "step": 6154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999992226755445e-05,
+      "loss": 0.6178,
+      "step": 6155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.99999216184354e-05,
+      "loss": 0.6545,
+      "step": 6156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999992096661732e-05,
+      "loss": 0.749,
+      "step": 6157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999920312100216e-05,
+      "loss": 0.6038,
+      "step": 6158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999919654884086e-05,
+      "loss": 0.5761,
+      "step": 6159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9999918994968933e-05,
+      "loss": 0.6275,
+      "step": 6160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999991833235475e-05,
+      "loss": 0.574,
+      "step": 6161
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999917667041547e-05,
+      "loss": 0.6373,
+      "step": 6162
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999991699902932e-05,
+      "loss": 0.5746,
+      "step": 6163
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999916328318066e-05,
+      "loss": 0.5813,
+      "step": 6164
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999915654907788e-05,
+      "loss": 0.5975,
+      "step": 6165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999914978798487e-05,
+      "loss": 0.6218,
+      "step": 6166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999914299990162e-05,
+      "loss": 0.635,
+      "step": 6167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999913618482814e-05,
+      "loss": 0.5856,
+      "step": 6168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999991293427644e-05,
+      "loss": 0.7044,
+      "step": 6169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999912247371047e-05,
+      "loss": 0.5522,
+      "step": 6170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999991155776663e-05,
+      "loss": 0.6496,
+      "step": 6171
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999910865463187e-05,
+      "loss": 0.4724,
+      "step": 6172
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999910170460724e-05,
+      "loss": 0.5789,
+      "step": 6173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999909472759238e-05,
+      "loss": 0.6986,
+      "step": 6174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999990877235873e-05,
+      "loss": 0.6211,
+      "step": 6175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.99999080692592e-05,
+      "loss": 0.5762,
+      "step": 6176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999907363460646e-05,
+      "loss": 0.5559,
+      "step": 6177
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999990665496307e-05,
+      "loss": 0.6134,
+      "step": 6178
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999905943766477e-05,
+      "loss": 0.6002,
+      "step": 6179
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999905229870856e-05,
+      "loss": 0.5969,
+      "step": 6180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999904513276222e-05,
+      "loss": 0.5947,
+      "step": 6181
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999903793982562e-05,
+      "loss": 0.6519,
+      "step": 6182
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999903071989882e-05,
+      "loss": 0.5809,
+      "step": 6183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999990234729818e-05,
+      "loss": 0.5807,
+      "step": 6184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999901619907457e-05,
+      "loss": 0.5911,
+      "step": 6185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999900889817714e-05,
+      "loss": 0.6046,
+      "step": 6186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999900157028956e-05,
+      "loss": 0.6167,
+      "step": 6187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999899421541174e-05,
+      "loss": 0.635,
+      "step": 6188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999898683354373e-05,
+      "loss": 0.5214,
+      "step": 6189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999897942468552e-05,
+      "loss": 0.6187,
+      "step": 6190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999897198883712e-05,
+      "loss": 0.5679,
+      "step": 6191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999896452599855e-05,
+      "loss": 0.6053,
+      "step": 6192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999989570361698e-05,
+      "loss": 0.5877,
+      "step": 6193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999894951935083e-05,
+      "loss": 0.5751,
+      "step": 6194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999894197554167e-05,
+      "loss": 0.6185,
+      "step": 6195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999893440474235e-05,
+      "loss": 0.5192,
+      "step": 6196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999892680695286e-05,
+      "loss": 0.5374,
+      "step": 6197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999989191821732e-05,
+      "loss": 0.7282,
+      "step": 6198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999989115304033e-05,
+      "loss": 0.6171,
+      "step": 6199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999890385164327e-05,
+      "loss": 0.6224,
+      "step": 6200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999988961458931e-05,
+      "loss": 0.6402,
+      "step": 6201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999988884131527e-05,
+      "loss": 0.612,
+      "step": 6202
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999888065342217e-05,
+      "loss": 0.6127,
+      "step": 6203
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999887286670145e-05,
+      "loss": 0.5881,
+      "step": 6204
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999988650529906e-05,
+      "loss": 0.5833,
+      "step": 6205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999885721228955e-05,
+      "loss": 0.5518,
+      "step": 6206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999988493445984e-05,
+      "loss": 0.5211,
+      "step": 6207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999884144991703e-05,
+      "loss": 0.5726,
+      "step": 6208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999883352824556e-05,
+      "loss": 0.5749,
+      "step": 6209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999988255795839e-05,
+      "loss": 0.5889,
+      "step": 6210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999881760393206e-05,
+      "loss": 0.5416,
+      "step": 6211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999880960129013e-05,
+      "loss": 0.5622,
+      "step": 6212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999880157165805e-05,
+      "loss": 0.6469,
+      "step": 6213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999987935150358e-05,
+      "loss": 0.6452,
+      "step": 6214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999987854314234e-05,
+      "loss": 0.5398,
+      "step": 6215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999877732082087e-05,
+      "loss": 0.5715,
+      "step": 6216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999876918322824e-05,
+      "loss": 0.6184,
+      "step": 6217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999876101864544e-05,
+      "loss": 0.6795,
+      "step": 6218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999987528270725e-05,
+      "loss": 0.647,
+      "step": 6219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999874460850942e-05,
+      "loss": 0.591,
+      "step": 6220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999873636295627e-05,
+      "loss": 0.5822,
+      "step": 6221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999872809041295e-05,
+      "loss": 0.5541,
+      "step": 6222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999987197908795e-05,
+      "loss": 0.6469,
+      "step": 6223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999871146435593e-05,
+      "loss": 0.6246,
+      "step": 6224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999870311084227e-05,
+      "loss": 0.5892,
+      "step": 6225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999869473033847e-05,
+      "loss": 0.5151,
+      "step": 6226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999868632284454e-05,
+      "loss": 0.6318,
+      "step": 6227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999867788836052e-05,
+      "loss": 0.6232,
+      "step": 6228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999986694268864e-05,
+      "loss": 0.6376,
+      "step": 6229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999866093842216e-05,
+      "loss": 0.5978,
+      "step": 6230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999986524229678e-05,
+      "loss": 0.5979,
+      "step": 6231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999864388052335e-05,
+      "loss": 0.5624,
+      "step": 6232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999986353110888e-05,
+      "loss": 0.5408,
+      "step": 6233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999862671466413e-05,
+      "loss": 0.5864,
+      "step": 6234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999861809124938e-05,
+      "loss": 0.6169,
+      "step": 6235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999860944084456e-05,
+      "loss": 0.5545,
+      "step": 6236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999860076344962e-05,
+      "loss": 0.7089,
+      "step": 6237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999859205906458e-05,
+      "loss": 0.618,
+      "step": 6238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999858332768945e-05,
+      "loss": 0.6869,
+      "step": 6239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999857456932425e-05,
+      "loss": 0.5533,
+      "step": 6240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999856578396896e-05,
+      "loss": 0.5889,
+      "step": 6241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999985569716236e-05,
+      "loss": 0.5605,
+      "step": 6242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999854813228817e-05,
+      "loss": 0.6795,
+      "step": 6243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999853926596266e-05,
+      "loss": 0.57,
+      "step": 6244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999853037264706e-05,
+      "loss": 0.6595,
+      "step": 6245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999985214523414e-05,
+      "loss": 0.6353,
+      "step": 6246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999851250504567e-05,
+      "loss": 0.6474,
+      "step": 6247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999985035307599e-05,
+      "loss": 0.6573,
+      "step": 6248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.99998494529484e-05,
+      "loss": 0.6456,
+      "step": 6249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999984855012181e-05,
+      "loss": 0.6045,
+      "step": 6250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999984764459621e-05,
+      "loss": 0.6246,
+      "step": 6251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999984673637161e-05,
+      "loss": 0.6462,
+      "step": 6252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999845825447997e-05,
+      "loss": 0.6203,
+      "step": 6253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999844911825386e-05,
+      "loss": 0.5193,
+      "step": 6254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999843995503764e-05,
+      "loss": 0.5659,
+      "step": 6255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999843076483144e-05,
+      "loss": 0.6946,
+      "step": 6256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999842154763517e-05,
+      "loss": 0.687,
+      "step": 6257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9999841230344884e-05,
+      "loss": 0.5754,
+      "step": 6258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999984030322725e-05,
+      "loss": 0.7501,
+      "step": 6259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999839373410607e-05,
+      "loss": 0.6934,
+      "step": 6260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999838440894966e-05,
+      "loss": 0.7289,
+      "step": 6261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999983750568032e-05,
+      "loss": 0.5532,
+      "step": 6262
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999983656776667e-05,
+      "loss": 0.5664,
+      "step": 6263
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999983562715402e-05,
+      "loss": 0.5712,
+      "step": 6264
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999834683842364e-05,
+      "loss": 0.6844,
+      "step": 6265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999833737831706e-05,
+      "loss": 0.6498,
+      "step": 6266
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999983278912205e-05,
+      "loss": 0.5777,
+      "step": 6267
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999831837713392e-05,
+      "loss": 0.5362,
+      "step": 6268
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999983088360573e-05,
+      "loss": 0.5528,
+      "step": 6269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999829926799067e-05,
+      "loss": 0.604,
+      "step": 6270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999828967293406e-05,
+      "loss": 0.5879,
+      "step": 6271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999982800508874e-05,
+      "loss": 0.6946,
+      "step": 6272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999827040185078e-05,
+      "loss": 0.6221,
+      "step": 6273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999826072582415e-05,
+      "loss": 0.5009,
+      "step": 6274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999982510228075e-05,
+      "loss": 0.6235,
+      "step": 6275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999824129280084e-05,
+      "loss": 0.6062,
+      "step": 6276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999823153580423e-05,
+      "loss": 0.5452,
+      "step": 6277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999822175181763e-05,
+      "loss": 0.6623,
+      "step": 6278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.99998211940841e-05,
+      "loss": 0.5373,
+      "step": 6279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999820210287445e-05,
+      "loss": 0.5819,
+      "step": 6280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999819223791786e-05,
+      "loss": 0.6557,
+      "step": 6281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999981823459713e-05,
+      "loss": 0.5222,
+      "step": 6282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999981724270348e-05,
+      "loss": 0.7888,
+      "step": 6283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999981624811083e-05,
+      "loss": 0.6452,
+      "step": 6284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999981525081918e-05,
+      "loss": 0.6718,
+      "step": 6285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999814250828536e-05,
+      "loss": 0.4974,
+      "step": 6286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999813248138898e-05,
+      "loss": 0.5682,
+      "step": 6287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999812242750258e-05,
+      "loss": 0.57,
+      "step": 6288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999811234662626e-05,
+      "loss": 0.7386,
+      "step": 6289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999810223875997e-05,
+      "loss": 0.5723,
+      "step": 6290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999809210390373e-05,
+      "loss": 0.5841,
+      "step": 6291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999808194205756e-05,
+      "loss": 0.6181,
+      "step": 6292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999980717532214e-05,
+      "loss": 0.7004,
+      "step": 6293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999980615373953e-05,
+      "loss": 0.6137,
+      "step": 6294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999805129457925e-05,
+      "loss": 0.6731,
+      "step": 6295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999804102477328e-05,
+      "loss": 0.6301,
+      "step": 6296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999803072797738e-05,
+      "loss": 0.6353,
+      "step": 6297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999980204041915e-05,
+      "loss": 0.6504,
+      "step": 6298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999980100534157e-05,
+      "loss": 0.5376,
+      "step": 6299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999799967565e-05,
+      "loss": 0.5855,
+      "step": 6300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999798927089435e-05,
+      "loss": 0.5384,
+      "step": 6301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999979788391488e-05,
+      "loss": 0.7025,
+      "step": 6302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999979683804133e-05,
+      "loss": 0.6923,
+      "step": 6303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999979578946879e-05,
+      "loss": 0.577,
+      "step": 6304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999794738197255e-05,
+      "loss": 0.6327,
+      "step": 6305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999793684226732e-05,
+      "loss": 0.6025,
+      "step": 6306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999792627557216e-05,
+      "loss": 0.6989,
+      "step": 6307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999979156818871e-05,
+      "loss": 0.532,
+      "step": 6308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999790506121214e-05,
+      "loss": 0.6344,
+      "step": 6309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999789441354728e-05,
+      "loss": 0.5731,
+      "step": 6310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999788373889252e-05,
+      "loss": 0.6062,
+      "step": 6311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999787303724787e-05,
+      "loss": 0.5643,
+      "step": 6312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999978623086133e-05,
+      "loss": 0.6236,
+      "step": 6313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999785155298887e-05,
+      "loss": 0.6429,
+      "step": 6314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999978407703745e-05,
+      "loss": 0.6734,
+      "step": 6315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999978299607703e-05,
+      "loss": 0.634,
+      "step": 6316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999781912417622e-05,
+      "loss": 0.6639,
+      "step": 6317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999780826059222e-05,
+      "loss": 0.6742,
+      "step": 6318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999779737001836e-05,
+      "loss": 0.6566,
+      "step": 6319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999977864524547e-05,
+      "loss": 0.7329,
+      "step": 6320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999777550790104e-05,
+      "loss": 0.5846,
+      "step": 6321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999977645363576e-05,
+      "loss": 0.5322,
+      "step": 6322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999775353782426e-05,
+      "loss": 0.5558,
+      "step": 6323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999774251230108e-05,
+      "loss": 0.6731,
+      "step": 6324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999773145978804e-05,
+      "loss": 0.6192,
+      "step": 6325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999772038028514e-05,
+      "loss": 0.7027,
+      "step": 6326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999770927379236e-05,
+      "loss": 0.5678,
+      "step": 6327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999769814030978e-05,
+      "loss": 0.6843,
+      "step": 6328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999976869798373e-05,
+      "loss": 0.5589,
+      "step": 6329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999767579237502e-05,
+      "loss": 0.4963,
+      "step": 6330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999976645779229e-05,
+      "loss": 0.5871,
+      "step": 6331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999765333648093e-05,
+      "loss": 0.526,
+      "step": 6332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999764206804913e-05,
+      "loss": 0.4884,
+      "step": 6333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999976307726275e-05,
+      "loss": 0.6057,
+      "step": 6334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999761945021603e-05,
+      "loss": 0.5664,
+      "step": 6335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999760810081472e-05,
+      "loss": 0.5912,
+      "step": 6336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999759672442363e-05,
+      "loss": 0.5259,
+      "step": 6337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999975853210427e-05,
+      "loss": 0.6625,
+      "step": 6338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999757389067197e-05,
+      "loss": 0.7018,
+      "step": 6339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999975624333114e-05,
+      "loss": 0.5427,
+      "step": 6340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999755094896104e-05,
+      "loss": 0.5922,
+      "step": 6341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999753943762086e-05,
+      "loss": 0.5406,
+      "step": 6342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999975278992909e-05,
+      "loss": 0.5262,
+      "step": 6343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999975163339711e-05,
+      "loss": 0.5974,
+      "step": 6344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999975047416615e-05,
+      "loss": 0.6048,
+      "step": 6345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999749312236217e-05,
+      "loss": 0.6374,
+      "step": 6346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.99997481476073e-05,
+      "loss": 0.6084,
+      "step": 6347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999746980279406e-05,
+      "loss": 0.5921,
+      "step": 6348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999745810252534e-05,
+      "loss": 0.538,
+      "step": 6349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999744637526684e-05,
+      "loss": 0.6191,
+      "step": 6350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999743462101852e-05,
+      "loss": 0.5696,
+      "step": 6351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999742283978047e-05,
+      "loss": 0.5301,
+      "step": 6352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999974110315526e-05,
+      "loss": 0.5412,
+      "step": 6353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.99997399196335e-05,
+      "loss": 0.6212,
+      "step": 6354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999738733412766e-05,
+      "loss": 0.566,
+      "step": 6355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999737544493052e-05,
+      "loss": 0.6337,
+      "step": 6356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9999736352874362e-05,
+      "loss": 0.5565,
+      "step": 6357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999735158556697e-05,
+      "loss": 0.7428,
+      "step": 6358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999733961540056e-05,
+      "loss": 0.6116,
+      "step": 6359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999732761824443e-05,
+      "loss": 0.5793,
+      "step": 6360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999731559409855e-05,
+      "loss": 0.6411,
+      "step": 6361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999973035429629e-05,
+      "loss": 0.5236,
+      "step": 6362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999972914648375e-05,
+      "loss": 0.5028,
+      "step": 6363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999972793597224e-05,
+      "loss": 0.5735,
+      "step": 6364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999726722761754e-05,
+      "loss": 0.6703,
+      "step": 6365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999725506852297e-05,
+      "loss": 0.6468,
+      "step": 6366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999972428824387e-05,
+      "loss": 0.5948,
+      "step": 6367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999723066936464e-05,
+      "loss": 0.6409,
+      "step": 6368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999721842930087e-05,
+      "loss": 0.5496,
+      "step": 6369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999720616224742e-05,
+      "loss": 0.5426,
+      "step": 6370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999719386820424e-05,
+      "loss": 0.4933,
+      "step": 6371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999718154717138e-05,
+      "loss": 0.631,
+      "step": 6372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999971691991488e-05,
+      "loss": 0.5535,
+      "step": 6373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999715682413648e-05,
+      "loss": 0.5652,
+      "step": 6374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999714442213448e-05,
+      "loss": 0.64,
+      "step": 6375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999713199314276e-05,
+      "loss": 0.5412,
+      "step": 6376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999711953716138e-05,
+      "loss": 0.646,
+      "step": 6377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999710705419032e-05,
+      "loss": 0.6644,
+      "step": 6378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999709454422953e-05,
+      "loss": 0.5807,
+      "step": 6379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999970820072791e-05,
+      "loss": 0.531,
+      "step": 6380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999706944333893e-05,
+      "loss": 0.5716,
+      "step": 6381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999705685240914e-05,
+      "loss": 0.5664,
+      "step": 6382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999704423448967e-05,
+      "loss": 0.6929,
+      "step": 6383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999970315895805e-05,
+      "loss": 0.5571,
+      "step": 6384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999701891768173e-05,
+      "loss": 0.6067,
+      "step": 6385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999700621879322e-05,
+      "loss": 0.5757,
+      "step": 6386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999969934929151e-05,
+      "loss": 0.5521,
+      "step": 6387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999969807400473e-05,
+      "loss": 0.659,
+      "step": 6388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999696796018985e-05,
+      "loss": 0.5911,
+      "step": 6389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999695515334277e-05,
+      "loss": 0.5668,
+      "step": 6390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999694231950602e-05,
+      "loss": 0.5861,
+      "step": 6391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999692945867966e-05,
+      "loss": 0.5944,
+      "step": 6392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999691657086364e-05,
+      "loss": 0.5979,
+      "step": 6393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999690365605797e-05,
+      "loss": 0.6096,
+      "step": 6394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999689071426268e-05,
+      "loss": 0.5816,
+      "step": 6395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999687774547776e-05,
+      "loss": 0.4985,
+      "step": 6396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999686474970323e-05,
+      "loss": 0.5981,
+      "step": 6397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999685172693907e-05,
+      "loss": 0.6871,
+      "step": 6398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999968386771853e-05,
+      "loss": 0.6084,
+      "step": 6399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999968256004419e-05,
+      "loss": 0.5404,
+      "step": 6400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999968124967089e-05,
+      "loss": 0.5567,
+      "step": 6401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999679936598628e-05,
+      "loss": 0.6282,
+      "step": 6402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999967862082741e-05,
+      "loss": 0.6268,
+      "step": 6403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999677302357224e-05,
+      "loss": 0.5795,
+      "step": 6404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999967598118808e-05,
+      "loss": 0.6154,
+      "step": 6405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999674657319983e-05,
+      "loss": 0.5765,
+      "step": 6406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999673330752926e-05,
+      "loss": 0.5974,
+      "step": 6407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999672001486906e-05,
+      "loss": 0.5024,
+      "step": 6408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999670669521932e-05,
+      "loss": 0.5468,
+      "step": 6409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999669334857995e-05,
+      "loss": 0.5425,
+      "step": 6410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999667997495103e-05,
+      "loss": 0.747,
+      "step": 6411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999666657433256e-05,
+      "loss": 0.5921,
+      "step": 6412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999966531467245e-05,
+      "loss": 0.5704,
+      "step": 6413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999966396921269e-05,
+      "loss": 0.6263,
+      "step": 6414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999662621053973e-05,
+      "loss": 0.646,
+      "step": 6415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999661270196298e-05,
+      "loss": 0.5796,
+      "step": 6416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999659916639667e-05,
+      "loss": 0.5607,
+      "step": 6417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999658560384082e-05,
+      "loss": 0.6301,
+      "step": 6418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999657201429548e-05,
+      "loss": 0.7019,
+      "step": 6419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999655839776055e-05,
+      "loss": 0.5636,
+      "step": 6420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999654475423607e-05,
+      "loss": 0.6367,
+      "step": 6421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999653108372206e-05,
+      "loss": 0.6546,
+      "step": 6422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999651738621855e-05,
+      "loss": 0.7151,
+      "step": 6423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999650366172547e-05,
+      "loss": 0.6923,
+      "step": 6424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999648991024288e-05,
+      "loss": 0.5377,
+      "step": 6425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999964761317708e-05,
+      "loss": 0.5676,
+      "step": 6426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999646232630917e-05,
+      "loss": 0.5732,
+      "step": 6427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999644849385803e-05,
+      "loss": 0.5669,
+      "step": 6428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999643463441743e-05,
+      "loss": 0.5712,
+      "step": 6429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999642074798727e-05,
+      "loss": 0.5079,
+      "step": 6430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999640683456764e-05,
+      "loss": 0.5524,
+      "step": 6431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999639289415848e-05,
+      "loss": 0.6393,
+      "step": 6432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999637892675984e-05,
+      "loss": 0.5376,
+      "step": 6433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999963649323717e-05,
+      "loss": 0.5422,
+      "step": 6434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999963509109941e-05,
+      "loss": 0.5868,
+      "step": 6435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.99996336862627e-05,
+      "loss": 0.5534,
+      "step": 6436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999632278727042e-05,
+      "loss": 0.603,
+      "step": 6437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999963086849244e-05,
+      "loss": 0.5784,
+      "step": 6438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999629455558883e-05,
+      "loss": 0.6507,
+      "step": 6439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999628039926383e-05,
+      "loss": 0.6061,
+      "step": 6440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999962662159494e-05,
+      "loss": 0.5877,
+      "step": 6441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999962520056455e-05,
+      "loss": 0.5916,
+      "step": 6442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999962377683521e-05,
+      "loss": 0.6127,
+      "step": 6443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999962235040693e-05,
+      "loss": 0.5999,
+      "step": 6444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999620921279704e-05,
+      "loss": 0.6184,
+      "step": 6445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999619489453533e-05,
+      "loss": 0.6029,
+      "step": 6446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999961805492842e-05,
+      "loss": 0.607,
+      "step": 6447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999616617704358e-05,
+      "loss": 0.5955,
+      "step": 6448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999615177781355e-05,
+      "loss": 0.5572,
+      "step": 6449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999961373515941e-05,
+      "loss": 0.5993,
+      "step": 6450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999612289838524e-05,
+      "loss": 0.5715,
+      "step": 6451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999610841818692e-05,
+      "loss": 0.6665,
+      "step": 6452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999609391099922e-05,
+      "loss": 0.5966,
+      "step": 6453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999960793768221e-05,
+      "loss": 0.6152,
+      "step": 6454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9999606481565553e-05,
+      "loss": 0.6042,
+      "step": 6455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999960502274996e-05,
+      "loss": 0.5759,
+      "step": 6456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999603561235424e-05,
+      "loss": 0.5474,
+      "step": 6457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999602097021952e-05,
+      "loss": 0.6051,
+      "step": 6458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999600630109538e-05,
+      "loss": 0.6335,
+      "step": 6459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999599160498186e-05,
+      "loss": 0.5245,
+      "step": 6460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999597688187896e-05,
+      "loss": 0.5448,
+      "step": 6461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999596213178667e-05,
+      "loss": 0.6003,
+      "step": 6462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.99995947354705e-05,
+      "loss": 0.5921,
+      "step": 6463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999593255063394e-05,
+      "loss": 0.4946,
+      "step": 6464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999591771957357e-05,
+      "loss": 0.6139,
+      "step": 6465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999590286152378e-05,
+      "loss": 0.5058,
+      "step": 6466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999588797648465e-05,
+      "loss": 0.4648,
+      "step": 6467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999587306445612e-05,
+      "loss": 0.6566,
+      "step": 6468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999958581254383e-05,
+      "loss": 0.5775,
+      "step": 6469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999584315943107e-05,
+      "loss": 0.6044,
+      "step": 6470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999582816643453e-05,
+      "loss": 0.6699,
+      "step": 6471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999581314644864e-05,
+      "loss": 0.6006,
+      "step": 6472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999957980994734e-05,
+      "loss": 0.5384,
+      "step": 6473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999578302550886e-05,
+      "loss": 0.5448,
+      "step": 6474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.99995767924555e-05,
+      "loss": 0.6922,
+      "step": 6475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999575279661178e-05,
+      "loss": 0.4796,
+      "step": 6476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999573764167924e-05,
+      "loss": 0.5725,
+      "step": 6477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999957224597574e-05,
+      "loss": 0.6064,
+      "step": 6478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999570725084623e-05,
+      "loss": 0.5853,
+      "step": 6479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999569201494575e-05,
+      "loss": 0.5078,
+      "step": 6480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.99995676752056e-05,
+      "loss": 0.6199,
+      "step": 6481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999956614621769e-05,
+      "loss": 0.6276,
+      "step": 6482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999564614530852e-05,
+      "loss": 0.616,
+      "step": 6483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999563080145085e-05,
+      "loss": 0.5521,
+      "step": 6484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999956154306039e-05,
+      "loss": 0.6138,
+      "step": 6485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999956000327677e-05,
+      "loss": 0.6048,
+      "step": 6486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999558460794217e-05,
+      "loss": 0.6505,
+      "step": 6487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999556915612736e-05,
+      "loss": 0.5991,
+      "step": 6488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999955536773233e-05,
+      "loss": 0.5637,
+      "step": 6489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999553817152995e-05,
+      "loss": 0.6735,
+      "step": 6490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999552263874737e-05,
+      "loss": 0.5436,
+      "step": 6491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999955070789755e-05,
+      "loss": 0.5602,
+      "step": 6492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999954914922144e-05,
+      "loss": 0.5493,
+      "step": 6493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999547587846404e-05,
+      "loss": 0.5879,
+      "step": 6494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999546023772444e-05,
+      "loss": 0.586,
+      "step": 6495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999544456999558e-05,
+      "loss": 0.5689,
+      "step": 6496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999954288752775e-05,
+      "loss": 0.5419,
+      "step": 6497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999954131535702e-05,
+      "loss": 0.5766,
+      "step": 6498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999953974048736e-05,
+      "loss": 0.533,
+      "step": 6499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999538162918782e-05,
+      "loss": 0.6295,
+      "step": 6500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999536582651283e-05,
+      "loss": 0.5965,
+      "step": 6501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999534999684863e-05,
+      "loss": 0.527,
+      "step": 6502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999533414019524e-05,
+      "loss": 0.5814,
+      "step": 6503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999531825655257e-05,
+      "loss": 0.6444,
+      "step": 6504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999530234592076e-05,
+      "loss": 0.6242,
+      "step": 6505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999528640829973e-05,
+      "loss": 0.6605,
+      "step": 6506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999527044368945e-05,
+      "loss": 0.5343,
+      "step": 6507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999525445209003e-05,
+      "loss": 0.541,
+      "step": 6508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999523843350146e-05,
+      "loss": 0.5434,
+      "step": 6509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999522238792365e-05,
+      "loss": 0.6052,
+      "step": 6510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999952063153567e-05,
+      "loss": 0.4931,
+      "step": 6511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999519021580057e-05,
+      "loss": 0.4848,
+      "step": 6512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999517408925522e-05,
+      "loss": 0.5527,
+      "step": 6513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999515793572075e-05,
+      "loss": 0.7749,
+      "step": 6514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999514175519713e-05,
+      "loss": 0.6861,
+      "step": 6515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999512554768433e-05,
+      "loss": 0.6207,
+      "step": 6516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999951093131824e-05,
+      "loss": 0.5405,
+      "step": 6517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999509305169127e-05,
+      "loss": 0.6367,
+      "step": 6518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999507676321107e-05,
+      "loss": 0.5274,
+      "step": 6519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999950604477417e-05,
+      "loss": 0.5889,
+      "step": 6520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999504410528316e-05,
+      "loss": 0.6229,
+      "step": 6521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999502773583552e-05,
+      "loss": 0.5756,
+      "step": 6522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999501133939877e-05,
+      "loss": 0.6169,
+      "step": 6523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999499491597287e-05,
+      "loss": 0.683,
+      "step": 6524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999949784655579e-05,
+      "loss": 0.5384,
+      "step": 6525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999496198815376e-05,
+      "loss": 0.6302,
+      "step": 6526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999494548376056e-05,
+      "loss": 0.7276,
+      "step": 6527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999949289523782e-05,
+      "loss": 0.6061,
+      "step": 6528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999949123940068e-05,
+      "loss": 0.4949,
+      "step": 6529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999489580864625e-05,
+      "loss": 0.5963,
+      "step": 6530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999487919629666e-05,
+      "loss": 0.6078,
+      "step": 6531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999486255695796e-05,
+      "loss": 0.6946,
+      "step": 6532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999948458906302e-05,
+      "loss": 0.516,
+      "step": 6533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999482919731333e-05,
+      "loss": 0.5603,
+      "step": 6534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999481247700744e-05,
+      "loss": 0.5748,
+      "step": 6535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999479572971246e-05,
+      "loss": 0.5999,
+      "step": 6536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999477895542837e-05,
+      "loss": 0.5853,
+      "step": 6537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999476215415527e-05,
+      "loss": 0.6307,
+      "step": 6538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999474532589312e-05,
+      "loss": 0.6655,
+      "step": 6539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999947284706419e-05,
+      "loss": 0.4675,
+      "step": 6540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999471158840163e-05,
+      "loss": 0.6068,
+      "step": 6541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999469467917235e-05,
+      "loss": 0.5361,
+      "step": 6542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.99994677742954e-05,
+      "loss": 0.5836,
+      "step": 6543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999466077974662e-05,
+      "loss": 0.6351,
+      "step": 6544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999464378955028e-05,
+      "loss": 0.6121,
+      "step": 6545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999462677236482e-05,
+      "loss": 0.5835,
+      "step": 6546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999460972819042e-05,
+      "loss": 0.5669,
+      "step": 6547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999459265702697e-05,
+      "loss": 0.5692,
+      "step": 6548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999945755588745e-05,
+      "loss": 0.5959,
+      "step": 6549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999455843373305e-05,
+      "loss": 0.6382,
+      "step": 6550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999454128160264e-05,
+      "loss": 0.5771,
+      "step": 6551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999945241024832e-05,
+      "loss": 0.5765,
+      "step": 6552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9999450689637472e-05,
+      "loss": 0.567,
+      "step": 6553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999944896632773e-05,
+      "loss": 0.6043,
+      "step": 6554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999944724031909e-05,
+      "loss": 0.5321,
+      "step": 6555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999445511611553e-05,
+      "loss": 0.6028,
+      "step": 6556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999944378020512e-05,
+      "loss": 0.6027,
+      "step": 6557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999442046099787e-05,
+      "loss": 0.5931,
+      "step": 6558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999440309295558e-05,
+      "loss": 0.4799,
+      "step": 6559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999438569792437e-05,
+      "loss": 0.5856,
+      "step": 6560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999943682759042e-05,
+      "loss": 0.5656,
+      "step": 6561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999435082689504e-05,
+      "loss": 0.6832,
+      "step": 6562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999433335089697e-05,
+      "loss": 0.5241,
+      "step": 6563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999431584790996e-05,
+      "loss": 0.716,
+      "step": 6564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99994298317934e-05,
+      "loss": 0.5222,
+      "step": 6565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999428076096915e-05,
+      "loss": 0.5918,
+      "step": 6566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999426317701535e-05,
+      "loss": 0.72,
+      "step": 6567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999424556607263e-05,
+      "loss": 0.5989,
+      "step": 6568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99994227928141e-05,
+      "loss": 0.6174,
+      "step": 6569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999421026322045e-05,
+      "loss": 0.6123,
+      "step": 6570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99994192571311e-05,
+      "loss": 0.6227,
+      "step": 6571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999417485241266e-05,
+      "loss": 0.5995,
+      "step": 6572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999941571065254e-05,
+      "loss": 0.6297,
+      "step": 6573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999941393336493e-05,
+      "loss": 0.6045,
+      "step": 6574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999941215337843e-05,
+      "loss": 0.5513,
+      "step": 6575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999410370693037e-05,
+      "loss": 0.6135,
+      "step": 6576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999940858530876e-05,
+      "loss": 0.5318,
+      "step": 6577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999406797225594e-05,
+      "loss": 0.6068,
+      "step": 6578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999405006443546e-05,
+      "loss": 0.7016,
+      "step": 6579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999403212962608e-05,
+      "loss": 0.5292,
+      "step": 6580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999401416782785e-05,
+      "loss": 0.6025,
+      "step": 6581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999399617904075e-05,
+      "loss": 0.5743,
+      "step": 6582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999397816326484e-05,
+      "loss": 0.6366,
+      "step": 6583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999396012050005e-05,
+      "loss": 0.613,
+      "step": 6584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999394205074646e-05,
+      "loss": 0.7189,
+      "step": 6585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999392395400403e-05,
+      "loss": 0.5652,
+      "step": 6586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999390583027275e-05,
+      "loss": 0.4872,
+      "step": 6587
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999388767955267e-05,
+      "loss": 0.5508,
+      "step": 6588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999386950184374e-05,
+      "loss": 0.6308,
+      "step": 6589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999385129714604e-05,
+      "loss": 0.6253,
+      "step": 6590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999938330654595e-05,
+      "loss": 0.5534,
+      "step": 6591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999381480678416e-05,
+      "loss": 0.6368,
+      "step": 6592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999379652112004e-05,
+      "loss": 0.5053,
+      "step": 6593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999937782084671e-05,
+      "loss": 0.521,
+      "step": 6594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999937598688254e-05,
+      "loss": 0.7265,
+      "step": 6595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999937415021949e-05,
+      "loss": 0.4912,
+      "step": 6596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999372310857565e-05,
+      "loss": 0.6891,
+      "step": 6597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999937046879676e-05,
+      "loss": 0.5228,
+      "step": 6598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999936862403708e-05,
+      "loss": 0.6292,
+      "step": 6599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999366776578523e-05,
+      "loss": 0.631,
+      "step": 6600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999936492642109e-05,
+      "loss": 0.621,
+      "step": 6601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999363073564777e-05,
+      "loss": 0.641,
+      "step": 6602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999361218009598e-05,
+      "loss": 0.5953,
+      "step": 6603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999359359755538e-05,
+      "loss": 0.6433,
+      "step": 6604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999357498802604e-05,
+      "loss": 0.6429,
+      "step": 6605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99993556351508e-05,
+      "loss": 0.6147,
+      "step": 6606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999353768800122e-05,
+      "loss": 0.6482,
+      "step": 6607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999935189975057e-05,
+      "loss": 0.5416,
+      "step": 6608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999935002800215e-05,
+      "loss": 0.5826,
+      "step": 6609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999348153554857e-05,
+      "loss": 0.5607,
+      "step": 6610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999934627640869e-05,
+      "loss": 0.6182,
+      "step": 6611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999934439656366e-05,
+      "loss": 0.5255,
+      "step": 6612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999934251401975e-05,
+      "loss": 0.505,
+      "step": 6613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999934062877698e-05,
+      "loss": 0.5763,
+      "step": 6614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999933874083534e-05,
+      "loss": 0.5704,
+      "step": 6615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999336850194825e-05,
+      "loss": 0.6398,
+      "step": 6616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999334956855448e-05,
+      "loss": 0.5015,
+      "step": 6617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99993330608172e-05,
+      "loss": 0.6898,
+      "step": 6618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999331162080088e-05,
+      "loss": 0.5212,
+      "step": 6619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999932926064411e-05,
+      "loss": 0.7508,
+      "step": 6620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999327356509263e-05,
+      "loss": 0.5594,
+      "step": 6621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999325449675557e-05,
+      "loss": 0.5445,
+      "step": 6622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999323540142983e-05,
+      "loss": 0.5143,
+      "step": 6623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999321627911542e-05,
+      "loss": 0.5838,
+      "step": 6624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999931971298124e-05,
+      "loss": 0.5358,
+      "step": 6625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999317795352076e-05,
+      "loss": 0.5406,
+      "step": 6626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999315875024047e-05,
+      "loss": 0.5553,
+      "step": 6627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999931395199716e-05,
+      "loss": 0.553,
+      "step": 6628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999312026271407e-05,
+      "loss": 0.564,
+      "step": 6629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999310097846794e-05,
+      "loss": 0.657,
+      "step": 6630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999930816672332e-05,
+      "loss": 0.6072,
+      "step": 6631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999930623290099e-05,
+      "loss": 0.5057,
+      "step": 6632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999304296379797e-05,
+      "loss": 0.5346,
+      "step": 6633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999302357159745e-05,
+      "loss": 0.6045,
+      "step": 6634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999300415240833e-05,
+      "loss": 0.6184,
+      "step": 6635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999298470623064e-05,
+      "loss": 0.5625,
+      "step": 6636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999929652330644e-05,
+      "loss": 0.6366,
+      "step": 6637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999929457329096e-05,
+      "loss": 0.5194,
+      "step": 6638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999292620576617e-05,
+      "loss": 0.6683,
+      "step": 6639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999290665163426e-05,
+      "loss": 0.6443,
+      "step": 6640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999288707051375e-05,
+      "loss": 0.6137,
+      "step": 6641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999928674624047e-05,
+      "loss": 0.5587,
+      "step": 6642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999284782730712e-05,
+      "loss": 0.5989,
+      "step": 6643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99992828165221e-05,
+      "loss": 0.5652,
+      "step": 6644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999280847614634e-05,
+      "loss": 0.5719,
+      "step": 6645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999278876008318e-05,
+      "loss": 0.5729,
+      "step": 6646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999276901703145e-05,
+      "loss": 0.6217,
+      "step": 6647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999274924699125e-05,
+      "loss": 0.5438,
+      "step": 6648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999927294499625e-05,
+      "loss": 0.6439,
+      "step": 6649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999270962594524e-05,
+      "loss": 0.5417,
+      "step": 6650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999268977493954e-05,
+      "loss": 0.6322,
+      "step": 6651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999926698969453e-05,
+      "loss": 0.5913,
+      "step": 6652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9999264999196258e-05,
+      "loss": 0.6166,
+      "step": 6653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999926300599914e-05,
+      "loss": 0.6462,
+      "step": 6654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999261010103172e-05,
+      "loss": 0.5453,
+      "step": 6655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999259011508357e-05,
+      "loss": 0.6713,
+      "step": 6656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999925701021469e-05,
+      "loss": 0.5772,
+      "step": 6657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999255006222186e-05,
+      "loss": 0.6752,
+      "step": 6658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999925299953083e-05,
+      "loss": 0.6362,
+      "step": 6659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999250990140634e-05,
+      "loss": 0.6451,
+      "step": 6660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999248978051588e-05,
+      "loss": 0.584,
+      "step": 6661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999246963263702e-05,
+      "loss": 0.6253,
+      "step": 6662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999924494577697e-05,
+      "loss": 0.6005,
+      "step": 6663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.99992429255914e-05,
+      "loss": 0.5822,
+      "step": 6664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999924090270698e-05,
+      "loss": 0.5881,
+      "step": 6665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999238877123722e-05,
+      "loss": 0.5796,
+      "step": 6666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999236848841622e-05,
+      "loss": 0.6036,
+      "step": 6667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999923481786068e-05,
+      "loss": 0.6467,
+      "step": 6668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.99992327841809e-05,
+      "loss": 0.6293,
+      "step": 6669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999923074780228e-05,
+      "loss": 0.5444,
+      "step": 6670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999922870872482e-05,
+      "loss": 0.6095,
+      "step": 6671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999226666948523e-05,
+      "loss": 0.5827,
+      "step": 6672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999224622473385e-05,
+      "loss": 0.5885,
+      "step": 6673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999222575299414e-05,
+      "loss": 0.5589,
+      "step": 6674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999220525426604e-05,
+      "loss": 0.4893,
+      "step": 6675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999218472854956e-05,
+      "loss": 0.6142,
+      "step": 6676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999216417584475e-05,
+      "loss": 0.6615,
+      "step": 6677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999214359615154e-05,
+      "loss": 0.5731,
+      "step": 6678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999212298947003e-05,
+      "loss": 0.5824,
+      "step": 6679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999210235580016e-05,
+      "loss": 0.5415,
+      "step": 6680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999208169514195e-05,
+      "loss": 0.5485,
+      "step": 6681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999920610074954e-05,
+      "loss": 0.5885,
+      "step": 6682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999204029286057e-05,
+      "loss": 0.5598,
+      "step": 6683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999920195512374e-05,
+      "loss": 0.6019,
+      "step": 6684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999919987826259e-05,
+      "loss": 0.6346,
+      "step": 6685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999919779870261e-05,
+      "loss": 0.5416,
+      "step": 6686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999195716443804e-05,
+      "loss": 0.5561,
+      "step": 6687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999919363148616e-05,
+      "loss": 0.5679,
+      "step": 6688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999191543829693e-05,
+      "loss": 0.6509,
+      "step": 6689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999189453474396e-05,
+      "loss": 0.6516,
+      "step": 6690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999187360420272e-05,
+      "loss": 0.5768,
+      "step": 6691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999918526466732e-05,
+      "loss": 0.5524,
+      "step": 6692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999183166215535e-05,
+      "loss": 0.5278,
+      "step": 6693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999181065064932e-05,
+      "loss": 0.5736,
+      "step": 6694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.99991789612155e-05,
+      "loss": 0.5699,
+      "step": 6695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999176854667243e-05,
+      "loss": 0.528,
+      "step": 6696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999174745420163e-05,
+      "loss": 0.5115,
+      "step": 6697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999172633474257e-05,
+      "loss": 0.6525,
+      "step": 6698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999170518829528e-05,
+      "loss": 0.5586,
+      "step": 6699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999916840148598e-05,
+      "loss": 0.5118,
+      "step": 6700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999166281443604e-05,
+      "loss": 0.5955,
+      "step": 6701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999916415870241e-05,
+      "loss": 0.5631,
+      "step": 6702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999916203326239e-05,
+      "loss": 0.743,
+      "step": 6703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999159905123552e-05,
+      "loss": 0.6003,
+      "step": 6704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999157774285894e-05,
+      "loss": 0.6692,
+      "step": 6705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999915564074942e-05,
+      "loss": 0.5524,
+      "step": 6706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999153504514123e-05,
+      "loss": 0.5241,
+      "step": 6707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999915136558001e-05,
+      "loss": 0.5693,
+      "step": 6708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999914922394708e-05,
+      "loss": 0.5905,
+      "step": 6709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999147079615332e-05,
+      "loss": 0.5368,
+      "step": 6710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999144932584766e-05,
+      "loss": 0.6826,
+      "step": 6711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999142782855383e-05,
+      "loss": 0.5436,
+      "step": 6712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999140630427187e-05,
+      "loss": 0.5554,
+      "step": 6713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999913847530018e-05,
+      "loss": 0.5836,
+      "step": 6714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999913631747435e-05,
+      "loss": 0.6269,
+      "step": 6715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999134156949712e-05,
+      "loss": 0.5184,
+      "step": 6716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999131993726258e-05,
+      "loss": 0.6779,
+      "step": 6717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999129827803994e-05,
+      "loss": 0.6094,
+      "step": 6718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999127659182917e-05,
+      "loss": 0.5998,
+      "step": 6719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999912548786303e-05,
+      "loss": 0.6225,
+      "step": 6720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999912331384433e-05,
+      "loss": 0.6533,
+      "step": 6721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999912113712682e-05,
+      "loss": 0.6492,
+      "step": 6722
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999118957710503e-05,
+      "loss": 0.5896,
+      "step": 6723
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999116775595378e-05,
+      "loss": 0.559,
+      "step": 6724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999911459078144e-05,
+      "loss": 0.6251,
+      "step": 6725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999112403268696e-05,
+      "loss": 0.623,
+      "step": 6726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999110213057145e-05,
+      "loss": 0.661,
+      "step": 6727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999108020146792e-05,
+      "loss": 0.5496,
+      "step": 6728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999910582453763e-05,
+      "loss": 0.6749,
+      "step": 6729
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999910362622966e-05,
+      "loss": 0.5652,
+      "step": 6730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999101425222885e-05,
+      "loss": 0.5796,
+      "step": 6731
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999099221517307e-05,
+      "loss": 0.5521,
+      "step": 6732
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999097015112926e-05,
+      "loss": 0.5056,
+      "step": 6733
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999094806009746e-05,
+      "loss": 0.6632,
+      "step": 6734
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999092594207756e-05,
+      "loss": 0.5783,
+      "step": 6735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999090379706966e-05,
+      "loss": 0.689,
+      "step": 6736
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999088162507378e-05,
+      "loss": 0.6028,
+      "step": 6737
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999908594260899e-05,
+      "loss": 0.4755,
+      "step": 6738
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999083720011798e-05,
+      "loss": 0.7255,
+      "step": 6739
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999081494715808e-05,
+      "loss": 0.5606,
+      "step": 6740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999907926672102e-05,
+      "loss": 0.5405,
+      "step": 6741
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999077036027432e-05,
+      "loss": 0.5514,
+      "step": 6742
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999074802635046e-05,
+      "loss": 0.5477,
+      "step": 6743
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999072566543868e-05,
+      "loss": 0.6316,
+      "step": 6744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999070327753887e-05,
+      "loss": 0.7074,
+      "step": 6745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999068086265114e-05,
+      "loss": 0.6314,
+      "step": 6746
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999065842077544e-05,
+      "loss": 0.5656,
+      "step": 6747
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999063595191185e-05,
+      "loss": 0.5956,
+      "step": 6748
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999061345606027e-05,
+      "loss": 0.6097,
+      "step": 6749
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999059093322076e-05,
+      "loss": 0.5753,
+      "step": 6750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.999905683833933e-05,
+      "loss": 0.6189,
+      "step": 6751
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9999054580657793e-05,
+      "loss": 0.606,
+      "step": 6752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999052320277467e-05,
+      "loss": 0.6419,
+      "step": 6753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999905005719835e-05,
+      "loss": 0.5075,
+      "step": 6754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999047791420438e-05,
+      "loss": 0.5859,
+      "step": 6755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999904552294374e-05,
+      "loss": 0.6268,
+      "step": 6756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999043251768252e-05,
+      "loss": 0.5947,
+      "step": 6757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999040977893977e-05,
+      "loss": 0.5462,
+      "step": 6758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999038701320912e-05,
+      "loss": 0.5351,
+      "step": 6759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999903642204906e-05,
+      "loss": 0.7524,
+      "step": 6760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999034140078425e-05,
+      "loss": 0.6357,
+      "step": 6761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999031855409e-05,
+      "loss": 0.5879,
+      "step": 6762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999029568040788e-05,
+      "loss": 0.6906,
+      "step": 6763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999027277973794e-05,
+      "loss": 0.5312,
+      "step": 6764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999024985208018e-05,
+      "loss": 0.6379,
+      "step": 6765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999022689743455e-05,
+      "loss": 0.5318,
+      "step": 6766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999020391580107e-05,
+      "loss": 0.491,
+      "step": 6767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999901809071798e-05,
+      "loss": 0.56,
+      "step": 6768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999015787157073e-05,
+      "loss": 0.6221,
+      "step": 6769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999013480897384e-05,
+      "loss": 0.5727,
+      "step": 6770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999011171938913e-05,
+      "loss": 0.5749,
+      "step": 6771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999008860281662e-05,
+      "loss": 0.5948,
+      "step": 6772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999006545925632e-05,
+      "loss": 0.5464,
+      "step": 6773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999004228870824e-05,
+      "loss": 0.5431,
+      "step": 6774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9999001909117236e-05,
+      "loss": 0.5627,
+      "step": 6775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998999586664872e-05,
+      "loss": 0.6198,
+      "step": 6776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998997261513733e-05,
+      "loss": 0.6472,
+      "step": 6777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998994933663814e-05,
+      "loss": 0.5393,
+      "step": 6778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998992603115123e-05,
+      "loss": 0.5671,
+      "step": 6779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998990269867657e-05,
+      "loss": 0.5746,
+      "step": 6780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998987933921415e-05,
+      "loss": 0.6125,
+      "step": 6781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.99989855952764e-05,
+      "loss": 0.4887,
+      "step": 6782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998983253932613e-05,
+      "loss": 0.549,
+      "step": 6783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998980909890054e-05,
+      "loss": 0.6292,
+      "step": 6784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998978563148723e-05,
+      "loss": 0.5972,
+      "step": 6785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999897621370862e-05,
+      "loss": 0.531,
+      "step": 6786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998973861569746e-05,
+      "loss": 0.5748,
+      "step": 6787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998971506732104e-05,
+      "loss": 0.5927,
+      "step": 6788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999896914919569e-05,
+      "loss": 0.6838,
+      "step": 6789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999896678896051e-05,
+      "loss": 0.5972,
+      "step": 6790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998964426026562e-05,
+      "loss": 0.7308,
+      "step": 6791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998962060393844e-05,
+      "loss": 0.5842,
+      "step": 6792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999895969206236e-05,
+      "loss": 0.7264,
+      "step": 6793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998957321032116e-05,
+      "loss": 0.6592,
+      "step": 6794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998954947303102e-05,
+      "loss": 0.5895,
+      "step": 6795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998952570875322e-05,
+      "loss": 0.5378,
+      "step": 6796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998950191748784e-05,
+      "loss": 0.5255,
+      "step": 6797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998947809923476e-05,
+      "loss": 0.5559,
+      "step": 6798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998945425399407e-05,
+      "loss": 0.6279,
+      "step": 6799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998943038176578e-05,
+      "loss": 0.7146,
+      "step": 6800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998940648254984e-05,
+      "loss": 0.5085,
+      "step": 6801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999893825563463e-05,
+      "loss": 0.5648,
+      "step": 6802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998935860315517e-05,
+      "loss": 0.6723,
+      "step": 6803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998933462297646e-05,
+      "loss": 0.6121,
+      "step": 6804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998931061581014e-05,
+      "loss": 0.6631,
+      "step": 6805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998928658165626e-05,
+      "loss": 0.5481,
+      "step": 6806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999892625205148e-05,
+      "loss": 0.6144,
+      "step": 6807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998923843238574e-05,
+      "loss": 0.6071,
+      "step": 6808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998921431726913e-05,
+      "loss": 0.5616,
+      "step": 6809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998919017516497e-05,
+      "loss": 0.5481,
+      "step": 6810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998916600607325e-05,
+      "loss": 0.6043,
+      "step": 6811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998914180999402e-05,
+      "loss": 0.5817,
+      "step": 6812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999891175869272e-05,
+      "loss": 0.5828,
+      "step": 6813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999890933368729e-05,
+      "loss": 0.5065,
+      "step": 6814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998906905983105e-05,
+      "loss": 0.5612,
+      "step": 6815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998904475580167e-05,
+      "loss": 0.6197,
+      "step": 6816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999890204247848e-05,
+      "loss": 0.4845,
+      "step": 6817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999889960667804e-05,
+      "loss": 0.5713,
+      "step": 6818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998897168178857e-05,
+      "loss": 0.6653,
+      "step": 6819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999889472698092e-05,
+      "loss": 0.6017,
+      "step": 6820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998892283084236e-05,
+      "loss": 0.633,
+      "step": 6821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998889836488802e-05,
+      "loss": 0.5593,
+      "step": 6822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999888738719462e-05,
+      "loss": 0.5478,
+      "step": 6823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998884935201696e-05,
+      "loss": 0.6259,
+      "step": 6824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998882480510024e-05,
+      "loss": 0.6273,
+      "step": 6825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998880023119603e-05,
+      "loss": 0.561,
+      "step": 6826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998877563030447e-05,
+      "loss": 0.5335,
+      "step": 6827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999887510024254e-05,
+      "loss": 0.5603,
+      "step": 6828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998872634755892e-05,
+      "loss": 0.6297,
+      "step": 6829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.99988701665705e-05,
+      "loss": 0.6677,
+      "step": 6830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998867695686366e-05,
+      "loss": 0.6018,
+      "step": 6831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998865222103494e-05,
+      "loss": 0.5784,
+      "step": 6832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999886274582188e-05,
+      "loss": 0.5988,
+      "step": 6833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998860266841524e-05,
+      "loss": 0.5857,
+      "step": 6834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999885778516243e-05,
+      "loss": 0.6261,
+      "step": 6835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998855300784597e-05,
+      "loss": 0.5682,
+      "step": 6836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999885281370803e-05,
+      "loss": 0.6347,
+      "step": 6837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999885032393272e-05,
+      "loss": 0.6307,
+      "step": 6838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998847831458682e-05,
+      "loss": 0.5891,
+      "step": 6839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998845336285902e-05,
+      "loss": 0.5707,
+      "step": 6840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998842838414388e-05,
+      "loss": 0.5668,
+      "step": 6841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998840337844138e-05,
+      "loss": 0.5671,
+      "step": 6842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998837834575156e-05,
+      "loss": 0.5691,
+      "step": 6843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998835328607443e-05,
+      "loss": 0.566,
+      "step": 6844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998832819940995e-05,
+      "loss": 0.5869,
+      "step": 6845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998830308575815e-05,
+      "loss": 0.577,
+      "step": 6846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998827794511903e-05,
+      "loss": 0.5723,
+      "step": 6847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999882527774926e-05,
+      "loss": 0.6032,
+      "step": 6848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998822758287892e-05,
+      "loss": 0.7092,
+      "step": 6849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9998820236127795e-05,
+      "loss": 0.6262,
+      "step": 6850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998817711268964e-05,
+      "loss": 0.585,
+      "step": 6851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999881518371141e-05,
+      "loss": 0.5847,
+      "step": 6852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998812653455126e-05,
+      "loss": 0.5875,
+      "step": 6853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999881012050012e-05,
+      "loss": 0.6699,
+      "step": 6854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998807584846385e-05,
+      "loss": 0.627,
+      "step": 6855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998805046493926e-05,
+      "loss": 0.6614,
+      "step": 6856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999880250544274e-05,
+      "loss": 0.6254,
+      "step": 6857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998799961692836e-05,
+      "loss": 0.5331,
+      "step": 6858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998797415244205e-05,
+      "loss": 0.5428,
+      "step": 6859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998794866096853e-05,
+      "loss": 0.6259,
+      "step": 6860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999879231425078e-05,
+      "loss": 0.5797,
+      "step": 6861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998789759705984e-05,
+      "loss": 0.5667,
+      "step": 6862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999878720246247e-05,
+      "loss": 0.4779,
+      "step": 6863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999878464252024e-05,
+      "loss": 0.544,
+      "step": 6864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998782079879286e-05,
+      "loss": 0.6492,
+      "step": 6865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998779514539618e-05,
+      "loss": 0.6638,
+      "step": 6866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998776946501226e-05,
+      "loss": 0.5622,
+      "step": 6867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998774375764122e-05,
+      "loss": 0.6318,
+      "step": 6868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998771802328303e-05,
+      "loss": 0.58,
+      "step": 6869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998769226193766e-05,
+      "loss": 0.6329,
+      "step": 6870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998766647360518e-05,
+      "loss": 0.6079,
+      "step": 6871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998764065828552e-05,
+      "loss": 0.5868,
+      "step": 6872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998761481597878e-05,
+      "loss": 0.596,
+      "step": 6873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998758894668485e-05,
+      "loss": 0.6038,
+      "step": 6874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998756305040385e-05,
+      "loss": 0.6661,
+      "step": 6875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998753712713573e-05,
+      "loss": 0.5195,
+      "step": 6876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999875111768805e-05,
+      "loss": 0.5437,
+      "step": 6877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999874851996382e-05,
+      "loss": 0.6182,
+      "step": 6878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998745919540877e-05,
+      "loss": 0.5534,
+      "step": 6879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998743316419227e-05,
+      "loss": 0.6957,
+      "step": 6880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999874071059887e-05,
+      "loss": 0.5447,
+      "step": 6881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998738102079806e-05,
+      "loss": 0.5216,
+      "step": 6882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998735490862035e-05,
+      "loss": 0.588,
+      "step": 6883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999873287694556e-05,
+      "loss": 0.6622,
+      "step": 6884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999873026033038e-05,
+      "loss": 0.4942,
+      "step": 6885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998727641016495e-05,
+      "loss": 0.5728,
+      "step": 6886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998725019003906e-05,
+      "loss": 0.594,
+      "step": 6887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998722394292616e-05,
+      "loss": 0.6541,
+      "step": 6888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998719766882625e-05,
+      "loss": 0.6017,
+      "step": 6889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998717136773932e-05,
+      "loss": 0.564,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998714503966535e-05,
+      "loss": 0.6136,
+      "step": 6891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998711868460444e-05,
+      "loss": 0.6699,
+      "step": 6892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999870923025565e-05,
+      "loss": 0.6456,
+      "step": 6893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999870658935216e-05,
+      "loss": 0.6035,
+      "step": 6894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999870394574997e-05,
+      "loss": 0.6133,
+      "step": 6895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998701299449083e-05,
+      "loss": 0.5048,
+      "step": 6896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998698650449502e-05,
+      "loss": 0.5708,
+      "step": 6897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998695998751224e-05,
+      "loss": 0.6336,
+      "step": 6898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999869334435425e-05,
+      "loss": 0.5156,
+      "step": 6899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998690687258584e-05,
+      "loss": 0.5446,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998688027464226e-05,
+      "loss": 0.6009,
+      "step": 6901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998685364971174e-05,
+      "loss": 0.5885,
+      "step": 6902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999868269977943e-05,
+      "loss": 0.5587,
+      "step": 6903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998680031888992e-05,
+      "loss": 0.6581,
+      "step": 6904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998677361299867e-05,
+      "loss": 0.5956,
+      "step": 6905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998674688012053e-05,
+      "loss": 0.5491,
+      "step": 6906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999867201202555e-05,
+      "loss": 0.5513,
+      "step": 6907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998669333340354e-05,
+      "loss": 0.582,
+      "step": 6908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998666651956475e-05,
+      "loss": 0.6273,
+      "step": 6909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998663967873908e-05,
+      "loss": 0.502,
+      "step": 6910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999866128109265e-05,
+      "loss": 0.5981,
+      "step": 6911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998658591612714e-05,
+      "loss": 0.4828,
+      "step": 6912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998655899434087e-05,
+      "loss": 0.7397,
+      "step": 6913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999865320455678e-05,
+      "loss": 0.5685,
+      "step": 6914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999865050698079e-05,
+      "loss": 0.5687,
+      "step": 6915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998647806706116e-05,
+      "loss": 0.5977,
+      "step": 6916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998645103732762e-05,
+      "loss": 0.5784,
+      "step": 6917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998642398060725e-05,
+      "loss": 0.5955,
+      "step": 6918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998639689690006e-05,
+      "loss": 0.5822,
+      "step": 6919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999863697862061e-05,
+      "loss": 0.6078,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998634264852537e-05,
+      "loss": 0.6186,
+      "step": 6921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998631548385783e-05,
+      "loss": 0.6294,
+      "step": 6922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999862882922035e-05,
+      "loss": 0.6794,
+      "step": 6923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998626107356245e-05,
+      "loss": 0.5954,
+      "step": 6924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998623382793462e-05,
+      "loss": 0.5542,
+      "step": 6925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998620655532005e-05,
+      "loss": 0.5927,
+      "step": 6926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999861792557187e-05,
+      "loss": 0.6354,
+      "step": 6927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998615192913065e-05,
+      "loss": 0.5366,
+      "step": 6928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998612457555585e-05,
+      "loss": 0.5737,
+      "step": 6929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998609719499435e-05,
+      "loss": 0.6886,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999860697874461e-05,
+      "loss": 0.5632,
+      "step": 6931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998604235291116e-05,
+      "loss": 0.723,
+      "step": 6932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999860148913895e-05,
+      "loss": 0.5266,
+      "step": 6933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998598740288117e-05,
+      "loss": 0.548,
+      "step": 6934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998595988738617e-05,
+      "loss": 0.4726,
+      "step": 6935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998593234490446e-05,
+      "loss": 0.5195,
+      "step": 6936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998590477543607e-05,
+      "loss": 0.4944,
+      "step": 6937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998587717898105e-05,
+      "loss": 0.5967,
+      "step": 6938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999858495555394e-05,
+      "loss": 0.6803,
+      "step": 6939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998582190511106e-05,
+      "loss": 0.573,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998579422769605e-05,
+      "loss": 0.6142,
+      "step": 6941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998576652329444e-05,
+      "loss": 0.6232,
+      "step": 6942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998573879190622e-05,
+      "loss": 0.615,
+      "step": 6943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998571103353137e-05,
+      "loss": 0.6237,
+      "step": 6944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999856832481699e-05,
+      "loss": 0.6188,
+      "step": 6945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998565543582184e-05,
+      "loss": 0.5671,
+      "step": 6946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999856275964872e-05,
+      "loss": 0.5801,
+      "step": 6947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9998559973016592e-05,
+      "loss": 0.6419,
+      "step": 6948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.999855718368581e-05,
+      "loss": 0.7242,
+      "step": 6949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999855439165637e-05,
+      "loss": 0.6499,
+      "step": 6950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998551596928273e-05,
+      "loss": 0.6598,
+      "step": 6951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999854879950152e-05,
+      "loss": 0.5524,
+      "step": 6952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999854599937611e-05,
+      "loss": 0.6118,
+      "step": 6953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999854319655205e-05,
+      "loss": 0.61,
+      "step": 6954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998540391029332e-05,
+      "loss": 0.5887,
+      "step": 6955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998537582807965e-05,
+      "loss": 0.6109,
+      "step": 6956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998534771887943e-05,
+      "loss": 0.6386,
+      "step": 6957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999853195826927e-05,
+      "loss": 0.6182,
+      "step": 6958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999852914195195e-05,
+      "loss": 0.5362,
+      "step": 6959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999852632293598e-05,
+      "loss": 0.6292,
+      "step": 6960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998523501221356e-05,
+      "loss": 0.5708,
+      "step": 6961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999852067680809e-05,
+      "loss": 0.5463,
+      "step": 6962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999851784969617e-05,
+      "loss": 0.5316,
+      "step": 6963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998515019885606e-05,
+      "loss": 0.6122,
+      "step": 6964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998512187376398e-05,
+      "loss": 0.4742,
+      "step": 6965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998509352168544e-05,
+      "loss": 0.5014,
+      "step": 6966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998506514262045e-05,
+      "loss": 0.5854,
+      "step": 6967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.99985036736569e-05,
+      "loss": 0.5776,
+      "step": 6968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998500830353115e-05,
+      "loss": 0.5374,
+      "step": 6969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998497984350686e-05,
+      "loss": 0.6011,
+      "step": 6970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998495135649617e-05,
+      "loss": 0.5023,
+      "step": 6971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998492284249908e-05,
+      "loss": 0.6001,
+      "step": 6972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999848943015156e-05,
+      "loss": 0.6077,
+      "step": 6973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999848657335457e-05,
+      "loss": 0.6038,
+      "step": 6974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998483713858943e-05,
+      "loss": 0.5775,
+      "step": 6975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999848085166468e-05,
+      "loss": 0.5273,
+      "step": 6976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999847798677178e-05,
+      "loss": 0.655,
+      "step": 6977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999847511918024e-05,
+      "loss": 0.5768,
+      "step": 6978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999847224889007e-05,
+      "loss": 0.6133,
+      "step": 6979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998469375901262e-05,
+      "loss": 0.5876,
+      "step": 6980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998466500213825e-05,
+      "loss": 0.5271,
+      "step": 6981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999846362182775e-05,
+      "loss": 0.6076,
+      "step": 6982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998460740743044e-05,
+      "loss": 0.6054,
+      "step": 6983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999845785695971e-05,
+      "loss": 0.5055,
+      "step": 6984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998454970477743e-05,
+      "loss": 0.5972,
+      "step": 6985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998452081297145e-05,
+      "loss": 0.7194,
+      "step": 6986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999844918941792e-05,
+      "loss": 0.575,
+      "step": 6987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998446294840067e-05,
+      "loss": 0.6175,
+      "step": 6988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998443397563587e-05,
+      "loss": 0.5085,
+      "step": 6989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999844049758848e-05,
+      "loss": 0.5905,
+      "step": 6990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998437594914745e-05,
+      "loss": 0.6234,
+      "step": 6991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998434689542385e-05,
+      "loss": 0.5309,
+      "step": 6992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998431781471404e-05,
+      "loss": 0.6325,
+      "step": 6993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998428870701797e-05,
+      "loss": 0.6754,
+      "step": 6994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998425957233567e-05,
+      "loss": 0.4998,
+      "step": 6995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998423041066716e-05,
+      "loss": 0.5932,
+      "step": 6996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998420122201246e-05,
+      "loss": 0.5526,
+      "step": 6997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998417200637153e-05,
+      "loss": 0.5747,
+      "step": 6998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998414276374443e-05,
+      "loss": 0.609,
+      "step": 6999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999841134941311e-05,
+      "loss": 0.5564,
+      "step": 7000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999840841975316e-05,
+      "loss": 0.5967,
+      "step": 7001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998405487394598e-05,
+      "loss": 0.5983,
+      "step": 7002
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998402552337415e-05,
+      "loss": 0.568,
+      "step": 7003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998399614581616e-05,
+      "loss": 0.5548,
+      "step": 7004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998396674127204e-05,
+      "loss": 0.5662,
+      "step": 7005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998393730974176e-05,
+      "loss": 0.6149,
+      "step": 7006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998390785122537e-05,
+      "loss": 0.5954,
+      "step": 7007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998387836572286e-05,
+      "loss": 0.5447,
+      "step": 7008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999838488532342e-05,
+      "loss": 0.5738,
+      "step": 7009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998381931375948e-05,
+      "loss": 0.6069,
+      "step": 7010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999837897472986e-05,
+      "loss": 0.5966,
+      "step": 7011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998376015385167e-05,
+      "loss": 0.5443,
+      "step": 7012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998373053341867e-05,
+      "loss": 0.541,
+      "step": 7013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998370088599957e-05,
+      "loss": 0.5771,
+      "step": 7014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999836712115944e-05,
+      "loss": 0.6153,
+      "step": 7015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998364151020316e-05,
+      "loss": 0.5206,
+      "step": 7016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999836117818259e-05,
+      "loss": 0.6623,
+      "step": 7017
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998358202646256e-05,
+      "loss": 0.5448,
+      "step": 7018
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998355224411322e-05,
+      "loss": 0.6079,
+      "step": 7019
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998352243477785e-05,
+      "loss": 0.5775,
+      "step": 7020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999834925984564e-05,
+      "loss": 0.5831,
+      "step": 7021
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998346273514898e-05,
+      "loss": 0.5576,
+      "step": 7022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999834328448556e-05,
+      "loss": 0.5535,
+      "step": 7023
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998340292757616e-05,
+      "loss": 0.4893,
+      "step": 7024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998337298331078e-05,
+      "loss": 0.5697,
+      "step": 7025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998334301205937e-05,
+      "loss": 0.6012,
+      "step": 7026
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998331301382203e-05,
+      "loss": 0.5458,
+      "step": 7027
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999832829885987e-05,
+      "loss": 0.5621,
+      "step": 7028
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999832529363894e-05,
+      "loss": 0.6286,
+      "step": 7029
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998322285719417e-05,
+      "loss": 0.5239,
+      "step": 7030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998319275101302e-05,
+      "loss": 0.5515,
+      "step": 7031
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998316261784594e-05,
+      "loss": 0.5136,
+      "step": 7032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999831324576929e-05,
+      "loss": 0.5693,
+      "step": 7033
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998310227055396e-05,
+      "loss": 0.7212,
+      "step": 7034
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998307205642913e-05,
+      "loss": 0.659,
+      "step": 7035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999830418153184e-05,
+      "loss": 0.6951,
+      "step": 7036
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998301154722176e-05,
+      "loss": 0.5216,
+      "step": 7037
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998298125213925e-05,
+      "loss": 0.516,
+      "step": 7038
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998295093007088e-05,
+      "loss": 0.5106,
+      "step": 7039
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998292058101662e-05,
+      "loss": 0.5852,
+      "step": 7040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998289020497653e-05,
+      "loss": 0.5964,
+      "step": 7041
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998285980195055e-05,
+      "loss": 0.6134,
+      "step": 7042
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998282937193877e-05,
+      "loss": 0.5013,
+      "step": 7043
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998279891494114e-05,
+      "loss": 0.558,
+      "step": 7044
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9998276843095764e-05,
+      "loss": 0.6626,
+      "step": 7045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999827379199884e-05,
+      "loss": 0.5476,
+      "step": 7046
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999827073820333e-05,
+      "loss": 0.5314,
+      "step": 7047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998267681709244e-05,
+      "loss": 0.6834,
+      "step": 7048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998264622516578e-05,
+      "loss": 0.6427,
+      "step": 7049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999826156062533e-05,
+      "loss": 0.6613,
+      "step": 7050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998258496035507e-05,
+      "loss": 0.5838,
+      "step": 7051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998255428747107e-05,
+      "loss": 0.6136,
+      "step": 7052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999825235876013e-05,
+      "loss": 0.5688,
+      "step": 7053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998249286074582e-05,
+      "loss": 0.5933,
+      "step": 7054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998246210690454e-05,
+      "loss": 0.5736,
+      "step": 7055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998243132607757e-05,
+      "loss": 0.5959,
+      "step": 7056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998240051826484e-05,
+      "loss": 0.4998,
+      "step": 7057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998236968346642e-05,
+      "loss": 0.7227,
+      "step": 7058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999823388216823e-05,
+      "loss": 0.5884,
+      "step": 7059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998230793291244e-05,
+      "loss": 0.6999,
+      "step": 7060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998227701715692e-05,
+      "loss": 0.6214,
+      "step": 7061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999822460744157e-05,
+      "loss": 0.677,
+      "step": 7062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999822151046888e-05,
+      "loss": 0.7073,
+      "step": 7063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998218410797627e-05,
+      "loss": 0.6092,
+      "step": 7064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998215308427805e-05,
+      "loss": 0.4955,
+      "step": 7065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998212203359418e-05,
+      "loss": 0.6361,
+      "step": 7066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998209095592468e-05,
+      "loss": 0.5427,
+      "step": 7067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998205985126953e-05,
+      "loss": 0.6012,
+      "step": 7068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998202871962875e-05,
+      "loss": 0.6228,
+      "step": 7069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999819975610024e-05,
+      "loss": 0.6354,
+      "step": 7070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999819663753904e-05,
+      "loss": 0.5657,
+      "step": 7071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999819351627928e-05,
+      "loss": 0.5482,
+      "step": 7072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998190392320962e-05,
+      "loss": 0.545,
+      "step": 7073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998187265664083e-05,
+      "loss": 0.5864,
+      "step": 7074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998184136308652e-05,
+      "loss": 0.6074,
+      "step": 7075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999818100425466e-05,
+      "loss": 0.6385,
+      "step": 7076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998177869502114e-05,
+      "loss": 0.5138,
+      "step": 7077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998174732051013e-05,
+      "loss": 0.4775,
+      "step": 7078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998171591901357e-05,
+      "loss": 0.5759,
+      "step": 7079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999816844905315e-05,
+      "loss": 0.5529,
+      "step": 7080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998165303506388e-05,
+      "loss": 0.603,
+      "step": 7081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998162155261075e-05,
+      "loss": 0.6028,
+      "step": 7082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998159004317207e-05,
+      "loss": 0.6131,
+      "step": 7083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998155850674794e-05,
+      "loss": 0.5413,
+      "step": 7084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998152694333832e-05,
+      "loss": 0.4637,
+      "step": 7085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999814953529432e-05,
+      "loss": 0.62,
+      "step": 7086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998146373556265e-05,
+      "loss": 0.4519,
+      "step": 7087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999814320911966e-05,
+      "loss": 0.6641,
+      "step": 7088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999814004198451e-05,
+      "loss": 0.5603,
+      "step": 7089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998136872150814e-05,
+      "loss": 0.4926,
+      "step": 7090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998133699618577e-05,
+      "loss": 0.6002,
+      "step": 7091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998130524387797e-05,
+      "loss": 0.4788,
+      "step": 7092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998127346458473e-05,
+      "loss": 0.5386,
+      "step": 7093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998124165830606e-05,
+      "loss": 0.6529,
+      "step": 7094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998120982504204e-05,
+      "loss": 0.5757,
+      "step": 7095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998117796479257e-05,
+      "loss": 0.6498,
+      "step": 7096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999811460775577e-05,
+      "loss": 0.5367,
+      "step": 7097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999811141633375e-05,
+      "loss": 0.5633,
+      "step": 7098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998108222213194e-05,
+      "loss": 0.5277,
+      "step": 7099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.99981050253941e-05,
+      "loss": 0.5342,
+      "step": 7100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998101825876467e-05,
+      "loss": 0.6258,
+      "step": 7101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998098623660302e-05,
+      "loss": 0.5763,
+      "step": 7102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998095418745605e-05,
+      "loss": 0.5949,
+      "step": 7103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998092211132377e-05,
+      "loss": 0.6579,
+      "step": 7104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998089000820613e-05,
+      "loss": 0.6352,
+      "step": 7105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999808578781032e-05,
+      "loss": 0.579,
+      "step": 7106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998082572101497e-05,
+      "loss": 0.5655,
+      "step": 7107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998079353694142e-05,
+      "loss": 0.5376,
+      "step": 7108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999807613258826e-05,
+      "loss": 0.5522,
+      "step": 7109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998072908783852e-05,
+      "loss": 0.5497,
+      "step": 7110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998069682280915e-05,
+      "loss": 0.5294,
+      "step": 7111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998066453079456e-05,
+      "loss": 0.5821,
+      "step": 7112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998063221179472e-05,
+      "loss": 0.5788,
+      "step": 7113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999805998658096e-05,
+      "loss": 0.7236,
+      "step": 7114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998056749283926e-05,
+      "loss": 0.6146,
+      "step": 7115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999805350928837e-05,
+      "loss": 0.6146,
+      "step": 7116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998050266594297e-05,
+      "loss": 0.6118,
+      "step": 7117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998047021201698e-05,
+      "loss": 0.7408,
+      "step": 7118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999804377311058e-05,
+      "loss": 0.6263,
+      "step": 7119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998040522320946e-05,
+      "loss": 0.5765,
+      "step": 7120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998037268832793e-05,
+      "loss": 0.6086,
+      "step": 7121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998034012646124e-05,
+      "loss": 0.5452,
+      "step": 7122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998030753760934e-05,
+      "loss": 0.6962,
+      "step": 7123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998027492177233e-05,
+      "loss": 0.5645,
+      "step": 7124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998024227895017e-05,
+      "loss": 0.5857,
+      "step": 7125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999802096091429e-05,
+      "loss": 0.5621,
+      "step": 7126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998017691235047e-05,
+      "loss": 0.5533,
+      "step": 7127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998014418857293e-05,
+      "loss": 0.5195,
+      "step": 7128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998011143781028e-05,
+      "loss": 0.5251,
+      "step": 7129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999800786600625e-05,
+      "loss": 0.5175,
+      "step": 7130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999800458553297e-05,
+      "loss": 0.563,
+      "step": 7131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998001302361177e-05,
+      "loss": 0.588,
+      "step": 7132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999799801649088e-05,
+      "loss": 0.6508,
+      "step": 7133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999799472792207e-05,
+      "loss": 0.7021,
+      "step": 7134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999799143665476e-05,
+      "loss": 0.6362,
+      "step": 7135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9997988142688945e-05,
+      "loss": 0.5261,
+      "step": 7136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9997984846024626e-05,
+      "loss": 0.5762,
+      "step": 7137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9997981546661805e-05,
+      "loss": 0.6429,
+      "step": 7138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999797824460048e-05,
+      "loss": 0.6072,
+      "step": 7139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9997974939840653e-05,
+      "loss": 0.6115,
+      "step": 7140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9997971632382328e-05,
+      "loss": 0.5683,
+      "step": 7141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9997968322225505e-05,
+      "loss": 0.5605,
+      "step": 7142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999796500937018e-05,
+      "loss": 0.5681,
+      "step": 7143
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999796169381636e-05,
+      "loss": 0.702,
+      "step": 7144
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999795837556404e-05,
+      "loss": 0.5706,
+      "step": 7145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999795505461323e-05,
+      "loss": 0.5862,
+      "step": 7146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997951730963924e-05,
+      "loss": 0.5944,
+      "step": 7147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999794840461612e-05,
+      "loss": 0.5313,
+      "step": 7148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997945075569827e-05,
+      "loss": 0.5772,
+      "step": 7149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999794174382504e-05,
+      "loss": 0.6156,
+      "step": 7150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999793840938176e-05,
+      "loss": 0.6109,
+      "step": 7151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997935072239994e-05,
+      "loss": 0.6074,
+      "step": 7152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997931732399735e-05,
+      "loss": 0.5927,
+      "step": 7153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999792838986099e-05,
+      "loss": 0.5772,
+      "step": 7154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997925044623754e-05,
+      "loss": 0.6467,
+      "step": 7155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997921696688033e-05,
+      "loss": 0.6555,
+      "step": 7156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999791834605383e-05,
+      "loss": 0.5801,
+      "step": 7157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997914992721136e-05,
+      "loss": 0.5889,
+      "step": 7158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997911636689964e-05,
+      "loss": 0.6258,
+      "step": 7159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997908277960305e-05,
+      "loss": 0.7126,
+      "step": 7160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997904916532165e-05,
+      "loss": 0.6346,
+      "step": 7161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997901552405543e-05,
+      "loss": 0.5711,
+      "step": 7162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999789818558044e-05,
+      "loss": 0.6436,
+      "step": 7163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997894816056858e-05,
+      "loss": 0.5694,
+      "step": 7164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997891443834798e-05,
+      "loss": 0.5723,
+      "step": 7165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999788806891426e-05,
+      "loss": 0.6075,
+      "step": 7166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997884691295247e-05,
+      "loss": 0.5115,
+      "step": 7167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997881310977757e-05,
+      "loss": 0.5483,
+      "step": 7168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999787792796179e-05,
+      "loss": 0.5615,
+      "step": 7169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999787454224735e-05,
+      "loss": 0.5252,
+      "step": 7170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999787115383444e-05,
+      "loss": 0.6995,
+      "step": 7171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997867762723055e-05,
+      "loss": 0.5504,
+      "step": 7172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.99978643689132e-05,
+      "loss": 0.5319,
+      "step": 7173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997860972404873e-05,
+      "loss": 0.6467,
+      "step": 7174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997857573198078e-05,
+      "loss": 0.5975,
+      "step": 7175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999785417129281e-05,
+      "loss": 0.6188,
+      "step": 7176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999785076668908e-05,
+      "loss": 0.5865,
+      "step": 7177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997847359386883e-05,
+      "loss": 0.69,
+      "step": 7178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999784394938622e-05,
+      "loss": 0.6255,
+      "step": 7179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997840536687092e-05,
+      "loss": 0.6929,
+      "step": 7180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.99978371212895e-05,
+      "loss": 0.5666,
+      "step": 7181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997833703193445e-05,
+      "loss": 0.6036,
+      "step": 7182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997830282398927e-05,
+      "loss": 0.6416,
+      "step": 7183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997826858905945e-05,
+      "loss": 0.5474,
+      "step": 7184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997823432714506e-05,
+      "loss": 0.708,
+      "step": 7185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999782000382461e-05,
+      "loss": 0.6393,
+      "step": 7186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997816572236256e-05,
+      "loss": 0.66,
+      "step": 7187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999781313794944e-05,
+      "loss": 0.557,
+      "step": 7188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999780970096417e-05,
+      "loss": 0.6254,
+      "step": 7189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999780626128044e-05,
+      "loss": 0.4879,
+      "step": 7190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997802818898262e-05,
+      "loss": 0.6481,
+      "step": 7191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999779937381763e-05,
+      "loss": 0.5021,
+      "step": 7192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999779592603854e-05,
+      "loss": 0.5362,
+      "step": 7193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997792475561e-05,
+      "loss": 0.644,
+      "step": 7194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999778902238501e-05,
+      "loss": 0.5211,
+      "step": 7195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999778556651057e-05,
+      "loss": 0.6447,
+      "step": 7196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999778210793768e-05,
+      "loss": 0.6004,
+      "step": 7197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997778646666345e-05,
+      "loss": 0.5475,
+      "step": 7198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997775182696562e-05,
+      "loss": 0.5782,
+      "step": 7199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999777171602833e-05,
+      "loss": 0.6857,
+      "step": 7200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997768246661655e-05,
+      "loss": 0.5863,
+      "step": 7201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997764774596534e-05,
+      "loss": 0.5824,
+      "step": 7202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999776129983297e-05,
+      "loss": 0.6205,
+      "step": 7203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997757822370966e-05,
+      "loss": 0.5415,
+      "step": 7204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997754342210517e-05,
+      "loss": 0.6197,
+      "step": 7205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999775085935163e-05,
+      "loss": 0.5913,
+      "step": 7206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.99977473737943e-05,
+      "loss": 0.6507,
+      "step": 7207
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999774388553853e-05,
+      "loss": 0.5681,
+      "step": 7208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997740394584325e-05,
+      "loss": 0.5341,
+      "step": 7209
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997736900931685e-05,
+      "loss": 0.6012,
+      "step": 7210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999773340458061e-05,
+      "loss": 0.574,
+      "step": 7211
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997729905531095e-05,
+      "loss": 0.6386,
+      "step": 7212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999772640378315e-05,
+      "loss": 0.564,
+      "step": 7213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997722899336768e-05,
+      "loss": 0.554,
+      "step": 7214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997719392191957e-05,
+      "loss": 0.6941,
+      "step": 7215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997715882348716e-05,
+      "loss": 0.6524,
+      "step": 7216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997712369807042e-05,
+      "loss": 0.5914,
+      "step": 7217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997708854566938e-05,
+      "loss": 0.5115,
+      "step": 7218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999770533662841e-05,
+      "loss": 0.6087,
+      "step": 7219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999770181599145e-05,
+      "loss": 0.6325,
+      "step": 7220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997698292656068e-05,
+      "loss": 0.5969,
+      "step": 7221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997694766622257e-05,
+      "loss": 0.6391,
+      "step": 7222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999769123789002e-05,
+      "loss": 0.6954,
+      "step": 7223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999768770645936e-05,
+      "loss": 0.5474,
+      "step": 7224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997684172330282e-05,
+      "loss": 0.4992,
+      "step": 7225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997680635502778e-05,
+      "loss": 0.5798,
+      "step": 7226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997677095976856e-05,
+      "loss": 0.5727,
+      "step": 7227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997673553752512e-05,
+      "loss": 0.6008,
+      "step": 7228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999767000882975e-05,
+      "loss": 0.5957,
+      "step": 7229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997666461208573e-05,
+      "loss": 0.6453,
+      "step": 7230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999766291088897e-05,
+      "loss": 0.6493,
+      "step": 7231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999765935787096e-05,
+      "loss": 0.6364,
+      "step": 7232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997655802154533e-05,
+      "loss": 0.5141,
+      "step": 7233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999765224373969e-05,
+      "loss": 0.5795,
+      "step": 7234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997648682626435e-05,
+      "loss": 0.634,
+      "step": 7235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997645118814768e-05,
+      "loss": 0.4744,
+      "step": 7236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999764155230469e-05,
+      "loss": 0.4928,
+      "step": 7237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.99976379830962e-05,
+      "loss": 0.6195,
+      "step": 7238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997634411189308e-05,
+      "loss": 0.5512,
+      "step": 7239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997630836583998e-05,
+      "loss": 0.6249,
+      "step": 7240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997627259280285e-05,
+      "loss": 0.5141,
+      "step": 7241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997623679278166e-05,
+      "loss": 0.5592,
+      "step": 7242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999762009657764e-05,
+      "loss": 0.5534,
+      "step": 7243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9997616511178712e-05,
+      "loss": 0.6391,
+      "step": 7244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999761292308138e-05,
+      "loss": 0.6264,
+      "step": 7245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997609332285646e-05,
+      "loss": 0.6803,
+      "step": 7246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997605738791508e-05,
+      "loss": 0.5651,
+      "step": 7247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999760214259897e-05,
+      "loss": 0.5228,
+      "step": 7248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999759854370803e-05,
+      "loss": 0.5297,
+      "step": 7249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997594942118698e-05,
+      "loss": 0.6319,
+      "step": 7250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997591337830965e-05,
+      "loss": 0.6067,
+      "step": 7251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997587730844836e-05,
+      "loss": 0.5274,
+      "step": 7252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999758412116031e-05,
+      "loss": 0.5833,
+      "step": 7253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999758050877739e-05,
+      "loss": 0.5727,
+      "step": 7254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997576893696078e-05,
+      "loss": 0.594,
+      "step": 7255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999757327591637e-05,
+      "loss": 0.5123,
+      "step": 7256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997569655438274e-05,
+      "loss": 0.5253,
+      "step": 7257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997566032261784e-05,
+      "loss": 0.571,
+      "step": 7258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997562406386905e-05,
+      "loss": 0.6615,
+      "step": 7259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997558777813637e-05,
+      "loss": 0.4701,
+      "step": 7260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997555146541982e-05,
+      "loss": 0.5526,
+      "step": 7261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999755151257194e-05,
+      "loss": 0.6484,
+      "step": 7262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999754787590351e-05,
+      "loss": 0.5699,
+      "step": 7263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997544236536698e-05,
+      "loss": 0.5577,
+      "step": 7264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997540594471503e-05,
+      "loss": 0.5537,
+      "step": 7265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997536949707924e-05,
+      "loss": 0.5694,
+      "step": 7266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999753330224596e-05,
+      "loss": 0.6108,
+      "step": 7267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997529652085616e-05,
+      "loss": 0.5173,
+      "step": 7268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997525999226893e-05,
+      "loss": 0.5914,
+      "step": 7269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997522343669794e-05,
+      "loss": 0.6287,
+      "step": 7270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997518685414313e-05,
+      "loss": 0.5735,
+      "step": 7271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997515024460455e-05,
+      "loss": 0.5179,
+      "step": 7272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997511360808225e-05,
+      "loss": 0.6853,
+      "step": 7273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997507694457616e-05,
+      "loss": 0.6314,
+      "step": 7274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997504025408634e-05,
+      "loss": 0.5273,
+      "step": 7275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999750035366128e-05,
+      "loss": 0.5947,
+      "step": 7276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997496679215553e-05,
+      "loss": 0.5884,
+      "step": 7277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997493002071453e-05,
+      "loss": 0.586,
+      "step": 7278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997489322228985e-05,
+      "loss": 0.5828,
+      "step": 7279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997485639688147e-05,
+      "loss": 0.5932,
+      "step": 7280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999748195444894e-05,
+      "loss": 0.5438,
+      "step": 7281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999747826651137e-05,
+      "loss": 0.6001,
+      "step": 7282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999747457587543e-05,
+      "loss": 0.5922,
+      "step": 7283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997470882541128e-05,
+      "loss": 0.5495,
+      "step": 7284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999746718650846e-05,
+      "loss": 0.7051,
+      "step": 7285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999746348777743e-05,
+      "loss": 0.5854,
+      "step": 7286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997459786348035e-05,
+      "loss": 0.6314,
+      "step": 7287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997456082220278e-05,
+      "loss": 0.5964,
+      "step": 7288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997452375394165e-05,
+      "loss": 0.6369,
+      "step": 7289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997448665869694e-05,
+      "loss": 0.6033,
+      "step": 7290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999744495364686e-05,
+      "loss": 0.6307,
+      "step": 7291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997441238725674e-05,
+      "loss": 0.6146,
+      "step": 7292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999743752110613e-05,
+      "loss": 0.5192,
+      "step": 7293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997433800788228e-05,
+      "loss": 0.7979,
+      "step": 7294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997430077771975e-05,
+      "loss": 0.5699,
+      "step": 7295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997426352057366e-05,
+      "loss": 0.6212,
+      "step": 7296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997422623644405e-05,
+      "loss": 0.6199,
+      "step": 7297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997418892533096e-05,
+      "loss": 0.5812,
+      "step": 7298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997415158723434e-05,
+      "loss": 0.6336,
+      "step": 7299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997411422215424e-05,
+      "loss": 0.6119,
+      "step": 7300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997407683009068e-05,
+      "loss": 0.5358,
+      "step": 7301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999740394110436e-05,
+      "loss": 0.5134,
+      "step": 7302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999740019650131e-05,
+      "loss": 0.5337,
+      "step": 7303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997396449199914e-05,
+      "loss": 0.6531,
+      "step": 7304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997392699200176e-05,
+      "loss": 0.5189,
+      "step": 7305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997388946502093e-05,
+      "loss": 0.5661,
+      "step": 7306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997385191105668e-05,
+      "loss": 0.5373,
+      "step": 7307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.99973814330109e-05,
+      "loss": 0.6366,
+      "step": 7308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997377672217795e-05,
+      "loss": 0.6264,
+      "step": 7309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999737390872635e-05,
+      "loss": 0.4972,
+      "step": 7310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997370142536567e-05,
+      "loss": 0.5863,
+      "step": 7311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997366373648445e-05,
+      "loss": 0.6088,
+      "step": 7312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999736260206199e-05,
+      "loss": 0.5729,
+      "step": 7313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.99973588277772e-05,
+      "loss": 0.5258,
+      "step": 7314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997355050794075e-05,
+      "loss": 0.5764,
+      "step": 7315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999735127111262e-05,
+      "loss": 0.5141,
+      "step": 7316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997347488732828e-05,
+      "loss": 0.5304,
+      "step": 7317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999734370365471e-05,
+      "loss": 0.5451,
+      "step": 7318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997339915878256e-05,
+      "loss": 0.5674,
+      "step": 7319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999733612540348e-05,
+      "loss": 0.6084,
+      "step": 7320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997332332230374e-05,
+      "loss": 0.5777,
+      "step": 7321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999732853635894e-05,
+      "loss": 0.5916,
+      "step": 7322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999732473778918e-05,
+      "loss": 0.5283,
+      "step": 7323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997320936521098e-05,
+      "loss": 0.6394,
+      "step": 7324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997317132554692e-05,
+      "loss": 0.6351,
+      "step": 7325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999731332588996e-05,
+      "loss": 0.56,
+      "step": 7326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999730951652691e-05,
+      "loss": 0.5931,
+      "step": 7327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997305704465538e-05,
+      "loss": 0.6331,
+      "step": 7328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997301889705846e-05,
+      "loss": 0.6529,
+      "step": 7329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997298072247838e-05,
+      "loss": 0.616,
+      "step": 7330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999729425209151e-05,
+      "loss": 0.5514,
+      "step": 7331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997290429236865e-05,
+      "loss": 0.5535,
+      "step": 7332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997286603683906e-05,
+      "loss": 0.5399,
+      "step": 7333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997282775432632e-05,
+      "loss": 0.6071,
+      "step": 7334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997278944483047e-05,
+      "loss": 0.5895,
+      "step": 7335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997275110835146e-05,
+      "loss": 0.6427,
+      "step": 7336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997271274488934e-05,
+      "loss": 0.5451,
+      "step": 7337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997267435444414e-05,
+      "loss": 0.6384,
+      "step": 7338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997263593701582e-05,
+      "loss": 0.6385,
+      "step": 7339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997259749260446e-05,
+      "loss": 0.6335,
+      "step": 7340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9997255902121e-05,
+      "loss": 0.6175,
+      "step": 7341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999725205228325e-05,
+      "loss": 0.573,
+      "step": 7342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.999724819974719e-05,
+      "loss": 0.5768,
+      "step": 7343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997244344512832e-05,
+      "loss": 0.5491,
+      "step": 7344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999724048658017e-05,
+      "loss": 0.674,
+      "step": 7345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.99972366259492e-05,
+      "loss": 0.5816,
+      "step": 7346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997232762619934e-05,
+      "loss": 0.5104,
+      "step": 7347
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999722889659237e-05,
+      "loss": 0.6153,
+      "step": 7348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997225027866504e-05,
+      "loss": 0.5036,
+      "step": 7349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999722115644234e-05,
+      "loss": 0.6971,
+      "step": 7350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999721728231988e-05,
+      "loss": 0.5902,
+      "step": 7351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997213405499125e-05,
+      "loss": 0.7557,
+      "step": 7352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997209525980078e-05,
+      "loss": 0.6343,
+      "step": 7353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997205643762732e-05,
+      "loss": 0.5475,
+      "step": 7354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997201758847095e-05,
+      "loss": 0.5128,
+      "step": 7355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997197871233167e-05,
+      "loss": 0.5016,
+      "step": 7356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999719398092095e-05,
+      "loss": 0.5211,
+      "step": 7357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997190087910444e-05,
+      "loss": 0.5824,
+      "step": 7358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997186192201646e-05,
+      "loss": 0.6191,
+      "step": 7359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997182293794563e-05,
+      "loss": 0.6086,
+      "step": 7360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997178392689193e-05,
+      "loss": 0.6028,
+      "step": 7361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999717448888554e-05,
+      "loss": 0.5868,
+      "step": 7362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997170582383602e-05,
+      "loss": 0.6687,
+      "step": 7363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997166673183378e-05,
+      "loss": 0.563,
+      "step": 7364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997162761284876e-05,
+      "loss": 0.5674,
+      "step": 7365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999715884668809e-05,
+      "loss": 0.6001,
+      "step": 7366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997154929393027e-05,
+      "loss": 0.6602,
+      "step": 7367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997151009399685e-05,
+      "loss": 0.6115,
+      "step": 7368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997147086708063e-05,
+      "loss": 0.5993,
+      "step": 7369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997143161318163e-05,
+      "loss": 0.4675,
+      "step": 7370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997139233229992e-05,
+      "loss": 0.5535,
+      "step": 7371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997135302443544e-05,
+      "loss": 0.535,
+      "step": 7372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999713136895882e-05,
+      "loss": 0.6139,
+      "step": 7373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997127432775827e-05,
+      "loss": 0.6237,
+      "step": 7374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997123493894562e-05,
+      "loss": 0.5987,
+      "step": 7375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997119552315024e-05,
+      "loss": 0.5729,
+      "step": 7376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999711560803722e-05,
+      "loss": 0.5703,
+      "step": 7377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997111661061147e-05,
+      "loss": 0.6115,
+      "step": 7378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997107711386805e-05,
+      "loss": 0.5428,
+      "step": 7379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.99971037590142e-05,
+      "loss": 0.5226,
+      "step": 7380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997099803943327e-05,
+      "loss": 0.5958,
+      "step": 7381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997095846174193e-05,
+      "loss": 0.5145,
+      "step": 7382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997091885706795e-05,
+      "loss": 0.643,
+      "step": 7383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997087922541133e-05,
+      "loss": 0.5365,
+      "step": 7384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999708395667721e-05,
+      "loss": 0.5942,
+      "step": 7385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999707998811503e-05,
+      "loss": 0.5377,
+      "step": 7386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999707601685459e-05,
+      "loss": 0.6126,
+      "step": 7387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997072042895892e-05,
+      "loss": 0.5701,
+      "step": 7388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999706806623894e-05,
+      "loss": 0.6234,
+      "step": 7389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999706408688373e-05,
+      "loss": 0.5397,
+      "step": 7390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997060104830267e-05,
+      "loss": 0.6796,
+      "step": 7391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999705612007855e-05,
+      "loss": 0.62,
+      "step": 7392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997052132628582e-05,
+      "loss": 0.5642,
+      "step": 7393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999704814248036e-05,
+      "loss": 0.6171,
+      "step": 7394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999704414963389e-05,
+      "loss": 0.5833,
+      "step": 7395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997040154089172e-05,
+      "loss": 0.5979,
+      "step": 7396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997036155846206e-05,
+      "loss": 0.6145,
+      "step": 7397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997032154904995e-05,
+      "loss": 0.6137,
+      "step": 7398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997028151265534e-05,
+      "loss": 0.5003,
+      "step": 7399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999702414492783e-05,
+      "loss": 0.4947,
+      "step": 7400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997020135891884e-05,
+      "loss": 0.6516,
+      "step": 7401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997016124157692e-05,
+      "loss": 0.5928,
+      "step": 7402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997012109725263e-05,
+      "loss": 0.5751,
+      "step": 7403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999700809259459e-05,
+      "loss": 0.5254,
+      "step": 7404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999700407276568e-05,
+      "loss": 0.5984,
+      "step": 7405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9997000050238532e-05,
+      "loss": 0.691,
+      "step": 7406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996996025013145e-05,
+      "loss": 0.5674,
+      "step": 7407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996991997089524e-05,
+      "loss": 0.5482,
+      "step": 7408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996987966467665e-05,
+      "loss": 0.5302,
+      "step": 7409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996983933147577e-05,
+      "loss": 0.6747,
+      "step": 7410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999697989712925e-05,
+      "loss": 0.5272,
+      "step": 7411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996975858412696e-05,
+      "loss": 0.698,
+      "step": 7412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996971816997914e-05,
+      "loss": 0.5039,
+      "step": 7413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996967772884898e-05,
+      "loss": 0.5628,
+      "step": 7414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999696372607365e-05,
+      "loss": 0.6316,
+      "step": 7415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999695967656418e-05,
+      "loss": 0.6308,
+      "step": 7416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999695562435649e-05,
+      "loss": 0.6288,
+      "step": 7417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996951569450568e-05,
+      "loss": 0.5976,
+      "step": 7418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999694751184642e-05,
+      "loss": 0.5805,
+      "step": 7419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999694345154405e-05,
+      "loss": 0.544,
+      "step": 7420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999693938854346e-05,
+      "loss": 0.6659,
+      "step": 7421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999693532284465e-05,
+      "loss": 0.6319,
+      "step": 7422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999693125444762e-05,
+      "loss": 0.5789,
+      "step": 7423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996927183352372e-05,
+      "loss": 0.5712,
+      "step": 7424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996923109558905e-05,
+      "loss": 0.7122,
+      "step": 7425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996919033067225e-05,
+      "loss": 0.5751,
+      "step": 7426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996914953877327e-05,
+      "loss": 0.5154,
+      "step": 7427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996910871989213e-05,
+      "loss": 0.6605,
+      "step": 7428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996906787402886e-05,
+      "loss": 0.5582,
+      "step": 7429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999690270011835e-05,
+      "loss": 0.5676,
+      "step": 7430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996898610135605e-05,
+      "loss": 0.5819,
+      "step": 7431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999689451745465e-05,
+      "loss": 0.734,
+      "step": 7432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996890422075482e-05,
+      "loss": 0.6692,
+      "step": 7433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999688632399811e-05,
+      "loss": 0.5256,
+      "step": 7434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999688222322253e-05,
+      "loss": 0.6484,
+      "step": 7435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996878119748742e-05,
+      "loss": 0.5972,
+      "step": 7436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996874013576756e-05,
+      "loss": 0.7024,
+      "step": 7437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996869904706564e-05,
+      "loss": 0.6364,
+      "step": 7438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999686579313817e-05,
+      "loss": 0.6093,
+      "step": 7439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9996861678871576e-05,
+      "loss": 0.6058,
+      "step": 7440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999685756190678e-05,
+      "loss": 0.5246,
+      "step": 7441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999685344224379e-05,
+      "loss": 0.5868,
+      "step": 7442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996849319882598e-05,
+      "loss": 0.6293,
+      "step": 7443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999684519482321e-05,
+      "loss": 0.6793,
+      "step": 7444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999684106706563e-05,
+      "loss": 0.5885,
+      "step": 7445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996836936609853e-05,
+      "loss": 0.5981,
+      "step": 7446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996832803455883e-05,
+      "loss": 0.6218,
+      "step": 7447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996828667603723e-05,
+      "loss": 0.5359,
+      "step": 7448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999682452905337e-05,
+      "loss": 0.5272,
+      "step": 7449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999682038780483e-05,
+      "loss": 0.5975,
+      "step": 7450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.99968162438581e-05,
+      "loss": 0.5734,
+      "step": 7451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996812097213183e-05,
+      "loss": 0.5188,
+      "step": 7452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999680794787008e-05,
+      "loss": 0.634,
+      "step": 7453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999680379582879e-05,
+      "loss": 0.6115,
+      "step": 7454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996799641089316e-05,
+      "loss": 0.626,
+      "step": 7455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999679548365166e-05,
+      "loss": 0.5901,
+      "step": 7456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999679132351582e-05,
+      "loss": 0.5811,
+      "step": 7457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996787160681804e-05,
+      "loss": 0.4565,
+      "step": 7458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996782995149606e-05,
+      "loss": 0.6173,
+      "step": 7459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999677882691923e-05,
+      "loss": 0.6898,
+      "step": 7460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996774655990677e-05,
+      "loss": 0.5847,
+      "step": 7461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996770482363948e-05,
+      "loss": 0.5486,
+      "step": 7462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999676630603904e-05,
+      "loss": 0.5731,
+      "step": 7463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999676212701596e-05,
+      "loss": 0.59,
+      "step": 7464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996757945294712e-05,
+      "loss": 0.5419,
+      "step": 7465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996753760875286e-05,
+      "loss": 0.5849,
+      "step": 7466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999674957375769e-05,
+      "loss": 0.5855,
+      "step": 7467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996745383941926e-05,
+      "loss": 0.6502,
+      "step": 7468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996741191427996e-05,
+      "loss": 0.6379,
+      "step": 7469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996736996215896e-05,
+      "loss": 0.5573,
+      "step": 7470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999673279830563e-05,
+      "loss": 0.6286,
+      "step": 7471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996728597697203e-05,
+      "loss": 0.5841,
+      "step": 7472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999672439439061e-05,
+      "loss": 0.5901,
+      "step": 7473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999672018838585e-05,
+      "loss": 0.58,
+      "step": 7474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996715979682934e-05,
+      "loss": 0.5838,
+      "step": 7475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996711768281853e-05,
+      "loss": 0.6478,
+      "step": 7476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996707554182616e-05,
+      "loss": 0.5509,
+      "step": 7477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996703337385223e-05,
+      "loss": 0.5751,
+      "step": 7478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996699117889668e-05,
+      "loss": 0.6344,
+      "step": 7479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999669489569596e-05,
+      "loss": 0.5809,
+      "step": 7480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996690670804097e-05,
+      "loss": 0.5503,
+      "step": 7481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996686443214082e-05,
+      "loss": 0.5585,
+      "step": 7482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999668221292591e-05,
+      "loss": 0.633,
+      "step": 7483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999667797993959e-05,
+      "loss": 0.5846,
+      "step": 7484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996673744255122e-05,
+      "loss": 0.6187,
+      "step": 7485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996669505872497e-05,
+      "loss": 0.5691,
+      "step": 7486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996665264791734e-05,
+      "loss": 0.5969,
+      "step": 7487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999666102101282e-05,
+      "loss": 0.5294,
+      "step": 7488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996656774535757e-05,
+      "loss": 0.6161,
+      "step": 7489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996652525360557e-05,
+      "loss": 0.5343,
+      "step": 7490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999664827348721e-05,
+      "loss": 0.5245,
+      "step": 7491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996644018915717e-05,
+      "loss": 0.636,
+      "step": 7492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996639761646088e-05,
+      "loss": 0.4902,
+      "step": 7493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999663550167832e-05,
+      "loss": 0.5853,
+      "step": 7494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999663123901241e-05,
+      "loss": 0.5441,
+      "step": 7495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996626973648366e-05,
+      "loss": 0.6275,
+      "step": 7496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999662270558618e-05,
+      "loss": 0.5923,
+      "step": 7497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996618434825865e-05,
+      "loss": 0.5751,
+      "step": 7498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996614161367413e-05,
+      "loss": 0.6125,
+      "step": 7499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999660988521083e-05,
+      "loss": 0.477,
+      "step": 7500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999660560635611e-05,
+      "loss": 0.6533,
+      "step": 7501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996601324803264e-05,
+      "loss": 0.6461,
+      "step": 7502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996597040552288e-05,
+      "loss": 0.671,
+      "step": 7503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996592753603183e-05,
+      "loss": 0.57,
+      "step": 7504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999658846395595e-05,
+      "loss": 0.5172,
+      "step": 7505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996584171610593e-05,
+      "loss": 0.5844,
+      "step": 7506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999657987656711e-05,
+      "loss": 0.5685,
+      "step": 7507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996575578825507e-05,
+      "loss": 0.6785,
+      "step": 7508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996571278385776e-05,
+      "loss": 0.6537,
+      "step": 7509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996566975247926e-05,
+      "loss": 0.6239,
+      "step": 7510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996562669411955e-05,
+      "loss": 0.5817,
+      "step": 7511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996558360877868e-05,
+      "loss": 0.626,
+      "step": 7512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996554049645657e-05,
+      "loss": 0.5682,
+      "step": 7513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996549735715337e-05,
+      "loss": 0.577,
+      "step": 7514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.99965454190869e-05,
+      "loss": 0.5658,
+      "step": 7515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999654109976034e-05,
+      "loss": 0.4996,
+      "step": 7516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996536777735677e-05,
+      "loss": 0.5643,
+      "step": 7517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996532453012897e-05,
+      "loss": 0.6356,
+      "step": 7518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999652812559201e-05,
+      "loss": 0.6752,
+      "step": 7519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996523795473012e-05,
+      "loss": 0.565,
+      "step": 7520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.99965194626559e-05,
+      "loss": 0.6079,
+      "step": 7521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996515127140687e-05,
+      "loss": 0.5476,
+      "step": 7522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996510788927366e-05,
+      "loss": 0.5666,
+      "step": 7523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999650644801594e-05,
+      "loss": 0.5652,
+      "step": 7524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999650210440641e-05,
+      "loss": 0.5635,
+      "step": 7525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996497758098776e-05,
+      "loss": 0.6021,
+      "step": 7526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996493409093042e-05,
+      "loss": 0.611,
+      "step": 7527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996489057389207e-05,
+      "loss": 0.6693,
+      "step": 7528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996484702987273e-05,
+      "loss": 0.591,
+      "step": 7529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996480345887245e-05,
+      "loss": 0.5628,
+      "step": 7530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996475986089115e-05,
+      "loss": 0.6074,
+      "step": 7531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999647162359289e-05,
+      "loss": 0.5203,
+      "step": 7532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999646725839857e-05,
+      "loss": 0.6313,
+      "step": 7533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999646289050616e-05,
+      "loss": 0.5948,
+      "step": 7534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999645851991566e-05,
+      "loss": 0.506,
+      "step": 7535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996454146627063e-05,
+      "loss": 0.5839,
+      "step": 7536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999644977064038e-05,
+      "loss": 0.5145,
+      "step": 7537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996445391955606e-05,
+      "loss": 0.6176,
+      "step": 7538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9996441010572746e-05,
+      "loss": 0.5148,
+      "step": 7539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.99964366264918e-05,
+      "loss": 0.6289,
+      "step": 7540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996432239712772e-05,
+      "loss": 0.5216,
+      "step": 7541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996427850235654e-05,
+      "loss": 0.5053,
+      "step": 7542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996423458060456e-05,
+      "loss": 0.5655,
+      "step": 7543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996419063187176e-05,
+      "loss": 0.5151,
+      "step": 7544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999641466561582e-05,
+      "loss": 0.6174,
+      "step": 7545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999641026534638e-05,
+      "loss": 0.5592,
+      "step": 7546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996405862378864e-05,
+      "loss": 0.6019,
+      "step": 7547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999640145671327e-05,
+      "loss": 0.654,
+      "step": 7548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996397048349602e-05,
+      "loss": 0.6788,
+      "step": 7549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999639263728786e-05,
+      "loss": 0.5782,
+      "step": 7550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996388223528045e-05,
+      "loss": 0.5888,
+      "step": 7551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999638380707016e-05,
+      "loss": 0.5244,
+      "step": 7552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996379387914198e-05,
+      "loss": 0.5452,
+      "step": 7553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999637496606017e-05,
+      "loss": 0.5918,
+      "step": 7554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996370541508074e-05,
+      "loss": 0.6606,
+      "step": 7555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999636611425791e-05,
+      "loss": 0.6624,
+      "step": 7556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996361684309682e-05,
+      "loss": 0.5864,
+      "step": 7557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996357251663388e-05,
+      "loss": 0.579,
+      "step": 7558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999635281631903e-05,
+      "loss": 0.6968,
+      "step": 7559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999634837827661e-05,
+      "loss": 0.5236,
+      "step": 7560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999634393753613e-05,
+      "loss": 0.635,
+      "step": 7561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996339494097588e-05,
+      "loss": 0.6503,
+      "step": 7562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999633504796099e-05,
+      "loss": 0.5997,
+      "step": 7563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999633059912633e-05,
+      "loss": 0.6784,
+      "step": 7564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999632614759362e-05,
+      "loss": 0.5786,
+      "step": 7565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999632169336285e-05,
+      "loss": 0.6725,
+      "step": 7566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996317236434028e-05,
+      "loss": 0.5942,
+      "step": 7567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996312776807153e-05,
+      "loss": 0.6131,
+      "step": 7568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996308314482224e-05,
+      "loss": 0.5791,
+      "step": 7569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996303849459248e-05,
+      "loss": 0.533,
+      "step": 7570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999629938173822e-05,
+      "loss": 0.6087,
+      "step": 7571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996294911319144e-05,
+      "loss": 0.6799,
+      "step": 7572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996290438202026e-05,
+      "loss": 0.6391,
+      "step": 7573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999628596238686e-05,
+      "loss": 0.5399,
+      "step": 7574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996281483873648e-05,
+      "loss": 0.4942,
+      "step": 7575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996277002662395e-05,
+      "loss": 0.664,
+      "step": 7576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996272518753098e-05,
+      "loss": 0.5308,
+      "step": 7577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999626803214576e-05,
+      "loss": 0.5655,
+      "step": 7578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996263542840382e-05,
+      "loss": 0.6126,
+      "step": 7579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999625905083697e-05,
+      "loss": 0.5997,
+      "step": 7580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999625455613552e-05,
+      "loss": 0.5309,
+      "step": 7581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999625005873603e-05,
+      "loss": 0.6603,
+      "step": 7582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999624555863851e-05,
+      "loss": 0.6377,
+      "step": 7583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996241055842957e-05,
+      "loss": 0.5696,
+      "step": 7584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999623655034937e-05,
+      "loss": 0.6129,
+      "step": 7585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999623204215775e-05,
+      "loss": 0.5852,
+      "step": 7586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.99962275312681e-05,
+      "loss": 0.5222,
+      "step": 7587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996223017680423e-05,
+      "loss": 0.5378,
+      "step": 7588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999621850139472e-05,
+      "loss": 0.6736,
+      "step": 7589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999621398241099e-05,
+      "loss": 0.5055,
+      "step": 7590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996209460729234e-05,
+      "loss": 0.655,
+      "step": 7591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996204936349455e-05,
+      "loss": 0.6251,
+      "step": 7592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996200409271656e-05,
+      "loss": 0.6791,
+      "step": 7593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996195879495834e-05,
+      "loss": 0.5045,
+      "step": 7594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999619134702199e-05,
+      "loss": 0.5652,
+      "step": 7595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996186811850128e-05,
+      "loss": 0.6552,
+      "step": 7596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996182273980248e-05,
+      "loss": 0.544,
+      "step": 7597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996177733412355e-05,
+      "loss": 0.5798,
+      "step": 7598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996173190146445e-05,
+      "loss": 0.5686,
+      "step": 7599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999616864418252e-05,
+      "loss": 0.5766,
+      "step": 7600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996164095520585e-05,
+      "loss": 0.6396,
+      "step": 7601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996159544160635e-05,
+      "loss": 0.6052,
+      "step": 7602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996154990102678e-05,
+      "loss": 0.6193,
+      "step": 7603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999615043334671e-05,
+      "loss": 0.6048,
+      "step": 7604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996145873892738e-05,
+      "loss": 0.5747,
+      "step": 7605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996141311740754e-05,
+      "loss": 0.5817,
+      "step": 7606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996136746890768e-05,
+      "loss": 0.6484,
+      "step": 7607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996132179342778e-05,
+      "loss": 0.5207,
+      "step": 7608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996127609096785e-05,
+      "loss": 0.5723,
+      "step": 7609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999612303615279e-05,
+      "loss": 0.6115,
+      "step": 7610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996118460510793e-05,
+      "loss": 0.6403,
+      "step": 7611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996113882170797e-05,
+      "loss": 0.5475,
+      "step": 7612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996109301132808e-05,
+      "loss": 0.562,
+      "step": 7613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999610471739682e-05,
+      "loss": 0.5007,
+      "step": 7614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996100130962833e-05,
+      "loss": 0.6492,
+      "step": 7615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996095541830852e-05,
+      "loss": 0.5614,
+      "step": 7616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996090950000884e-05,
+      "loss": 0.6646,
+      "step": 7617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996086355472923e-05,
+      "loss": 0.6057,
+      "step": 7618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996081758246965e-05,
+      "loss": 0.5644,
+      "step": 7619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996077158323025e-05,
+      "loss": 0.6525,
+      "step": 7620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996072555701095e-05,
+      "loss": 0.6206,
+      "step": 7621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996067950381178e-05,
+      "loss": 0.6294,
+      "step": 7622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996063342363272e-05,
+      "loss": 0.6221,
+      "step": 7623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996058731647387e-05,
+      "loss": 0.5788,
+      "step": 7624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996054118233515e-05,
+      "loss": 0.6532,
+      "step": 7625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996049502121663e-05,
+      "loss": 0.6253,
+      "step": 7626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996044883311833e-05,
+      "loss": 0.702,
+      "step": 7627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999604026180402e-05,
+      "loss": 0.6587,
+      "step": 7628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999603563759823e-05,
+      "loss": 0.6776,
+      "step": 7629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996031010694463e-05,
+      "loss": 0.6812,
+      "step": 7630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999602638109272e-05,
+      "loss": 0.5863,
+      "step": 7631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996021748793006e-05,
+      "loss": 0.5713,
+      "step": 7632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996017113795315e-05,
+      "loss": 0.5834,
+      "step": 7633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996012476099658e-05,
+      "loss": 0.5711,
+      "step": 7634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996007835706025e-05,
+      "loss": 0.5751,
+      "step": 7635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9996003192614423e-05,
+      "loss": 0.6766,
+      "step": 7636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9995998546824858e-05,
+      "loss": 0.5915,
+      "step": 7637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999599389833732e-05,
+      "loss": 0.5501,
+      "step": 7638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999598924715182e-05,
+      "loss": 0.6201,
+      "step": 7639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995984593268357e-05,
+      "loss": 0.6309,
+      "step": 7640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995979936686928e-05,
+      "loss": 0.5869,
+      "step": 7641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995975277407537e-05,
+      "loss": 0.6734,
+      "step": 7642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999597061543019e-05,
+      "loss": 0.5982,
+      "step": 7643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999596595075488e-05,
+      "loss": 0.6389,
+      "step": 7644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999596128338161e-05,
+      "loss": 0.6241,
+      "step": 7645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995956613310386e-05,
+      "loss": 0.6678,
+      "step": 7646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995951940541206e-05,
+      "loss": 0.6259,
+      "step": 7647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995947265074073e-05,
+      "loss": 0.6552,
+      "step": 7648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995942586908988e-05,
+      "loss": 0.5266,
+      "step": 7649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999593790604595e-05,
+      "loss": 0.699,
+      "step": 7650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995933222484963e-05,
+      "loss": 0.6106,
+      "step": 7651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995928536226024e-05,
+      "loss": 0.6038,
+      "step": 7652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999592384726914e-05,
+      "loss": 0.553,
+      "step": 7653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995919155614306e-05,
+      "loss": 0.6245,
+      "step": 7654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995914461261533e-05,
+      "loss": 0.6074,
+      "step": 7655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995909764210812e-05,
+      "loss": 0.591,
+      "step": 7656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995905064462145e-05,
+      "loss": 0.5319,
+      "step": 7657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999590036201554e-05,
+      "loss": 0.5548,
+      "step": 7658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995895656870994e-05,
+      "loss": 0.6234,
+      "step": 7659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999589094902851e-05,
+      "loss": 0.5269,
+      "step": 7660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995886238488087e-05,
+      "loss": 0.5516,
+      "step": 7661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999588152524973e-05,
+      "loss": 0.54,
+      "step": 7662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995876809313436e-05,
+      "loss": 0.5615,
+      "step": 7663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995872090679207e-05,
+      "loss": 0.614,
+      "step": 7664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995867369347045e-05,
+      "loss": 0.5947,
+      "step": 7665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995862645316956e-05,
+      "loss": 0.6106,
+      "step": 7666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995857918588934e-05,
+      "loss": 0.6857,
+      "step": 7667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999585318916298e-05,
+      "loss": 0.5469,
+      "step": 7668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995848457039104e-05,
+      "loss": 0.5208,
+      "step": 7669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.99958437222173e-05,
+      "loss": 0.6389,
+      "step": 7670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999583898469757e-05,
+      "loss": 0.5758,
+      "step": 7671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995834244479918e-05,
+      "loss": 0.5824,
+      "step": 7672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995829501564344e-05,
+      "loss": 0.595,
+      "step": 7673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995824755950845e-05,
+      "loss": 0.6543,
+      "step": 7674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999582000763943e-05,
+      "loss": 0.5621,
+      "step": 7675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995815256630095e-05,
+      "loss": 0.599,
+      "step": 7676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995810502922844e-05,
+      "loss": 0.5974,
+      "step": 7677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995805746517675e-05,
+      "loss": 0.6584,
+      "step": 7678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999580098741459e-05,
+      "loss": 0.5404,
+      "step": 7679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995796225613597e-05,
+      "loss": 0.553,
+      "step": 7680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999579146111469e-05,
+      "loss": 0.6577,
+      "step": 7681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999578669391787e-05,
+      "loss": 0.543,
+      "step": 7682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999578192402314e-05,
+      "loss": 0.5451,
+      "step": 7683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995777151430504e-05,
+      "loss": 0.5416,
+      "step": 7684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999577237613996e-05,
+      "loss": 0.5349,
+      "step": 7685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999576759815151e-05,
+      "loss": 0.5494,
+      "step": 7686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995762817465157e-05,
+      "loss": 0.6421,
+      "step": 7687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.99957580340809e-05,
+      "loss": 0.545,
+      "step": 7688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999575324799874e-05,
+      "loss": 0.683,
+      "step": 7689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999574845921868e-05,
+      "loss": 0.5712,
+      "step": 7690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999574366774072e-05,
+      "loss": 0.508,
+      "step": 7691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995738873564865e-05,
+      "loss": 0.5197,
+      "step": 7692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995734076691114e-05,
+      "loss": 0.6284,
+      "step": 7693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999572927711946e-05,
+      "loss": 0.6529,
+      "step": 7694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999572447484992e-05,
+      "loss": 0.5452,
+      "step": 7695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995719669882485e-05,
+      "loss": 0.5581,
+      "step": 7696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999571486221716e-05,
+      "loss": 0.6704,
+      "step": 7697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999571005185394e-05,
+      "loss": 0.5346,
+      "step": 7698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995705238792834e-05,
+      "loss": 0.5404,
+      "step": 7699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995700423033843e-05,
+      "loss": 0.5544,
+      "step": 7700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999569560457696e-05,
+      "loss": 0.6304,
+      "step": 7701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995690783422198e-05,
+      "loss": 0.6007,
+      "step": 7702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999568595956955e-05,
+      "loss": 0.5595,
+      "step": 7703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995681133019017e-05,
+      "loss": 0.5968,
+      "step": 7704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999567630377061e-05,
+      "loss": 0.5126,
+      "step": 7705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995671471824317e-05,
+      "loss": 0.5195,
+      "step": 7706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995666637180148e-05,
+      "loss": 0.5805,
+      "step": 7707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995661799838097e-05,
+      "loss": 0.6752,
+      "step": 7708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995656959798174e-05,
+      "loss": 0.6365,
+      "step": 7709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995652117060378e-05,
+      "loss": 0.5689,
+      "step": 7710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995647271624707e-05,
+      "loss": 0.6503,
+      "step": 7711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995642423491165e-05,
+      "loss": 0.6334,
+      "step": 7712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999563757265975e-05,
+      "loss": 0.766,
+      "step": 7713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995632719130467e-05,
+      "loss": 0.5674,
+      "step": 7714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999562786290332e-05,
+      "loss": 0.5621,
+      "step": 7715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.99956230039783e-05,
+      "loss": 0.7428,
+      "step": 7716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995618142355418e-05,
+      "loss": 0.6913,
+      "step": 7717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999561327803467e-05,
+      "loss": 0.539,
+      "step": 7718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999560841101606e-05,
+      "loss": 0.683,
+      "step": 7719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999560354129959e-05,
+      "loss": 0.6242,
+      "step": 7720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999559866888526e-05,
+      "loss": 0.6079,
+      "step": 7721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999559379377307e-05,
+      "loss": 0.5372,
+      "step": 7722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999558891596302e-05,
+      "loss": 0.5464,
+      "step": 7723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995584035455117e-05,
+      "loss": 0.4973,
+      "step": 7724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999557915224936e-05,
+      "loss": 0.552,
+      "step": 7725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995574266345747e-05,
+      "loss": 0.6436,
+      "step": 7726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995569377744284e-05,
+      "loss": 0.582,
+      "step": 7727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999556448644497e-05,
+      "loss": 0.5836,
+      "step": 7728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995559592447803e-05,
+      "loss": 0.5733,
+      "step": 7729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999555469575279e-05,
+      "loss": 0.5091,
+      "step": 7730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995549796359932e-05,
+      "loss": 0.5959,
+      "step": 7731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995544894269226e-05,
+      "loss": 0.6502,
+      "step": 7732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995539989480675e-05,
+      "loss": 0.5689,
+      "step": 7733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995535081994284e-05,
+      "loss": 0.5544,
+      "step": 7734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999553017181005e-05,
+      "loss": 0.5369,
+      "step": 7735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9995525258927974e-05,
+      "loss": 0.6731,
+      "step": 7736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999552034334806e-05,
+      "loss": 0.6514,
+      "step": 7737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995515425070308e-05,
+      "loss": 0.5727,
+      "step": 7738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999551050409472e-05,
+      "loss": 0.5959,
+      "step": 7739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.99955055804213e-05,
+      "loss": 0.5465,
+      "step": 7740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995500654050045e-05,
+      "loss": 0.7336,
+      "step": 7741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995495724980953e-05,
+      "loss": 0.5989,
+      "step": 7742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995490793214034e-05,
+      "loss": 0.7606,
+      "step": 7743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995485858749287e-05,
+      "loss": 0.5971,
+      "step": 7744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995480921586707e-05,
+      "loss": 0.5364,
+      "step": 7745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995475981726304e-05,
+      "loss": 0.684,
+      "step": 7746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995471039168073e-05,
+      "loss": 0.5995,
+      "step": 7747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995466093912016e-05,
+      "loss": 0.6404,
+      "step": 7748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999546114595814e-05,
+      "loss": 0.5014,
+      "step": 7749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999545619530644e-05,
+      "loss": 0.6279,
+      "step": 7750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999545124195692e-05,
+      "loss": 0.6169,
+      "step": 7751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995446285909582e-05,
+      "loss": 0.4869,
+      "step": 7752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995441327164423e-05,
+      "loss": 0.5531,
+      "step": 7753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999543636572145e-05,
+      "loss": 0.6574,
+      "step": 7754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999543140158066e-05,
+      "loss": 0.5004,
+      "step": 7755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995426434742062e-05,
+      "loss": 0.5248,
+      "step": 7756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999542146520565e-05,
+      "loss": 0.524,
+      "step": 7757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995416492971422e-05,
+      "loss": 0.5282,
+      "step": 7758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999541151803939e-05,
+      "loss": 0.5538,
+      "step": 7759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995406540409546e-05,
+      "loss": 0.6291,
+      "step": 7760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995401560081896e-05,
+      "loss": 0.5486,
+      "step": 7761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995396577056443e-05,
+      "loss": 0.5794,
+      "step": 7762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995391591333183e-05,
+      "loss": 0.6793,
+      "step": 7763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999538660291212e-05,
+      "loss": 0.6182,
+      "step": 7764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995381611793253e-05,
+      "loss": 0.6119,
+      "step": 7765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995376617976593e-05,
+      "loss": 0.7037,
+      "step": 7766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995371621462126e-05,
+      "loss": 0.5045,
+      "step": 7767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995366622249867e-05,
+      "loss": 0.5989,
+      "step": 7768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999536162033981e-05,
+      "loss": 0.5894,
+      "step": 7769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995356615731957e-05,
+      "loss": 0.6016,
+      "step": 7770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999535160842631e-05,
+      "loss": 0.62,
+      "step": 7771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995346598422876e-05,
+      "loss": 0.5944,
+      "step": 7772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995341585721647e-05,
+      "loss": 0.5175,
+      "step": 7773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995336570322628e-05,
+      "loss": 0.634,
+      "step": 7774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995331552225827e-05,
+      "loss": 0.5271,
+      "step": 7775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999532653143123e-05,
+      "loss": 0.6233,
+      "step": 7776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995321507938854e-05,
+      "loss": 0.6083,
+      "step": 7777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995316481748693e-05,
+      "loss": 0.528,
+      "step": 7778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999531145286075e-05,
+      "loss": 0.5665,
+      "step": 7779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995306421275026e-05,
+      "loss": 0.6029,
+      "step": 7780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999530138699152e-05,
+      "loss": 0.675,
+      "step": 7781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995296350010233e-05,
+      "loss": 0.5781,
+      "step": 7782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995291310331174e-05,
+      "loss": 0.6162,
+      "step": 7783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995286267954336e-05,
+      "loss": 0.5714,
+      "step": 7784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995281222879725e-05,
+      "loss": 0.5798,
+      "step": 7785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995276175107337e-05,
+      "loss": 0.5857,
+      "step": 7786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995271124637184e-05,
+      "loss": 0.51,
+      "step": 7787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995266071469257e-05,
+      "loss": 0.5786,
+      "step": 7788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999526101560356e-05,
+      "loss": 0.5889,
+      "step": 7789
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995255957040096e-05,
+      "loss": 0.6324,
+      "step": 7790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995250895778868e-05,
+      "loss": 0.6222,
+      "step": 7791
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995245831819874e-05,
+      "loss": 0.5196,
+      "step": 7792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995240765163114e-05,
+      "loss": 0.5925,
+      "step": 7793
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995235695808595e-05,
+      "loss": 0.5679,
+      "step": 7794
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995230623756313e-05,
+      "loss": 0.5974,
+      "step": 7795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999522554900627e-05,
+      "loss": 0.6467,
+      "step": 7796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999522047155847e-05,
+      "loss": 0.6114,
+      "step": 7797
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995215391412918e-05,
+      "loss": 0.5592,
+      "step": 7798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995210308569605e-05,
+      "loss": 0.5177,
+      "step": 7799
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999520522302854e-05,
+      "loss": 0.7073,
+      "step": 7800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995200134789723e-05,
+      "loss": 0.5683,
+      "step": 7801
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995195043853153e-05,
+      "loss": 0.5787,
+      "step": 7802
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995189950218834e-05,
+      "loss": 0.5569,
+      "step": 7803
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995184853886766e-05,
+      "loss": 0.6452,
+      "step": 7804
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995179754856956e-05,
+      "loss": 0.5268,
+      "step": 7805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995174653129396e-05,
+      "loss": 0.6908,
+      "step": 7806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999516954870409e-05,
+      "loss": 0.5863,
+      "step": 7807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999516444158104e-05,
+      "loss": 0.5185,
+      "step": 7808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995159331760253e-05,
+      "loss": 0.5206,
+      "step": 7809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995154219241724e-05,
+      "loss": 0.592,
+      "step": 7810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995149104025456e-05,
+      "loss": 0.6174,
+      "step": 7811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999514398611145e-05,
+      "loss": 0.6409,
+      "step": 7812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995138865499707e-05,
+      "loss": 0.6427,
+      "step": 7813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995133742190233e-05,
+      "loss": 0.5476,
+      "step": 7814
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995128616183022e-05,
+      "loss": 0.5926,
+      "step": 7815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999512348747808e-05,
+      "loss": 0.5529,
+      "step": 7816
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995118356075406e-05,
+      "loss": 0.4942,
+      "step": 7817
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995113221975006e-05,
+      "loss": 0.5653,
+      "step": 7818
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995108085176877e-05,
+      "loss": 0.6035,
+      "step": 7819
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995102945681023e-05,
+      "loss": 0.5675,
+      "step": 7820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995097803487444e-05,
+      "loss": 0.5029,
+      "step": 7821
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999509265859614e-05,
+      "loss": 0.5204,
+      "step": 7822
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995087511007112e-05,
+      "loss": 0.5897,
+      "step": 7823
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995082360720363e-05,
+      "loss": 0.5793,
+      "step": 7824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.99950772077359e-05,
+      "loss": 0.6475,
+      "step": 7825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995072052053713e-05,
+      "loss": 0.6309,
+      "step": 7826
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995066893673808e-05,
+      "loss": 0.5614,
+      "step": 7827
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995061732596195e-05,
+      "loss": 0.5352,
+      "step": 7828
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995056568820863e-05,
+      "loss": 0.5836,
+      "step": 7829
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999505140234782e-05,
+      "loss": 0.679,
+      "step": 7830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995046233177064e-05,
+      "loss": 0.6574,
+      "step": 7831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995041061308604e-05,
+      "loss": 0.6538,
+      "step": 7832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999503588674243e-05,
+      "loss": 0.6503,
+      "step": 7833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999503070947855e-05,
+      "loss": 0.5834,
+      "step": 7834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9995025529516968e-05,
+      "loss": 0.6622,
+      "step": 7835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999502034685768e-05,
+      "loss": 0.572,
+      "step": 7836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9995015161500687e-05,
+      "loss": 0.5449,
+      "step": 7837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9995009973445993e-05,
+      "loss": 0.6197,
+      "step": 7838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99950047826936e-05,
+      "loss": 0.6718,
+      "step": 7839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999499958924351e-05,
+      "loss": 0.6715,
+      "step": 7840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999499439309572e-05,
+      "loss": 0.6106,
+      "step": 7841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999498919425024e-05,
+      "loss": 0.6619,
+      "step": 7842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999498399270706e-05,
+      "loss": 0.5975,
+      "step": 7843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999497878846619e-05,
+      "loss": 0.4708,
+      "step": 7844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994973581527624e-05,
+      "loss": 0.5337,
+      "step": 7845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994968371891374e-05,
+      "loss": 0.6317,
+      "step": 7846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994963159557432e-05,
+      "loss": 0.4998,
+      "step": 7847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994957944525802e-05,
+      "loss": 0.675,
+      "step": 7848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994952726796488e-05,
+      "loss": 0.5604,
+      "step": 7849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999494750636949e-05,
+      "loss": 0.6418,
+      "step": 7850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994942283244808e-05,
+      "loss": 0.5155,
+      "step": 7851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994937057422442e-05,
+      "loss": 0.5603,
+      "step": 7852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99949318289024e-05,
+      "loss": 0.5962,
+      "step": 7853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994926597684677e-05,
+      "loss": 0.495,
+      "step": 7854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994921363769278e-05,
+      "loss": 0.6294,
+      "step": 7855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99949161271562e-05,
+      "loss": 0.5736,
+      "step": 7856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999491088784545e-05,
+      "loss": 0.619,
+      "step": 7857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994905645837023e-05,
+      "loss": 0.5767,
+      "step": 7858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999490040113093e-05,
+      "loss": 0.7013,
+      "step": 7859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994895153727163e-05,
+      "loss": 0.6079,
+      "step": 7860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999488990362573e-05,
+      "loss": 0.5328,
+      "step": 7861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999488465082663e-05,
+      "loss": 0.6401,
+      "step": 7862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999487939532986e-05,
+      "loss": 0.5121,
+      "step": 7863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994874137135426e-05,
+      "loss": 0.6483,
+      "step": 7864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999486887624333e-05,
+      "loss": 0.5694,
+      "step": 7865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999486361265357e-05,
+      "loss": 0.7095,
+      "step": 7866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999485834636615e-05,
+      "loss": 0.64,
+      "step": 7867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994853077381073e-05,
+      "loss": 0.669,
+      "step": 7868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994847805698338e-05,
+      "loss": 0.5665,
+      "step": 7869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994842531317945e-05,
+      "loss": 0.5521,
+      "step": 7870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994837254239902e-05,
+      "loss": 0.5416,
+      "step": 7871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99948319744642e-05,
+      "loss": 0.561,
+      "step": 7872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999482669199085e-05,
+      "loss": 0.6136,
+      "step": 7873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999482140681985e-05,
+      "loss": 0.5673,
+      "step": 7874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994816118951196e-05,
+      "loss": 0.608,
+      "step": 7875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994810828384898e-05,
+      "loss": 0.5368,
+      "step": 7876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994805535120953e-05,
+      "loss": 0.5798,
+      "step": 7877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994800239159363e-05,
+      "loss": 0.5259,
+      "step": 7878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999479494050013e-05,
+      "loss": 0.6281,
+      "step": 7879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994789639143256e-05,
+      "loss": 0.7129,
+      "step": 7880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999478433508874e-05,
+      "loss": 0.5648,
+      "step": 7881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999477902833659e-05,
+      "loss": 0.5008,
+      "step": 7882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994773718886796e-05,
+      "loss": 0.6561,
+      "step": 7883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999476840673937e-05,
+      "loss": 0.6287,
+      "step": 7884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994763091894305e-05,
+      "loss": 0.5603,
+      "step": 7885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999475777435161e-05,
+      "loss": 0.5744,
+      "step": 7886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994752454111282e-05,
+      "loss": 0.5791,
+      "step": 7887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994747131173327e-05,
+      "loss": 0.5928,
+      "step": 7888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999474180553774e-05,
+      "loss": 0.6871,
+      "step": 7889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994736477204527e-05,
+      "loss": 0.4922,
+      "step": 7890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994731146173684e-05,
+      "loss": 0.6674,
+      "step": 7891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999472581244522e-05,
+      "loss": 0.566,
+      "step": 7892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999472047601913e-05,
+      "loss": 0.5477,
+      "step": 7893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999471513689542e-05,
+      "loss": 0.6774,
+      "step": 7894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999470979507409e-05,
+      "loss": 0.593,
+      "step": 7895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999470445055514e-05,
+      "loss": 0.602,
+      "step": 7896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994699103338573e-05,
+      "loss": 0.6615,
+      "step": 7897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999469375342439e-05,
+      "loss": 0.6894,
+      "step": 7898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994688400812593e-05,
+      "loss": 0.5548,
+      "step": 7899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994683045503183e-05,
+      "loss": 0.5809,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994677687496162e-05,
+      "loss": 0.6375,
+      "step": 7901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999467232679153e-05,
+      "loss": 0.5842,
+      "step": 7902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994666963389288e-05,
+      "loss": 0.4714,
+      "step": 7903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999466159728944e-05,
+      "loss": 0.5519,
+      "step": 7904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994656228491986e-05,
+      "loss": 0.5622,
+      "step": 7905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994650856996928e-05,
+      "loss": 0.5785,
+      "step": 7906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994645482804263e-05,
+      "loss": 0.7467,
+      "step": 7907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994640105914002e-05,
+      "loss": 0.6117,
+      "step": 7908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994634726326137e-05,
+      "loss": 0.7025,
+      "step": 7909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994629344040676e-05,
+      "loss": 0.4987,
+      "step": 7910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994623959057614e-05,
+      "loss": 0.7396,
+      "step": 7911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994618571376963e-05,
+      "loss": 0.5987,
+      "step": 7912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994613180998712e-05,
+      "loss": 0.5806,
+      "step": 7913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994607787922867e-05,
+      "loss": 0.6183,
+      "step": 7914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994602392149436e-05,
+      "loss": 0.5353,
+      "step": 7915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994596993678415e-05,
+      "loss": 0.5639,
+      "step": 7916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99945915925098e-05,
+      "loss": 0.501,
+      "step": 7917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99945861886436e-05,
+      "loss": 0.6061,
+      "step": 7918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994580782079817e-05,
+      "loss": 0.5237,
+      "step": 7919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999457537281845e-05,
+      "loss": 0.6413,
+      "step": 7920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994569960859496e-05,
+      "loss": 0.5858,
+      "step": 7921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994564546202967e-05,
+      "loss": 0.5451,
+      "step": 7922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994559128848854e-05,
+      "loss": 0.6516,
+      "step": 7923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994553708797165e-05,
+      "loss": 0.5544,
+      "step": 7924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99945482860479e-05,
+      "loss": 0.6325,
+      "step": 7925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994542860601058e-05,
+      "loss": 0.6095,
+      "step": 7926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999453743245664e-05,
+      "loss": 0.5601,
+      "step": 7927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994532001614652e-05,
+      "loss": 0.5286,
+      "step": 7928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999452656807509e-05,
+      "loss": 0.4772,
+      "step": 7929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994521131837965e-05,
+      "loss": 0.6143,
+      "step": 7930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994515692903265e-05,
+      "loss": 0.6055,
+      "step": 7931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994510251271003e-05,
+      "loss": 0.567,
+      "step": 7932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994504806941174e-05,
+      "loss": 0.6199,
+      "step": 7933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994499359913783e-05,
+      "loss": 0.6048,
+      "step": 7934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999449391018883e-05,
+      "loss": 0.6509,
+      "step": 7935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994488457766315e-05,
+      "loss": 0.5841,
+      "step": 7936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994483002646243e-05,
+      "loss": 0.5952,
+      "step": 7937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994477544828614e-05,
+      "loss": 0.5132,
+      "step": 7938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994472084313422e-05,
+      "loss": 0.6355,
+      "step": 7939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999446662110068e-05,
+      "loss": 0.5705,
+      "step": 7940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994461155190385e-05,
+      "loss": 0.5388,
+      "step": 7941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994455686582535e-05,
+      "loss": 0.6248,
+      "step": 7942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999445021527714e-05,
+      "loss": 0.509,
+      "step": 7943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994444741274192e-05,
+      "loss": 0.6112,
+      "step": 7944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.99944392645737e-05,
+      "loss": 0.5777,
+      "step": 7945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999443378517566e-05,
+      "loss": 0.5434,
+      "step": 7946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994428303080077e-05,
+      "loss": 0.6273,
+      "step": 7947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999442281828695e-05,
+      "loss": 0.539,
+      "step": 7948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994417330796282e-05,
+      "loss": 0.5992,
+      "step": 7949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994411840608073e-05,
+      "loss": 0.6338,
+      "step": 7950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994406347722325e-05,
+      "loss": 0.5133,
+      "step": 7951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999440085213904e-05,
+      "loss": 0.5149,
+      "step": 7952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999439535385822e-05,
+      "loss": 0.6375,
+      "step": 7953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999438985287987e-05,
+      "loss": 0.6807,
+      "step": 7954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994384349203982e-05,
+      "loss": 0.5955,
+      "step": 7955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994378842830565e-05,
+      "loss": 0.5175,
+      "step": 7956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999437333375962e-05,
+      "loss": 0.5474,
+      "step": 7957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994367821991146e-05,
+      "loss": 0.6001,
+      "step": 7958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994362307525143e-05,
+      "loss": 0.5679,
+      "step": 7959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994356790361618e-05,
+      "loss": 0.6049,
+      "step": 7960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994351270500567e-05,
+      "loss": 0.6141,
+      "step": 7961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994345747941994e-05,
+      "loss": 0.547,
+      "step": 7962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994340222685902e-05,
+      "loss": 0.5913,
+      "step": 7963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994334694732292e-05,
+      "loss": 0.6643,
+      "step": 7964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999432916408116e-05,
+      "loss": 0.6755,
+      "step": 7965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994323630732515e-05,
+      "loss": 0.5005,
+      "step": 7966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994318094686352e-05,
+      "loss": 0.5793,
+      "step": 7967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999431255594268e-05,
+      "loss": 0.5993,
+      "step": 7968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994307014501494e-05,
+      "loss": 0.6715,
+      "step": 7969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994301470362802e-05,
+      "loss": 0.5709,
+      "step": 7970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994295923526598e-05,
+      "loss": 0.5992,
+      "step": 7971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994290373992885e-05,
+      "loss": 0.6604,
+      "step": 7972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999428482176167e-05,
+      "loss": 0.5466,
+      "step": 7973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994279266832948e-05,
+      "loss": 0.64,
+      "step": 7974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994273709206723e-05,
+      "loss": 0.5854,
+      "step": 7975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994268148882996e-05,
+      "loss": 0.584,
+      "step": 7976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994262585861775e-05,
+      "loss": 0.5665,
+      "step": 7977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999425702014305e-05,
+      "loss": 0.5544,
+      "step": 7978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999425145172683e-05,
+      "loss": 0.5469,
+      "step": 7979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994245880613117e-05,
+      "loss": 0.6686,
+      "step": 7980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999424030680191e-05,
+      "loss": 0.4911,
+      "step": 7981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999423473029321e-05,
+      "loss": 0.6174,
+      "step": 7982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994229151087017e-05,
+      "loss": 0.6157,
+      "step": 7983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994223569183337e-05,
+      "loss": 0.5842,
+      "step": 7984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999421798458217e-05,
+      "loss": 0.6268,
+      "step": 7985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994212397283513e-05,
+      "loss": 0.5174,
+      "step": 7986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994206807287372e-05,
+      "loss": 0.5685,
+      "step": 7987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994201214593753e-05,
+      "loss": 0.6117,
+      "step": 7988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999419561920265e-05,
+      "loss": 0.5587,
+      "step": 7989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994190021114063e-05,
+      "loss": 0.5696,
+      "step": 7990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994184420328003e-05,
+      "loss": 0.5224,
+      "step": 7991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999417881684446e-05,
+      "loss": 0.5347,
+      "step": 7992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999417321066345e-05,
+      "loss": 0.6549,
+      "step": 7993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999416760178496e-05,
+      "loss": 0.5582,
+      "step": 7994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994161990208996e-05,
+      "loss": 0.599,
+      "step": 7995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994156375935565e-05,
+      "loss": 0.5924,
+      "step": 7996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999415075896466e-05,
+      "loss": 0.674,
+      "step": 7997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999414513929629e-05,
+      "loss": 0.5287,
+      "step": 7998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994139516930454e-05,
+      "loss": 0.6336,
+      "step": 7999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994133891867152e-05,
+      "loss": 0.5551,
+      "step": 8000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994128264106386e-05,
+      "loss": 0.6619,
+      "step": 8001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994122633648162e-05,
+      "loss": 0.536,
+      "step": 8002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994117000492474e-05,
+      "loss": 0.5596,
+      "step": 8003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994111364639324e-05,
+      "loss": 0.5035,
+      "step": 8004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994105726088724e-05,
+      "loss": 0.5686,
+      "step": 8005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994100084840665e-05,
+      "loss": 0.6808,
+      "step": 8006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999409444089515e-05,
+      "loss": 0.6779,
+      "step": 8007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994088794252186e-05,
+      "loss": 0.6125,
+      "step": 8008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994083144911764e-05,
+      "loss": 0.6191,
+      "step": 8009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.99940774928739e-05,
+      "loss": 0.6219,
+      "step": 8010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994071838138582e-05,
+      "loss": 0.6005,
+      "step": 8011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999406618070582e-05,
+      "loss": 0.5755,
+      "step": 8012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999406052057561e-05,
+      "loss": 0.5662,
+      "step": 8013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999405485774796e-05,
+      "loss": 0.5427,
+      "step": 8014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994049192222868e-05,
+      "loss": 0.6434,
+      "step": 8015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999404352400033e-05,
+      "loss": 0.5596,
+      "step": 8016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994037853080356e-05,
+      "loss": 0.6121,
+      "step": 8017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994032179462947e-05,
+      "loss": 0.5386,
+      "step": 8018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.99940265031481e-05,
+      "loss": 0.5152,
+      "step": 8019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994020824135817e-05,
+      "loss": 0.6453,
+      "step": 8020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994015142426103e-05,
+      "loss": 0.5636,
+      "step": 8021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994009458018958e-05,
+      "loss": 0.5848,
+      "step": 8022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9994003770914382e-05,
+      "loss": 0.6607,
+      "step": 8023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9993998081112376e-05,
+      "loss": 0.7163,
+      "step": 8024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999399238861295e-05,
+      "loss": 0.623,
+      "step": 8025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9993986693416092e-05,
+      "loss": 0.6361,
+      "step": 8026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999398099552181e-05,
+      "loss": 0.6464,
+      "step": 8027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999397529493011e-05,
+      "loss": 0.6289,
+      "step": 8028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9993969591640988e-05,
+      "loss": 0.4955,
+      "step": 8029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9993963885654445e-05,
+      "loss": 0.6095,
+      "step": 8030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9993958176970486e-05,
+      "loss": 0.612,
+      "step": 8031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999395246558911e-05,
+      "loss": 0.4752,
+      "step": 8032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9993946751510322e-05,
+      "loss": 0.5736,
+      "step": 8033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993941034734118e-05,
+      "loss": 0.568,
+      "step": 8034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993935315260504e-05,
+      "loss": 0.5462,
+      "step": 8035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999392959308948e-05,
+      "loss": 0.5315,
+      "step": 8036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999392386822105e-05,
+      "loss": 0.6151,
+      "step": 8037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999391814065521e-05,
+      "loss": 0.529,
+      "step": 8038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993912410391964e-05,
+      "loss": 0.585,
+      "step": 8039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993906677431315e-05,
+      "loss": 0.5781,
+      "step": 8040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999390094177327e-05,
+      "loss": 0.5936,
+      "step": 8041
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993895203417817e-05,
+      "loss": 0.5562,
+      "step": 8042
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993889462364967e-05,
+      "loss": 0.5699,
+      "step": 8043
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993883718614718e-05,
+      "loss": 0.5924,
+      "step": 8044
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993877972167075e-05,
+      "loss": 0.6083,
+      "step": 8045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999387222302204e-05,
+      "loss": 0.6258,
+      "step": 8046
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999386647117961e-05,
+      "loss": 0.5643,
+      "step": 8047
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993860716639786e-05,
+      "loss": 0.6501,
+      "step": 8048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993854959402577e-05,
+      "loss": 0.5584,
+      "step": 8049
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993849199467977e-05,
+      "loss": 0.5795,
+      "step": 8050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999384343683599e-05,
+      "loss": 0.7254,
+      "step": 8051
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999383767150662e-05,
+      "loss": 0.6295,
+      "step": 8052
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993831903479863e-05,
+      "loss": 0.6207,
+      "step": 8053
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999382613275573e-05,
+      "loss": 0.6375,
+      "step": 8054
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993820359334212e-05,
+      "loss": 0.5656,
+      "step": 8055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993814583215317e-05,
+      "loss": 0.5685,
+      "step": 8056
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999380880439904e-05,
+      "loss": 0.6321,
+      "step": 8057
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993803022885393e-05,
+      "loss": 0.5179,
+      "step": 8058
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999379723867437e-05,
+      "loss": 0.5892,
+      "step": 8059
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993791451765975e-05,
+      "loss": 0.4522,
+      "step": 8060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993785662160208e-05,
+      "loss": 0.5468,
+      "step": 8061
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999377986985707e-05,
+      "loss": 0.5604,
+      "step": 8062
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993774074856568e-05,
+      "loss": 0.5488,
+      "step": 8063
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.99937682771587e-05,
+      "loss": 0.6509,
+      "step": 8064
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993762476763463e-05,
+      "loss": 0.5708,
+      "step": 8065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993756673670864e-05,
+      "loss": 0.5683,
+      "step": 8066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993750867880903e-05,
+      "loss": 0.4924,
+      "step": 8067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993745059393585e-05,
+      "loss": 0.5599,
+      "step": 8068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993739248208905e-05,
+      "loss": 0.6068,
+      "step": 8069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993733434326872e-05,
+      "loss": 0.5806,
+      "step": 8070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999372761774748e-05,
+      "loss": 0.6745,
+      "step": 8071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993721798470735e-05,
+      "loss": 0.6088,
+      "step": 8072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993715976496637e-05,
+      "loss": 0.6038,
+      "step": 8073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999371015182519e-05,
+      "loss": 0.5851,
+      "step": 8074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993704324456393e-05,
+      "loss": 0.6723,
+      "step": 8075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999369849439025e-05,
+      "loss": 0.5527,
+      "step": 8076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999369266162676e-05,
+      "loss": 0.5902,
+      "step": 8077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993686826165924e-05,
+      "loss": 0.5632,
+      "step": 8078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993680988007748e-05,
+      "loss": 0.5565,
+      "step": 8079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993675147152228e-05,
+      "loss": 0.5805,
+      "step": 8080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993669303599373e-05,
+      "loss": 0.6072,
+      "step": 8081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993663457349175e-05,
+      "loss": 0.5172,
+      "step": 8082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993657608401645e-05,
+      "loss": 0.5608,
+      "step": 8083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993651756756776e-05,
+      "loss": 0.4947,
+      "step": 8084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993645902414575e-05,
+      "loss": 0.6546,
+      "step": 8085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993640045375044e-05,
+      "loss": 0.4822,
+      "step": 8086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999363418563818e-05,
+      "loss": 0.6054,
+      "step": 8087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999362832320399e-05,
+      "loss": 0.5897,
+      "step": 8088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999362245807247e-05,
+      "loss": 0.6163,
+      "step": 8089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999361659024363e-05,
+      "loss": 0.5505,
+      "step": 8090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999361071971746e-05,
+      "loss": 0.6617,
+      "step": 8091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999360484649397e-05,
+      "loss": 0.5292,
+      "step": 8092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999359897057316e-05,
+      "loss": 0.5317,
+      "step": 8093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993593091955027e-05,
+      "loss": 0.5584,
+      "step": 8094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993587210639583e-05,
+      "loss": 0.6253,
+      "step": 8095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999358132662682e-05,
+      "loss": 0.7446,
+      "step": 8096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993575439916745e-05,
+      "loss": 0.5622,
+      "step": 8097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999356955050935e-05,
+      "loss": 0.5973,
+      "step": 8098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993563658404653e-05,
+      "loss": 0.558,
+      "step": 8099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999355776360264e-05,
+      "loss": 0.565,
+      "step": 8100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993551866103322e-05,
+      "loss": 0.5168,
+      "step": 8101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993545965906697e-05,
+      "loss": 0.6261,
+      "step": 8102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999354006301277e-05,
+      "loss": 0.6386,
+      "step": 8103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993534157421535e-05,
+      "loss": 0.5792,
+      "step": 8104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993528249133e-05,
+      "loss": 0.6199,
+      "step": 8105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993522338147168e-05,
+      "loss": 0.5708,
+      "step": 8106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993516424464034e-05,
+      "loss": 0.6218,
+      "step": 8107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.99935105080836e-05,
+      "loss": 0.6344,
+      "step": 8108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993504589005876e-05,
+      "loss": 0.503,
+      "step": 8109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999349866723086e-05,
+      "loss": 0.5463,
+      "step": 8110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999349274275855e-05,
+      "loss": 0.5761,
+      "step": 8111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993486815588946e-05,
+      "loss": 0.6622,
+      "step": 8112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993480885722058e-05,
+      "loss": 0.6948,
+      "step": 8113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993474953157885e-05,
+      "loss": 0.6245,
+      "step": 8114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999346901789642e-05,
+      "loss": 0.5673,
+      "step": 8115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999346307993767e-05,
+      "loss": 0.6099,
+      "step": 8116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993457139281642e-05,
+      "loss": 0.5564,
+      "step": 8117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993451195928336e-05,
+      "loss": 0.5389,
+      "step": 8118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993445249877746e-05,
+      "loss": 0.558,
+      "step": 8119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993439301129876e-05,
+      "loss": 0.5386,
+      "step": 8120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993433349684736e-05,
+      "loss": 0.6044,
+      "step": 8121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993427395542317e-05,
+      "loss": 0.5642,
+      "step": 8122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999342143870263e-05,
+      "loss": 0.5445,
+      "step": 8123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999341547916567e-05,
+      "loss": 0.4977,
+      "step": 8124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993409516931436e-05,
+      "loss": 0.6565,
+      "step": 8125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993403551999942e-05,
+      "loss": 0.5814,
+      "step": 8126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993397584371177e-05,
+      "loss": 0.5871,
+      "step": 8127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993391614045147e-05,
+      "loss": 0.558,
+      "step": 8128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993385641021856e-05,
+      "loss": 0.582,
+      "step": 8129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.99933796653013e-05,
+      "loss": 0.5826,
+      "step": 8130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993373686883486e-05,
+      "loss": 0.5367,
+      "step": 8131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9993367705768415e-05,
+      "loss": 0.6037,
+      "step": 8132
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993361721956085e-05,
+      "loss": 0.6265,
+      "step": 8133
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.99933557354465e-05,
+      "loss": 0.6005,
+      "step": 8134
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993349746239663e-05,
+      "loss": 0.545,
+      "step": 8135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993343754335574e-05,
+      "loss": 0.5377,
+      "step": 8136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993337759734234e-05,
+      "loss": 0.6521,
+      "step": 8137
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999333176243565e-05,
+      "loss": 0.6689,
+      "step": 8138
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993325762439813e-05,
+      "loss": 0.5197,
+      "step": 8139
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993319759746733e-05,
+      "loss": 0.5765,
+      "step": 8140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999331375435641e-05,
+      "loss": 0.6212,
+      "step": 8141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993307746268844e-05,
+      "loss": 0.5805,
+      "step": 8142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993301735484038e-05,
+      "loss": 0.4574,
+      "step": 8143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993295722001994e-05,
+      "loss": 0.5459,
+      "step": 8144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999328970582271e-05,
+      "loss": 0.5638,
+      "step": 8145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999328368694619e-05,
+      "loss": 0.7524,
+      "step": 8146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993277665372438e-05,
+      "loss": 0.5254,
+      "step": 8147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993271641101455e-05,
+      "loss": 0.6567,
+      "step": 8148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999326561413324e-05,
+      "loss": 0.6232,
+      "step": 8149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993259584467796e-05,
+      "loss": 0.674,
+      "step": 8150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993253552105127e-05,
+      "loss": 0.6,
+      "step": 8151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993247517045227e-05,
+      "loss": 0.6007,
+      "step": 8152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993241479288104e-05,
+      "loss": 0.518,
+      "step": 8153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999323543883376e-05,
+      "loss": 0.6674,
+      "step": 8154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993229395682198e-05,
+      "loss": 0.6192,
+      "step": 8155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999322334983341e-05,
+      "loss": 0.6043,
+      "step": 8156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993217301287407e-05,
+      "loss": 0.543,
+      "step": 8157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999321125004419e-05,
+      "loss": 0.6478,
+      "step": 8158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993205196103757e-05,
+      "loss": 0.5539,
+      "step": 8159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999319913946611e-05,
+      "loss": 0.4688,
+      "step": 8160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993193080131256e-05,
+      "loss": 0.6095,
+      "step": 8161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993187018099188e-05,
+      "loss": 0.507,
+      "step": 8162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993180953369916e-05,
+      "loss": 0.612,
+      "step": 8163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993174885943434e-05,
+      "loss": 0.6172,
+      "step": 8164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999316881581975e-05,
+      "loss": 0.5452,
+      "step": 8165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999316274299886e-05,
+      "loss": 0.616,
+      "step": 8166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999315666748077e-05,
+      "loss": 0.5765,
+      "step": 8167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993150589265484e-05,
+      "loss": 0.6111,
+      "step": 8168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993144508352994e-05,
+      "loss": 0.5679,
+      "step": 8169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993138424743313e-05,
+      "loss": 0.6033,
+      "step": 8170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993132338436433e-05,
+      "loss": 0.6002,
+      "step": 8171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993126249432363e-05,
+      "loss": 0.5917,
+      "step": 8172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.99931201577311e-05,
+      "loss": 0.4864,
+      "step": 8173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993114063332645e-05,
+      "loss": 0.7662,
+      "step": 8174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993107966237004e-05,
+      "loss": 0.5537,
+      "step": 8175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993101866444177e-05,
+      "loss": 0.4991,
+      "step": 8176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993095763954163e-05,
+      "loss": 0.526,
+      "step": 8177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999308965876697e-05,
+      "loss": 0.5815,
+      "step": 8178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993083550882592e-05,
+      "loss": 0.5246,
+      "step": 8179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993077440301035e-05,
+      "loss": 0.4913,
+      "step": 8180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993071327022298e-05,
+      "loss": 0.5612,
+      "step": 8181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993065211046385e-05,
+      "loss": 0.5449,
+      "step": 8182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993059092373298e-05,
+      "loss": 0.5562,
+      "step": 8183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993052971003038e-05,
+      "loss": 0.5707,
+      "step": 8184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993046846935605e-05,
+      "loss": 0.7198,
+      "step": 8185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993040720171003e-05,
+      "loss": 0.5553,
+      "step": 8186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993034590709234e-05,
+      "loss": 0.537,
+      "step": 8187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993028458550295e-05,
+      "loss": 0.5214,
+      "step": 8188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993022323694193e-05,
+      "loss": 0.6057,
+      "step": 8189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993016186140925e-05,
+      "loss": 0.6189,
+      "step": 8190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9993010045890497e-05,
+      "loss": 0.6112,
+      "step": 8191
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999300390294291e-05,
+      "loss": 0.6403,
+      "step": 8192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992997757298163e-05,
+      "loss": 0.5384,
+      "step": 8193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999299160895626e-05,
+      "loss": 0.632,
+      "step": 8194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992985457917204e-05,
+      "loss": 0.679,
+      "step": 8195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999297930418099e-05,
+      "loss": 0.5711,
+      "step": 8196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992973147747628e-05,
+      "loss": 0.6144,
+      "step": 8197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992966988617113e-05,
+      "loss": 0.552,
+      "step": 8198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992960826789453e-05,
+      "loss": 0.4909,
+      "step": 8199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992954662264643e-05,
+      "loss": 0.6523,
+      "step": 8200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992948495042688e-05,
+      "loss": 0.6119,
+      "step": 8201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999294232512359e-05,
+      "loss": 0.5527,
+      "step": 8202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999293615250735e-05,
+      "loss": 0.5596,
+      "step": 8203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999292997719397e-05,
+      "loss": 0.5928,
+      "step": 8204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999292379918345e-05,
+      "loss": 0.6127,
+      "step": 8205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992917618475797e-05,
+      "loss": 0.6321,
+      "step": 8206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992911435071006e-05,
+      "loss": 0.5057,
+      "step": 8207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992905248969082e-05,
+      "loss": 0.5027,
+      "step": 8208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992899060170026e-05,
+      "loss": 0.6076,
+      "step": 8209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992892868673837e-05,
+      "loss": 0.6079,
+      "step": 8210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992886674480526e-05,
+      "loss": 0.5759,
+      "step": 8211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992880477590083e-05,
+      "loss": 0.4815,
+      "step": 8212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992874278002517e-05,
+      "loss": 0.4672,
+      "step": 8213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992868075717826e-05,
+      "loss": 0.5516,
+      "step": 8214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992861870736015e-05,
+      "loss": 0.5059,
+      "step": 8215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992855663057083e-05,
+      "loss": 0.5684,
+      "step": 8216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992849452681034e-05,
+      "loss": 0.5773,
+      "step": 8217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992843239607867e-05,
+      "loss": 0.6337,
+      "step": 8218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992837023837585e-05,
+      "loss": 0.5463,
+      "step": 8219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999283080537019e-05,
+      "loss": 0.6074,
+      "step": 8220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999282458420568e-05,
+      "loss": 0.5833,
+      "step": 8221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992818360344064e-05,
+      "loss": 0.6055,
+      "step": 8222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992812133785336e-05,
+      "loss": 0.5213,
+      "step": 8223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992805904529503e-05,
+      "loss": 0.5318,
+      "step": 8224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992799672576568e-05,
+      "loss": 0.5332,
+      "step": 8225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992793437926527e-05,
+      "loss": 0.5411,
+      "step": 8226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992787200579384e-05,
+      "loss": 0.6093,
+      "step": 8227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999278096053514e-05,
+      "loss": 0.5947,
+      "step": 8228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.99927747177938e-05,
+      "loss": 0.6229,
+      "step": 8229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9992768472355365e-05,
+      "loss": 0.4878,
+      "step": 8230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992762224219832e-05,
+      "loss": 0.5497,
+      "step": 8231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992755973387206e-05,
+      "loss": 0.5366,
+      "step": 8232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992749719857486e-05,
+      "loss": 0.582,
+      "step": 8233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999274346363068e-05,
+      "loss": 0.5912,
+      "step": 8234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992737204706786e-05,
+      "loss": 0.5179,
+      "step": 8235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992730943085804e-05,
+      "loss": 0.6631,
+      "step": 8236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999272467876774e-05,
+      "loss": 0.5364,
+      "step": 8237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992718411752587e-05,
+      "loss": 0.6315,
+      "step": 8238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999271214204036e-05,
+      "loss": 0.6241,
+      "step": 8239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992705869631047e-05,
+      "loss": 0.6654,
+      "step": 8240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999269959452466e-05,
+      "loss": 0.5615,
+      "step": 8241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999269331672119e-05,
+      "loss": 0.5734,
+      "step": 8242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992687036220654e-05,
+      "loss": 0.661,
+      "step": 8243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999268075302304e-05,
+      "loss": 0.5796,
+      "step": 8244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999267446712836e-05,
+      "loss": 0.6556,
+      "step": 8245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992668178536603e-05,
+      "loss": 0.5553,
+      "step": 8246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999266188724778e-05,
+      "loss": 0.6291,
+      "step": 8247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992655593261893e-05,
+      "loss": 0.6156,
+      "step": 8248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992649296578944e-05,
+      "loss": 0.564,
+      "step": 8249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992642997198926e-05,
+      "loss": 0.5993,
+      "step": 8250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992636695121855e-05,
+      "loss": 0.4802,
+      "step": 8251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992630390347718e-05,
+      "loss": 0.6347,
+      "step": 8252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992624082876527e-05,
+      "loss": 0.57,
+      "step": 8253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992617772708275e-05,
+      "loss": 0.5422,
+      "step": 8254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992611459842975e-05,
+      "loss": 0.6775,
+      "step": 8255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992605144280617e-05,
+      "loss": 0.5677,
+      "step": 8256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999259882602121e-05,
+      "loss": 0.5646,
+      "step": 8257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992592505064757e-05,
+      "loss": 0.6295,
+      "step": 8258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992586181411253e-05,
+      "loss": 0.5241,
+      "step": 8259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992579855060707e-05,
+      "loss": 0.5668,
+      "step": 8260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992573526013113e-05,
+      "loss": 0.6087,
+      "step": 8261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999256719426848e-05,
+      "loss": 0.6027,
+      "step": 8262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.99925608598268e-05,
+      "loss": 0.5823,
+      "step": 8263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992554522688087e-05,
+      "loss": 0.5975,
+      "step": 8264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992548182852336e-05,
+      "loss": 0.587,
+      "step": 8265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992541840319548e-05,
+      "loss": 0.5445,
+      "step": 8266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999253549508973e-05,
+      "loss": 0.5677,
+      "step": 8267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992529147162876e-05,
+      "loss": 0.5806,
+      "step": 8268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999252279653899e-05,
+      "loss": 0.5258,
+      "step": 8269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992516443218083e-05,
+      "loss": 0.5583,
+      "step": 8270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992510087200144e-05,
+      "loss": 0.538,
+      "step": 8271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999250372848518e-05,
+      "loss": 0.5834,
+      "step": 8272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999249736707319e-05,
+      "loss": 0.5793,
+      "step": 8273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992491002964182e-05,
+      "loss": 0.585,
+      "step": 8274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992484636158154e-05,
+      "loss": 0.5772,
+      "step": 8275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992478266655105e-05,
+      "loss": 0.487,
+      "step": 8276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999247189445504e-05,
+      "loss": 0.6453,
+      "step": 8277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999246551955796e-05,
+      "loss": 0.6046,
+      "step": 8278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999245914196387e-05,
+      "loss": 0.5383,
+      "step": 8279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992452761672766e-05,
+      "loss": 0.5576,
+      "step": 8280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992446378684654e-05,
+      "loss": 0.5695,
+      "step": 8281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999243999299953e-05,
+      "loss": 0.5915,
+      "step": 8282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992433604617405e-05,
+      "loss": 0.6345,
+      "step": 8283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992427213538272e-05,
+      "loss": 0.5254,
+      "step": 8284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999242081976214e-05,
+      "loss": 0.5921,
+      "step": 8285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992414423289005e-05,
+      "loss": 0.5146,
+      "step": 8286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992408024118867e-05,
+      "loss": 0.5988,
+      "step": 8287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992401622251733e-05,
+      "loss": 0.527,
+      "step": 8288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992395217687605e-05,
+      "loss": 0.5482,
+      "step": 8289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992388810426483e-05,
+      "loss": 0.7071,
+      "step": 8290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992382400468368e-05,
+      "loss": 0.5826,
+      "step": 8291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992375987813262e-05,
+      "loss": 0.607,
+      "step": 8292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992369572461167e-05,
+      "loss": 0.5321,
+      "step": 8293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992363154412084e-05,
+      "loss": 0.6564,
+      "step": 8294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992356733666015e-05,
+      "loss": 0.516,
+      "step": 8295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992350310222966e-05,
+      "loss": 0.6634,
+      "step": 8296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992343884082933e-05,
+      "loss": 0.5842,
+      "step": 8297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992337455245917e-05,
+      "loss": 0.5282,
+      "step": 8298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992331023711925e-05,
+      "loss": 0.5954,
+      "step": 8299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992324589480952e-05,
+      "loss": 0.6451,
+      "step": 8300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992318152553013e-05,
+      "loss": 0.6711,
+      "step": 8301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992311712928094e-05,
+      "loss": 0.5028,
+      "step": 8302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992305270606205e-05,
+      "loss": 0.5364,
+      "step": 8303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992298825587343e-05,
+      "loss": 0.6133,
+      "step": 8304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992292377871515e-05,
+      "loss": 0.6956,
+      "step": 8305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999228592745872e-05,
+      "loss": 0.6258,
+      "step": 8306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992279474348965e-05,
+      "loss": 0.7268,
+      "step": 8307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992273018542244e-05,
+      "loss": 0.5334,
+      "step": 8308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999226656003856e-05,
+      "loss": 0.5824,
+      "step": 8309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992260098837915e-05,
+      "loss": 0.562,
+      "step": 8310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992253634940315e-05,
+      "loss": 0.543,
+      "step": 8311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999224716834576e-05,
+      "loss": 0.6027,
+      "step": 8312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999224069905425e-05,
+      "loss": 0.717,
+      "step": 8313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992234227065787e-05,
+      "loss": 0.5889,
+      "step": 8314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999222775238037e-05,
+      "loss": 0.7032,
+      "step": 8315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999222127499801e-05,
+      "loss": 0.5801,
+      "step": 8316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992214794918703e-05,
+      "loss": 0.5892,
+      "step": 8317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992208312142446e-05,
+      "loss": 0.5552,
+      "step": 8318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992201826669246e-05,
+      "loss": 0.5707,
+      "step": 8319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992195338499107e-05,
+      "loss": 0.5095,
+      "step": 8320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992188847632025e-05,
+      "loss": 0.5246,
+      "step": 8321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992182354068007e-05,
+      "loss": 0.5729,
+      "step": 8322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999217585780705e-05,
+      "loss": 0.5959,
+      "step": 8323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992169358849157e-05,
+      "loss": 0.5432,
+      "step": 8324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999216285719433e-05,
+      "loss": 0.4805,
+      "step": 8325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992156352842577e-05,
+      "loss": 0.4991,
+      "step": 8326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992149845793892e-05,
+      "loss": 0.6257,
+      "step": 8327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9992143336048276e-05,
+      "loss": 0.617,
+      "step": 8328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999213682360574e-05,
+      "loss": 0.5524,
+      "step": 8329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992130308466272e-05,
+      "loss": 0.7181,
+      "step": 8330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992123790629888e-05,
+      "loss": 0.4831,
+      "step": 8331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992117270096582e-05,
+      "loss": 0.5287,
+      "step": 8332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992110746866353e-05,
+      "loss": 0.5271,
+      "step": 8333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999210422093921e-05,
+      "loss": 0.5755,
+      "step": 8334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999209769231515e-05,
+      "loss": 0.5771,
+      "step": 8335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992091160994173e-05,
+      "loss": 0.5978,
+      "step": 8336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999208462697629e-05,
+      "loss": 0.6403,
+      "step": 8337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992078090261492e-05,
+      "loss": 0.5508,
+      "step": 8338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992071550849787e-05,
+      "loss": 0.584,
+      "step": 8339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992065008741176e-05,
+      "loss": 0.5979,
+      "step": 8340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999205846393566e-05,
+      "loss": 0.5227,
+      "step": 8341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999205191643324e-05,
+      "loss": 0.6844,
+      "step": 8342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999204536623392e-05,
+      "loss": 0.6834,
+      "step": 8343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992038813337697e-05,
+      "loss": 0.509,
+      "step": 8344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992032257744576e-05,
+      "loss": 0.68,
+      "step": 8345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992025699454562e-05,
+      "loss": 0.6591,
+      "step": 8346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999201913846765e-05,
+      "loss": 0.5768,
+      "step": 8347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992012574783846e-05,
+      "loss": 0.6595,
+      "step": 8348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9992006008403154e-05,
+      "loss": 0.5701,
+      "step": 8349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991999439325573e-05,
+      "loss": 0.5975,
+      "step": 8350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991992867551103e-05,
+      "loss": 0.6562,
+      "step": 8351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991986293079745e-05,
+      "loss": 0.5117,
+      "step": 8352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999197971591151e-05,
+      "loss": 0.6276,
+      "step": 8353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991973136046384e-05,
+      "loss": 0.6006,
+      "step": 8354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991966553484384e-05,
+      "loss": 0.6074,
+      "step": 8355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991959968225502e-05,
+      "loss": 0.6193,
+      "step": 8356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991953380269745e-05,
+      "loss": 0.4934,
+      "step": 8357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991946789617117e-05,
+      "loss": 0.7038,
+      "step": 8358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999194019626761e-05,
+      "loss": 0.5773,
+      "step": 8359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991933600221232e-05,
+      "loss": 0.5103,
+      "step": 8360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991927001477986e-05,
+      "loss": 0.5932,
+      "step": 8361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991920400037875e-05,
+      "loss": 0.5346,
+      "step": 8362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991913795900895e-05,
+      "loss": 0.6588,
+      "step": 8363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999190718906705e-05,
+      "loss": 0.6126,
+      "step": 8364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991900579536342e-05,
+      "loss": 0.6611,
+      "step": 8365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991893967308775e-05,
+      "loss": 0.611,
+      "step": 8366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999188735238435e-05,
+      "loss": 0.6017,
+      "step": 8367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991880734763066e-05,
+      "loss": 0.4639,
+      "step": 8368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999187411444493e-05,
+      "loss": 0.5429,
+      "step": 8369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991867491429936e-05,
+      "loss": 0.6315,
+      "step": 8370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991860865718092e-05,
+      "loss": 0.5641,
+      "step": 8371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991854237309397e-05,
+      "loss": 0.599,
+      "step": 8372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991847606203854e-05,
+      "loss": 0.5276,
+      "step": 8373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991840972401467e-05,
+      "loss": 0.5578,
+      "step": 8374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991834335902235e-05,
+      "loss": 0.6229,
+      "step": 8375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991827696706162e-05,
+      "loss": 0.6778,
+      "step": 8376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999182105481324e-05,
+      "loss": 0.572,
+      "step": 8377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991814410223486e-05,
+      "loss": 0.6094,
+      "step": 8378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991807762936894e-05,
+      "loss": 0.5426,
+      "step": 8379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991801112953463e-05,
+      "loss": 0.6317,
+      "step": 8380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991794460273205e-05,
+      "loss": 0.5904,
+      "step": 8381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999178780489611e-05,
+      "loss": 0.5735,
+      "step": 8382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991781146822182e-05,
+      "loss": 0.5817,
+      "step": 8383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991774486051428e-05,
+      "loss": 0.5581,
+      "step": 8384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999176782258385e-05,
+      "loss": 0.6508,
+      "step": 8385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991761156419444e-05,
+      "loss": 0.6476,
+      "step": 8386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991754487558217e-05,
+      "loss": 0.5691,
+      "step": 8387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999174781600017e-05,
+      "loss": 0.5077,
+      "step": 8388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.99917411417453e-05,
+      "loss": 0.5378,
+      "step": 8389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991734464793612e-05,
+      "loss": 0.4931,
+      "step": 8390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999172778514511e-05,
+      "loss": 0.5887,
+      "step": 8391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991721102799796e-05,
+      "loss": 0.6337,
+      "step": 8392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991714417757665e-05,
+      "loss": 0.6002,
+      "step": 8393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991707730018726e-05,
+      "loss": 0.6336,
+      "step": 8394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999170103958298e-05,
+      "loss": 0.4861,
+      "step": 8395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991694346450425e-05,
+      "loss": 0.5896,
+      "step": 8396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999168765062107e-05,
+      "loss": 0.5241,
+      "step": 8397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.99916809520949e-05,
+      "loss": 0.5167,
+      "step": 8398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999167425087194e-05,
+      "loss": 0.5718,
+      "step": 8399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999166754695218e-05,
+      "loss": 0.5525,
+      "step": 8400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991660840335615e-05,
+      "loss": 0.6506,
+      "step": 8401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999165413102226e-05,
+      "loss": 0.5626,
+      "step": 8402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991647419012107e-05,
+      "loss": 0.6551,
+      "step": 8403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999164070430516e-05,
+      "loss": 0.5214,
+      "step": 8404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999163398690143e-05,
+      "loss": 0.5826,
+      "step": 8405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991627266800905e-05,
+      "loss": 0.5577,
+      "step": 8406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991620544003595e-05,
+      "loss": 0.5244,
+      "step": 8407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.99916138185095e-05,
+      "loss": 0.6163,
+      "step": 8408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991607090318624e-05,
+      "loss": 0.6987,
+      "step": 8409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991600359430964e-05,
+      "loss": 0.5761,
+      "step": 8410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991593625846523e-05,
+      "loss": 0.518,
+      "step": 8411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991586889565306e-05,
+      "loss": 0.5712,
+      "step": 8412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991580150587312e-05,
+      "loss": 0.5652,
+      "step": 8413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991573408912548e-05,
+      "loss": 0.5386,
+      "step": 8414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991566664541007e-05,
+      "loss": 0.5502,
+      "step": 8415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.99915599174727e-05,
+      "loss": 0.5768,
+      "step": 8416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999155316770762e-05,
+      "loss": 0.6561,
+      "step": 8417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991546415245772e-05,
+      "loss": 0.5249,
+      "step": 8418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999153966008716e-05,
+      "loss": 0.6884,
+      "step": 8419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991532902231788e-05,
+      "loss": 0.64,
+      "step": 8420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999152614167965e-05,
+      "loss": 0.4926,
+      "step": 8421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999151937843076e-05,
+      "loss": 0.5861,
+      "step": 8422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991512612485105e-05,
+      "loss": 0.5439,
+      "step": 8423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991505843842693e-05,
+      "loss": 0.5298,
+      "step": 8424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999149907250353e-05,
+      "loss": 0.6126,
+      "step": 8425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9991492298467615e-05,
+      "loss": 0.6869,
+      "step": 8426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999148552173495e-05,
+      "loss": 0.6172,
+      "step": 8427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991478742305536e-05,
+      "loss": 0.6147,
+      "step": 8428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991471960179375e-05,
+      "loss": 0.7256,
+      "step": 8429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991465175356465e-05,
+      "loss": 0.575,
+      "step": 8430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999145838783682e-05,
+      "loss": 0.5793,
+      "step": 8431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999145159762043e-05,
+      "loss": 0.6278,
+      "step": 8432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991444804707297e-05,
+      "loss": 0.6367,
+      "step": 8433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999143800909743e-05,
+      "loss": 0.5757,
+      "step": 8434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991431210790825e-05,
+      "loss": 0.5511,
+      "step": 8435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991424409787488e-05,
+      "loss": 0.6281,
+      "step": 8436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991417606087418e-05,
+      "loss": 0.5767,
+      "step": 8437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991410799690616e-05,
+      "loss": 0.6657,
+      "step": 8438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991403990597088e-05,
+      "loss": 0.6152,
+      "step": 8439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991397178806834e-05,
+      "loss": 0.6119,
+      "step": 8440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999139036431985e-05,
+      "loss": 0.6774,
+      "step": 8441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991383547136148e-05,
+      "loss": 0.7239,
+      "step": 8442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991376727255723e-05,
+      "loss": 0.5195,
+      "step": 8443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991369904678582e-05,
+      "loss": 0.5518,
+      "step": 8444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999136307940472e-05,
+      "loss": 0.4928,
+      "step": 8445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991356251434144e-05,
+      "loss": 0.6943,
+      "step": 8446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991349420766856e-05,
+      "loss": 0.4914,
+      "step": 8447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991342587402852e-05,
+      "loss": 0.6791,
+      "step": 8448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999133575134214e-05,
+      "loss": 0.5368,
+      "step": 8449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999132891258472e-05,
+      "loss": 0.6118,
+      "step": 8450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999132207113059e-05,
+      "loss": 0.5576,
+      "step": 8451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991315226979758e-05,
+      "loss": 0.5287,
+      "step": 8452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991308380132227e-05,
+      "loss": 0.6645,
+      "step": 8453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991301530587993e-05,
+      "loss": 0.5943,
+      "step": 8454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991294678347057e-05,
+      "loss": 0.661,
+      "step": 8455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991287823409426e-05,
+      "loss": 0.5552,
+      "step": 8456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.99912809657751e-05,
+      "loss": 0.5903,
+      "step": 8457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999127410544408e-05,
+      "loss": 0.6252,
+      "step": 8458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991267242416367e-05,
+      "loss": 0.5764,
+      "step": 8459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991260376691967e-05,
+      "loss": 0.4976,
+      "step": 8460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999125350827088e-05,
+      "loss": 0.5663,
+      "step": 8461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991246637153103e-05,
+      "loss": 0.5201,
+      "step": 8462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991239763338644e-05,
+      "loss": 0.5555,
+      "step": 8463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991232886827504e-05,
+      "loss": 0.6092,
+      "step": 8464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999122600761968e-05,
+      "loss": 0.5698,
+      "step": 8465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999121912571518e-05,
+      "loss": 0.548,
+      "step": 8466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991212241114002e-05,
+      "loss": 0.6608,
+      "step": 8467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999120535381615e-05,
+      "loss": 0.7094,
+      "step": 8468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991198463821626e-05,
+      "loss": 0.6645,
+      "step": 8469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991191571130428e-05,
+      "loss": 0.584,
+      "step": 8470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991184675742562e-05,
+      "loss": 0.5189,
+      "step": 8471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999117777765803e-05,
+      "loss": 0.6596,
+      "step": 8472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991170876876832e-05,
+      "loss": 0.5922,
+      "step": 8473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999116397339897e-05,
+      "loss": 0.6374,
+      "step": 8474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991157067224446e-05,
+      "loss": 0.4781,
+      "step": 8475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999115015835326e-05,
+      "loss": 0.5675,
+      "step": 8476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991143246785418e-05,
+      "loss": 0.533,
+      "step": 8477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991136332520918e-05,
+      "loss": 0.6455,
+      "step": 8478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991129415559763e-05,
+      "loss": 0.6616,
+      "step": 8479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999112249590196e-05,
+      "loss": 0.6708,
+      "step": 8480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991115573547503e-05,
+      "loss": 0.6237,
+      "step": 8481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991108648496398e-05,
+      "loss": 0.5908,
+      "step": 8482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991101720748648e-05,
+      "loss": 0.5067,
+      "step": 8483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999109479030425e-05,
+      "loss": 0.5885,
+      "step": 8484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991087857163207e-05,
+      "loss": 0.5156,
+      "step": 8485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991080921325524e-05,
+      "loss": 0.5416,
+      "step": 8486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991073982791206e-05,
+      "loss": 0.5565,
+      "step": 8487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991067041560246e-05,
+      "loss": 0.7057,
+      "step": 8488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991060097632653e-05,
+      "loss": 0.5695,
+      "step": 8489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991053151008424e-05,
+      "loss": 0.6354,
+      "step": 8490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991046201687565e-05,
+      "loss": 0.4985,
+      "step": 8491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991039249670075e-05,
+      "loss": 0.6445,
+      "step": 8492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991032294955954e-05,
+      "loss": 0.6511,
+      "step": 8493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999102533754521e-05,
+      "loss": 0.5104,
+      "step": 8494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991018377437842e-05,
+      "loss": 0.5897,
+      "step": 8495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991011414633848e-05,
+      "loss": 0.5407,
+      "step": 8496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9991004449133237e-05,
+      "loss": 0.5622,
+      "step": 8497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990997480936004e-05,
+      "loss": 0.6604,
+      "step": 8498
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990990510042158e-05,
+      "loss": 0.6107,
+      "step": 8499
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990983536451694e-05,
+      "loss": 0.6019,
+      "step": 8500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990976560164616e-05,
+      "loss": 0.5434,
+      "step": 8501
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999096958118093e-05,
+      "loss": 0.6539,
+      "step": 8502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990962599500635e-05,
+      "loss": 0.6833,
+      "step": 8503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990955615123728e-05,
+      "loss": 0.519,
+      "step": 8504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990948628050217e-05,
+      "loss": 0.6431,
+      "step": 8505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990941638280103e-05,
+      "loss": 0.5802,
+      "step": 8506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990934645813388e-05,
+      "loss": 0.5646,
+      "step": 8507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990927650650075e-05,
+      "loss": 0.6401,
+      "step": 8508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999092065279016e-05,
+      "loss": 0.5746,
+      "step": 8509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999091365223365e-05,
+      "loss": 0.5979,
+      "step": 8510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990906648980546e-05,
+      "loss": 0.5503,
+      "step": 8511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990899643030848e-05,
+      "loss": 0.631,
+      "step": 8512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990892634384564e-05,
+      "loss": 0.5952,
+      "step": 8513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990885623041685e-05,
+      "loss": 0.5185,
+      "step": 8514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990878609002224e-05,
+      "loss": 0.4676,
+      "step": 8515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990871592266178e-05,
+      "loss": 0.5455,
+      "step": 8516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990864572833546e-05,
+      "loss": 0.6405,
+      "step": 8517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990857550704333e-05,
+      "loss": 0.6516,
+      "step": 8518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990850525878544e-05,
+      "loss": 0.4925,
+      "step": 8519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990843498356177e-05,
+      "loss": 0.5311,
+      "step": 8520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990836468137234e-05,
+      "loss": 0.6334,
+      "step": 8521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990829435221718e-05,
+      "loss": 0.5029,
+      "step": 8522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990822399609628e-05,
+      "loss": 0.5736,
+      "step": 8523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999081536130097e-05,
+      "loss": 0.4951,
+      "step": 8524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990808320295745e-05,
+      "loss": 0.6612,
+      "step": 8525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990801276593955e-05,
+      "loss": 0.6514,
+      "step": 8526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990794230195598e-05,
+      "loss": 0.5585,
+      "step": 8527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999078718110068e-05,
+      "loss": 0.6862,
+      "step": 8528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990780129309205e-05,
+      "loss": 0.5704,
+      "step": 8529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990773074821166e-05,
+      "loss": 0.5556,
+      "step": 8530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990766017636573e-05,
+      "loss": 0.5329,
+      "step": 8531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990758957755427e-05,
+      "loss": 0.6064,
+      "step": 8532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999075189517773e-05,
+      "loss": 0.4925,
+      "step": 8533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990744829903476e-05,
+      "loss": 0.5662,
+      "step": 8534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990737761932678e-05,
+      "loss": 0.6158,
+      "step": 8535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990730691265333e-05,
+      "loss": 0.533,
+      "step": 8536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990723617901442e-05,
+      "loss": 0.5768,
+      "step": 8537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990716541841012e-05,
+      "loss": 0.5799,
+      "step": 8538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990709463084035e-05,
+      "loss": 0.6227,
+      "step": 8539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990702381630522e-05,
+      "loss": 0.5218,
+      "step": 8540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990695297480473e-05,
+      "loss": 0.5442,
+      "step": 8541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990688210633884e-05,
+      "loss": 0.5145,
+      "step": 8542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990681121090766e-05,
+      "loss": 0.5358,
+      "step": 8543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990674028851115e-05,
+      "loss": 0.6205,
+      "step": 8544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999066693391493e-05,
+      "loss": 0.6027,
+      "step": 8545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990659836282225e-05,
+      "loss": 0.5409,
+      "step": 8546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999065273595299e-05,
+      "loss": 0.5605,
+      "step": 8547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999064563292723e-05,
+      "loss": 0.5698,
+      "step": 8548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990638527204953e-05,
+      "loss": 0.5757,
+      "step": 8549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999063141878615e-05,
+      "loss": 0.6533,
+      "step": 8550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990624307670834e-05,
+      "loss": 0.5545,
+      "step": 8551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990617193858995e-05,
+      "loss": 0.542,
+      "step": 8552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999061007735065e-05,
+      "loss": 0.6037,
+      "step": 8553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990602958145788e-05,
+      "loss": 0.6671,
+      "step": 8554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999059583624442e-05,
+      "loss": 0.5776,
+      "step": 8555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990588711646537e-05,
+      "loss": 0.5003,
+      "step": 8556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999058158435215e-05,
+      "loss": 0.506,
+      "step": 8557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990574454361257e-05,
+      "loss": 0.5082,
+      "step": 8558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990567321673865e-05,
+      "loss": 0.6901,
+      "step": 8559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999056018628997e-05,
+      "loss": 0.5397,
+      "step": 8560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990553048209574e-05,
+      "loss": 0.5476,
+      "step": 8561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990545907432682e-05,
+      "loss": 0.58,
+      "step": 8562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990538763959298e-05,
+      "loss": 0.7783,
+      "step": 8563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999053161778942e-05,
+      "loss": 0.581,
+      "step": 8564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999052446892305e-05,
+      "loss": 0.629,
+      "step": 8565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990517317360187e-05,
+      "loss": 0.6264,
+      "step": 8566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999051016310084e-05,
+      "loss": 0.6031,
+      "step": 8567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999050300614501e-05,
+      "loss": 0.6328,
+      "step": 8568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990495846492693e-05,
+      "loss": 0.6588,
+      "step": 8569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990488684143895e-05,
+      "loss": 0.674,
+      "step": 8570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990481519098618e-05,
+      "loss": 0.5736,
+      "step": 8571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990474351356866e-05,
+      "loss": 0.5547,
+      "step": 8572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990467180918635e-05,
+      "loss": 0.5541,
+      "step": 8573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990460007783933e-05,
+      "loss": 0.5821,
+      "step": 8574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990452831952755e-05,
+      "loss": 0.6457,
+      "step": 8575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999044565342511e-05,
+      "loss": 0.5628,
+      "step": 8576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990438472200995e-05,
+      "loss": 0.6644,
+      "step": 8577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990431288280415e-05,
+      "loss": 0.6039,
+      "step": 8578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990424101663374e-05,
+      "loss": 0.5444,
+      "step": 8579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990416912349865e-05,
+      "loss": 0.5899,
+      "step": 8580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.99904097203399e-05,
+      "loss": 0.5308,
+      "step": 8581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990402525633478e-05,
+      "loss": 0.4981,
+      "step": 8582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990395328230594e-05,
+      "loss": 0.5693,
+      "step": 8583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990388128131258e-05,
+      "loss": 0.5326,
+      "step": 8584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990380925335474e-05,
+      "loss": 0.598,
+      "step": 8585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990373719843236e-05,
+      "loss": 0.5426,
+      "step": 8586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990366511654546e-05,
+      "loss": 0.5531,
+      "step": 8587
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990359300769414e-05,
+      "loss": 0.5968,
+      "step": 8588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990352087187835e-05,
+      "loss": 0.5261,
+      "step": 8589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990344870909814e-05,
+      "loss": 0.5879,
+      "step": 8590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990337651935352e-05,
+      "loss": 0.658,
+      "step": 8591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999033043026445e-05,
+      "loss": 0.566,
+      "step": 8592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990323205897115e-05,
+      "loss": 0.5915,
+      "step": 8593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990315978833342e-05,
+      "loss": 0.5686,
+      "step": 8594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999030874907314e-05,
+      "loss": 0.6032,
+      "step": 8595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.99903015166165e-05,
+      "loss": 0.5048,
+      "step": 8596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990294281463435e-05,
+      "loss": 0.534,
+      "step": 8597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990287043613942e-05,
+      "loss": 0.5483,
+      "step": 8598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990279803068024e-05,
+      "loss": 0.5491,
+      "step": 8599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990272559825685e-05,
+      "loss": 0.5297,
+      "step": 8600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990265313886922e-05,
+      "loss": 0.5658,
+      "step": 8601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999025806525174e-05,
+      "loss": 0.5446,
+      "step": 8602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990250813920143e-05,
+      "loss": 0.5621,
+      "step": 8603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990243559892127e-05,
+      "loss": 0.5996,
+      "step": 8604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990236303167697e-05,
+      "loss": 0.6017,
+      "step": 8605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999022904374686e-05,
+      "loss": 0.64,
+      "step": 8606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999022178162961e-05,
+      "loss": 0.611,
+      "step": 8607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990214516815956e-05,
+      "loss": 0.6626,
+      "step": 8608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990207249305892e-05,
+      "loss": 0.6615,
+      "step": 8609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999019997909943e-05,
+      "loss": 0.5187,
+      "step": 8610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990192706196564e-05,
+      "loss": 0.6082,
+      "step": 8611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990185430597292e-05,
+      "loss": 0.587,
+      "step": 8612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999017815230163e-05,
+      "loss": 0.578,
+      "step": 8613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999017087130957e-05,
+      "loss": 0.6313,
+      "step": 8614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990163587621114e-05,
+      "loss": 0.6011,
+      "step": 8615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990156301236267e-05,
+      "loss": 0.5832,
+      "step": 8616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990149012155032e-05,
+      "loss": 0.6083,
+      "step": 8617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999014172037741e-05,
+      "loss": 0.6179,
+      "step": 8618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.99901344259034e-05,
+      "loss": 0.5465,
+      "step": 8619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990127128733007e-05,
+      "loss": 0.5957,
+      "step": 8620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999011982886623e-05,
+      "loss": 0.6377,
+      "step": 8621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9990112526303074e-05,
+      "loss": 0.5236,
+      "step": 8622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999010522104354e-05,
+      "loss": 0.5319,
+      "step": 8623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999009791308763e-05,
+      "loss": 0.5718,
+      "step": 8624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9990090602435348e-05,
+      "loss": 0.5792,
+      "step": 8625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999008328908669e-05,
+      "loss": 0.6374,
+      "step": 8626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9990075973041663e-05,
+      "loss": 0.6028,
+      "step": 8627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999006865430027e-05,
+      "loss": 0.66,
+      "step": 8628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999006133286251e-05,
+      "loss": 0.5642,
+      "step": 8629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9990054008728384e-05,
+      "loss": 0.5678,
+      "step": 8630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9990046681897897e-05,
+      "loss": 0.67,
+      "step": 8631
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999003935237105e-05,
+      "loss": 0.5616,
+      "step": 8632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9990032020147842e-05,
+      "loss": 0.6232,
+      "step": 8633
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999002468522828e-05,
+      "loss": 0.6627,
+      "step": 8634
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9990017347612364e-05,
+      "loss": 0.6092,
+      "step": 8635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9990010007300093e-05,
+      "loss": 0.5893,
+      "step": 8636
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9990002664291476e-05,
+      "loss": 0.6259,
+      "step": 8637
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989995318586506e-05,
+      "loss": 0.5259,
+      "step": 8638
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989987970185193e-05,
+      "loss": 0.5541,
+      "step": 8639
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989980619087533e-05,
+      "loss": 0.5924,
+      "step": 8640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998997326529353e-05,
+      "loss": 0.5831,
+      "step": 8641
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989965908803192e-05,
+      "loss": 0.581,
+      "step": 8642
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998995854961651e-05,
+      "loss": 0.5933,
+      "step": 8643
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989951187733495e-05,
+      "loss": 0.6138,
+      "step": 8644
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989943823154144e-05,
+      "loss": 0.6286,
+      "step": 8645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989936455878457e-05,
+      "loss": 0.5736,
+      "step": 8646
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989929085906447e-05,
+      "loss": 0.5715,
+      "step": 8647
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.99899217132381e-05,
+      "loss": 0.7134,
+      "step": 8648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998991433787343e-05,
+      "loss": 0.5897,
+      "step": 8649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989906959812436e-05,
+      "loss": 0.5568,
+      "step": 8650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998989957905512e-05,
+      "loss": 0.5535,
+      "step": 8651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989892195601484e-05,
+      "loss": 0.63,
+      "step": 8652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989884809451528e-05,
+      "loss": 0.5716,
+      "step": 8653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989877420605256e-05,
+      "loss": 0.5788,
+      "step": 8654
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989870029062667e-05,
+      "loss": 0.5609,
+      "step": 8655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998986263482377e-05,
+      "loss": 0.518,
+      "step": 8656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998985523788856e-05,
+      "loss": 0.5875,
+      "step": 8657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998984783825704e-05,
+      "loss": 0.5362,
+      "step": 8658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989840435929212e-05,
+      "loss": 0.6282,
+      "step": 8659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989833030905085e-05,
+      "loss": 0.5951,
+      "step": 8660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998982562318465e-05,
+      "loss": 0.6097,
+      "step": 8661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989818212767916e-05,
+      "loss": 0.5532,
+      "step": 8662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989810799654885e-05,
+      "loss": 0.607,
+      "step": 8663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989803383845558e-05,
+      "loss": 0.5781,
+      "step": 8664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989795965339932e-05,
+      "loss": 0.4789,
+      "step": 8665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989788544138018e-05,
+      "loss": 0.5829,
+      "step": 8666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998978112023981e-05,
+      "loss": 0.6438,
+      "step": 8667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989773693645315e-05,
+      "loss": 0.6742,
+      "step": 8668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989766264354534e-05,
+      "loss": 0.4911,
+      "step": 8669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989758832367468e-05,
+      "loss": 0.6338,
+      "step": 8670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998975139768412e-05,
+      "loss": 0.5967,
+      "step": 8671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998974396030449e-05,
+      "loss": 0.6006,
+      "step": 8672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998973652022858e-05,
+      "loss": 0.5218,
+      "step": 8673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989729077456395e-05,
+      "loss": 0.5453,
+      "step": 8674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989721631987935e-05,
+      "loss": 0.5474,
+      "step": 8675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989714183823203e-05,
+      "loss": 0.5452,
+      "step": 8676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989706732962203e-05,
+      "loss": 0.4545,
+      "step": 8677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998969927940493e-05,
+      "loss": 0.6341,
+      "step": 8678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989691823151393e-05,
+      "loss": 0.6529,
+      "step": 8679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989684364201588e-05,
+      "loss": 0.6454,
+      "step": 8680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989676902555527e-05,
+      "loss": 0.542,
+      "step": 8681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.99896694382132e-05,
+      "loss": 0.6451,
+      "step": 8682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989661971174617e-05,
+      "loss": 0.5794,
+      "step": 8683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989654501439778e-05,
+      "loss": 0.6498,
+      "step": 8684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989647029008684e-05,
+      "loss": 0.5721,
+      "step": 8685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989639553881335e-05,
+      "loss": 0.572,
+      "step": 8686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989632076057738e-05,
+      "loss": 0.5319,
+      "step": 8687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989624595537893e-05,
+      "loss": 0.6318,
+      "step": 8688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.99896171123218e-05,
+      "loss": 0.5359,
+      "step": 8689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989609626409464e-05,
+      "loss": 0.5822,
+      "step": 8690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989602137800885e-05,
+      "loss": 0.5145,
+      "step": 8691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989594646496067e-05,
+      "loss": 0.6111,
+      "step": 8692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989587152495008e-05,
+      "loss": 0.7932,
+      "step": 8693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989579655797714e-05,
+      "loss": 0.538,
+      "step": 8694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989572156404186e-05,
+      "loss": 0.534,
+      "step": 8695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989564654314426e-05,
+      "loss": 0.5377,
+      "step": 8696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989557149528435e-05,
+      "loss": 0.5943,
+      "step": 8697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989549642046217e-05,
+      "loss": 0.5014,
+      "step": 8698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998954213186777e-05,
+      "loss": 0.5756,
+      "step": 8699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989534618993103e-05,
+      "loss": 0.5219,
+      "step": 8700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989527103422214e-05,
+      "loss": 0.4913,
+      "step": 8701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.99895195851551e-05,
+      "loss": 0.6614,
+      "step": 8702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989512064191774e-05,
+      "loss": 0.5612,
+      "step": 8703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998950454053223e-05,
+      "loss": 0.5714,
+      "step": 8704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998949701417647e-05,
+      "loss": 0.6107,
+      "step": 8705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.99894894851245e-05,
+      "loss": 0.6071,
+      "step": 8706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989481953376318e-05,
+      "loss": 0.5402,
+      "step": 8707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998947441893193e-05,
+      "loss": 0.6458,
+      "step": 8708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989466881791337e-05,
+      "loss": 0.6224,
+      "step": 8709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989459341954536e-05,
+      "loss": 0.5845,
+      "step": 8710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998945179942154e-05,
+      "loss": 0.5386,
+      "step": 8711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989444254192337e-05,
+      "loss": 0.5696,
+      "step": 8712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998943670626694e-05,
+      "loss": 0.5065,
+      "step": 8713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989429155645345e-05,
+      "loss": 0.5611,
+      "step": 8714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989421602327558e-05,
+      "loss": 0.6199,
+      "step": 8715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998941404631358e-05,
+      "loss": 0.5465,
+      "step": 8716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989406487603412e-05,
+      "loss": 0.6439,
+      "step": 8717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989398926197056e-05,
+      "loss": 0.5553,
+      "step": 8718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989391362094514e-05,
+      "loss": 0.51,
+      "step": 8719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998938379529579e-05,
+      "loss": 0.5428,
+      "step": 8720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9989376225800887e-05,
+      "loss": 0.6343,
+      "step": 8721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.99893686536098e-05,
+      "loss": 0.6151,
+      "step": 8722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.998936107872254e-05,
+      "loss": 0.5715,
+      "step": 8723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.99893535011391e-05,
+      "loss": 0.5718,
+      "step": 8724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998934592085949e-05,
+      "loss": 0.5474,
+      "step": 8725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998933833788371e-05,
+      "loss": 0.518,
+      "step": 8726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989330752211756e-05,
+      "loss": 0.5978,
+      "step": 8727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989323163843638e-05,
+      "loss": 0.6148,
+      "step": 8728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998931557277936e-05,
+      "loss": 0.5446,
+      "step": 8729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998930797901891e-05,
+      "loss": 0.6077,
+      "step": 8730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989300382562304e-05,
+      "loss": 0.4573,
+      "step": 8731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989292783409538e-05,
+      "loss": 0.469,
+      "step": 8732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989285181560615e-05,
+      "loss": 0.5494,
+      "step": 8733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998927757701554e-05,
+      "loss": 0.5261,
+      "step": 8734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989269969774307e-05,
+      "loss": 0.5777,
+      "step": 8735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989262359836924e-05,
+      "loss": 0.5599,
+      "step": 8736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989254747203395e-05,
+      "loss": 0.6217,
+      "step": 8737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989247131873723e-05,
+      "loss": 0.6444,
+      "step": 8738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989239513847897e-05,
+      "loss": 0.5254,
+      "step": 8739
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998923189312594e-05,
+      "loss": 0.6416,
+      "step": 8740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998922426970783e-05,
+      "loss": 0.6411,
+      "step": 8741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989216643593593e-05,
+      "loss": 0.5897,
+      "step": 8742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989209014783215e-05,
+      "loss": 0.6307,
+      "step": 8743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.99892013832767e-05,
+      "loss": 0.5643,
+      "step": 8744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989193749074057e-05,
+      "loss": 0.5363,
+      "step": 8745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989186112175284e-05,
+      "loss": 0.6438,
+      "step": 8746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998917847258038e-05,
+      "loss": 0.6767,
+      "step": 8747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989170830289352e-05,
+      "loss": 0.5248,
+      "step": 8748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.99891631853022e-05,
+      "loss": 0.5552,
+      "step": 8749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998915553761893e-05,
+      "loss": 0.5909,
+      "step": 8750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998914788723953e-05,
+      "loss": 0.6428,
+      "step": 8751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998914023416402e-05,
+      "loss": 0.5818,
+      "step": 8752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989132578392395e-05,
+      "loss": 0.6142,
+      "step": 8753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989124919924653e-05,
+      "loss": 0.5896,
+      "step": 8754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.99891172587608e-05,
+      "loss": 0.5334,
+      "step": 8755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998910959490084e-05,
+      "loss": 0.5723,
+      "step": 8756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998910192834477e-05,
+      "loss": 0.5039,
+      "step": 8757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989094259092594e-05,
+      "loss": 0.5798,
+      "step": 8758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989086587144316e-05,
+      "loss": 0.521,
+      "step": 8759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998907891249994e-05,
+      "loss": 0.6666,
+      "step": 8760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989071235159462e-05,
+      "loss": 0.6262,
+      "step": 8761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989063555122886e-05,
+      "loss": 0.5114,
+      "step": 8762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989055872390215e-05,
+      "loss": 0.6948,
+      "step": 8763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998904818696145e-05,
+      "loss": 0.6556,
+      "step": 8764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.99890404988366e-05,
+      "loss": 0.607,
+      "step": 8765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989032808015657e-05,
+      "loss": 0.584,
+      "step": 8766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989025114498627e-05,
+      "loss": 0.5973,
+      "step": 8767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989017418285514e-05,
+      "loss": 0.6742,
+      "step": 8768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989009719376316e-05,
+      "loss": 0.6123,
+      "step": 8769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9989002017771043e-05,
+      "loss": 0.6261,
+      "step": 8770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988994313469687e-05,
+      "loss": 0.554,
+      "step": 8771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988986606472256e-05,
+      "loss": 0.6671,
+      "step": 8772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998897889677875e-05,
+      "loss": 0.5153,
+      "step": 8773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988971184389174e-05,
+      "loss": 0.5014,
+      "step": 8774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988963469303527e-05,
+      "loss": 0.6718,
+      "step": 8775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988955751521812e-05,
+      "loss": 0.6149,
+      "step": 8776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998894803104403e-05,
+      "loss": 0.6443,
+      "step": 8777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988940307870185e-05,
+      "loss": 0.57,
+      "step": 8778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988932582000277e-05,
+      "loss": 0.6209,
+      "step": 8779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998892485343431e-05,
+      "loss": 0.5059,
+      "step": 8780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998891712217229e-05,
+      "loss": 0.5291,
+      "step": 8781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998890938821421e-05,
+      "loss": 0.7517,
+      "step": 8782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988901651560075e-05,
+      "loss": 0.5549,
+      "step": 8783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998889391220989e-05,
+      "loss": 0.6664,
+      "step": 8784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998888617016366e-05,
+      "loss": 0.5772,
+      "step": 8785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998887842542138e-05,
+      "loss": 0.6288,
+      "step": 8786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988870677983056e-05,
+      "loss": 0.6096,
+      "step": 8787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988862927848686e-05,
+      "loss": 0.552,
+      "step": 8788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998885517501828e-05,
+      "loss": 0.5907,
+      "step": 8789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988847419491833e-05,
+      "loss": 0.672,
+      "step": 8790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998883966126935e-05,
+      "loss": 0.549,
+      "step": 8791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998883190035083e-05,
+      "loss": 0.5315,
+      "step": 8792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998882413673628e-05,
+      "loss": 0.6049,
+      "step": 8793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.99888163704257e-05,
+      "loss": 0.6168,
+      "step": 8794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988808601419087e-05,
+      "loss": 0.5341,
+      "step": 8795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988800829716454e-05,
+      "loss": 0.5707,
+      "step": 8796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988793055317792e-05,
+      "loss": 0.5514,
+      "step": 8797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988785278223113e-05,
+      "loss": 0.631,
+      "step": 8798
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998877749843241e-05,
+      "loss": 0.7031,
+      "step": 8799
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998876971594569e-05,
+      "loss": 0.5167,
+      "step": 8800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988761930762958e-05,
+      "loss": 0.6148,
+      "step": 8801
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988754142884207e-05,
+      "loss": 0.6636,
+      "step": 8802
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988746352309446e-05,
+      "loss": 0.6957,
+      "step": 8803
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998873855903868e-05,
+      "loss": 0.5185,
+      "step": 8804
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988730763071903e-05,
+      "loss": 0.5972,
+      "step": 8805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988722964409122e-05,
+      "loss": 0.6005,
+      "step": 8806
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988715163050337e-05,
+      "loss": 0.5714,
+      "step": 8807
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988707358995555e-05,
+      "loss": 0.5609,
+      "step": 8808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988699552244768e-05,
+      "loss": 0.5676,
+      "step": 8809
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988691742797987e-05,
+      "loss": 0.6064,
+      "step": 8810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988683930655212e-05,
+      "loss": 0.5597,
+      "step": 8811
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988676115816443e-05,
+      "loss": 0.7621,
+      "step": 8812
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988668298281686e-05,
+      "loss": 0.5766,
+      "step": 8813
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998866047805094e-05,
+      "loss": 0.6163,
+      "step": 8814
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988652655124205e-05,
+      "loss": 0.5056,
+      "step": 8815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998864482950149e-05,
+      "loss": 0.5463,
+      "step": 8816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998863700118279e-05,
+      "loss": 0.5118,
+      "step": 8817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988629170168112e-05,
+      "loss": 0.5314,
+      "step": 8818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988621336457456e-05,
+      "loss": 0.6209,
+      "step": 8819
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988613500050826e-05,
+      "loss": 0.5638,
+      "step": 8820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.998860566094822e-05,
+      "loss": 0.5748,
+      "step": 8821
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9988597819149645e-05,
+      "loss": 0.5808,
+      "step": 8822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988589974655097e-05,
+      "loss": 0.5577,
+      "step": 8823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988582127464585e-05,
+      "loss": 0.6044,
+      "step": 8824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988574277578107e-05,
+      "loss": 0.5499,
+      "step": 8825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988566424995665e-05,
+      "loss": 0.6046,
+      "step": 8826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988558569717266e-05,
+      "loss": 0.539,
+      "step": 8827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988550711742904e-05,
+      "loss": 0.7069,
+      "step": 8828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988542851072585e-05,
+      "loss": 0.5934,
+      "step": 8829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988534987706315e-05,
+      "loss": 0.5152,
+      "step": 8830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988527121644093e-05,
+      "loss": 0.589,
+      "step": 8831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988519252885917e-05,
+      "loss": 0.6088,
+      "step": 8832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988511381431798e-05,
+      "loss": 0.5689,
+      "step": 8833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998850350728173e-05,
+      "loss": 0.5496,
+      "step": 8834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998849563043572e-05,
+      "loss": 0.6438,
+      "step": 8835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988487750893767e-05,
+      "loss": 0.5524,
+      "step": 8836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998847986865587e-05,
+      "loss": 0.5709,
+      "step": 8837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988471983722043e-05,
+      "loss": 0.5607,
+      "step": 8838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988464096092274e-05,
+      "loss": 0.6162,
+      "step": 8839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988456205766576e-05,
+      "loss": 0.521,
+      "step": 8840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998844831274495e-05,
+      "loss": 0.5265,
+      "step": 8841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998844041702739e-05,
+      "loss": 0.5824,
+      "step": 8842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988432518613904e-05,
+      "loss": 0.5661,
+      "step": 8843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988424617504493e-05,
+      "loss": 0.5664,
+      "step": 8844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988416713699164e-05,
+      "loss": 0.5339,
+      "step": 8845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998840880719791e-05,
+      "loss": 0.587,
+      "step": 8846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998840089800074e-05,
+      "loss": 0.5506,
+      "step": 8847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988392986107654e-05,
+      "loss": 0.5734,
+      "step": 8848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988385071518653e-05,
+      "loss": 0.5888,
+      "step": 8849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998837715423374e-05,
+      "loss": 0.5045,
+      "step": 8850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988369234252916e-05,
+      "loss": 0.6427,
+      "step": 8851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988361311576184e-05,
+      "loss": 0.5431,
+      "step": 8852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988353386203552e-05,
+      "loss": 0.6982,
+      "step": 8853
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998834545813501e-05,
+      "loss": 0.6392,
+      "step": 8854
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998833752737057e-05,
+      "loss": 0.6198,
+      "step": 8855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988329593910235e-05,
+      "loss": 0.6046,
+      "step": 8856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988321657754e-05,
+      "loss": 0.6874,
+      "step": 8857
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998831371890187e-05,
+      "loss": 0.6843,
+      "step": 8858
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988305777353845e-05,
+      "loss": 0.5781,
+      "step": 8859
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988297833109933e-05,
+      "loss": 0.6232,
+      "step": 8860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988289886170128e-05,
+      "loss": 0.6419,
+      "step": 8861
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998828193653444e-05,
+      "loss": 0.5199,
+      "step": 8862
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998827398420287e-05,
+      "loss": 0.5276,
+      "step": 8863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998826602917542e-05,
+      "loss": 0.5562,
+      "step": 8864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988258071452086e-05,
+      "loss": 0.5333,
+      "step": 8865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988250111032873e-05,
+      "loss": 0.5884,
+      "step": 8866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988242147917788e-05,
+      "loss": 0.5983,
+      "step": 8867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988234182106828e-05,
+      "loss": 0.5953,
+      "step": 8868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.99882262136e-05,
+      "loss": 0.631,
+      "step": 8869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988218242397298e-05,
+      "loss": 0.5597,
+      "step": 8870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998821026849873e-05,
+      "loss": 0.6057,
+      "step": 8871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988202291904303e-05,
+      "loss": 0.721,
+      "step": 8872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988194312614006e-05,
+      "loss": 0.4756,
+      "step": 8873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988186330627854e-05,
+      "loss": 0.5239,
+      "step": 8874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988178345945843e-05,
+      "loss": 0.6708,
+      "step": 8875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988170358567977e-05,
+      "loss": 0.5699,
+      "step": 8876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988162368494252e-05,
+      "loss": 0.6185,
+      "step": 8877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998815437572468e-05,
+      "loss": 0.672,
+      "step": 8878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998814638025926e-05,
+      "loss": 0.5416,
+      "step": 8879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988138382097985e-05,
+      "loss": 0.5103,
+      "step": 8880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988130381240867e-05,
+      "loss": 0.5833,
+      "step": 8881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998812237768791e-05,
+      "loss": 0.5814,
+      "step": 8882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988114371439114e-05,
+      "loss": 0.5544,
+      "step": 8883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988106362494474e-05,
+      "loss": 0.6228,
+      "step": 8884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988098350853997e-05,
+      "loss": 0.6129,
+      "step": 8885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988090336517688e-05,
+      "loss": 0.5631,
+      "step": 8886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988082319485547e-05,
+      "loss": 0.5839,
+      "step": 8887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988074299757575e-05,
+      "loss": 0.5507,
+      "step": 8888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988066277333776e-05,
+      "loss": 0.5176,
+      "step": 8889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998805825221415e-05,
+      "loss": 0.5146,
+      "step": 8890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988050224398702e-05,
+      "loss": 0.5967,
+      "step": 8891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998804219388743e-05,
+      "loss": 0.5437,
+      "step": 8892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998803416068034e-05,
+      "loss": 0.6352,
+      "step": 8893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988026124777433e-05,
+      "loss": 0.5483,
+      "step": 8894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988018086178713e-05,
+      "loss": 0.671,
+      "step": 8895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9988010044884177e-05,
+      "loss": 0.6027,
+      "step": 8896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998800200089383e-05,
+      "loss": 0.5465,
+      "step": 8897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987993954207677e-05,
+      "loss": 0.6016,
+      "step": 8898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998798590482572e-05,
+      "loss": 0.5153,
+      "step": 8899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987977852747953e-05,
+      "loss": 0.5265,
+      "step": 8900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998796979797439e-05,
+      "loss": 0.4958,
+      "step": 8901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998796174050502e-05,
+      "loss": 0.6029,
+      "step": 8902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998795368033986e-05,
+      "loss": 0.5995,
+      "step": 8903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987945617478903e-05,
+      "loss": 0.5902,
+      "step": 8904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998793755192215e-05,
+      "loss": 0.5812,
+      "step": 8905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987929483669607e-05,
+      "loss": 0.5288,
+      "step": 8906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987921412721275e-05,
+      "loss": 0.6395,
+      "step": 8907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987913339077156e-05,
+      "loss": 0.682,
+      "step": 8908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987905262737253e-05,
+      "loss": 0.6245,
+      "step": 8909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998789718370157e-05,
+      "loss": 0.5379,
+      "step": 8910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.99878891019701e-05,
+      "loss": 0.4748,
+      "step": 8911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998788101754286e-05,
+      "loss": 0.5909,
+      "step": 8912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998787293041984e-05,
+      "loss": 0.6573,
+      "step": 8913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987864840601047e-05,
+      "loss": 0.6551,
+      "step": 8914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987856748086484e-05,
+      "loss": 0.6912,
+      "step": 8915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998784865287615e-05,
+      "loss": 0.6604,
+      "step": 8916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998784055497005e-05,
+      "loss": 0.6132,
+      "step": 8917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987832454368185e-05,
+      "loss": 0.6283,
+      "step": 8918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987824351070556e-05,
+      "loss": 0.6164,
+      "step": 8919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9987816245077168e-05,
+      "loss": 0.5779,
+      "step": 8920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998780813638802e-05,
+      "loss": 0.6429,
+      "step": 8921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987800025003117e-05,
+      "loss": 0.5391,
+      "step": 8922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998779191092246e-05,
+      "loss": 0.6057,
+      "step": 8923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987783794146055e-05,
+      "loss": 0.6364,
+      "step": 8924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987775674673896e-05,
+      "loss": 0.5455,
+      "step": 8925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998776755250599e-05,
+      "loss": 0.6621,
+      "step": 8926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998775942764234e-05,
+      "loss": 0.6038,
+      "step": 8927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987751300082948e-05,
+      "loss": 0.6697,
+      "step": 8928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998774316982781e-05,
+      "loss": 0.577,
+      "step": 8929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998773503687694e-05,
+      "loss": 0.6158,
+      "step": 8930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998772690123033e-05,
+      "loss": 0.5835,
+      "step": 8931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987718762887987e-05,
+      "loss": 0.5119,
+      "step": 8932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998771062184991e-05,
+      "loss": 0.5699,
+      "step": 8933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987702478116103e-05,
+      "loss": 0.4462,
+      "step": 8934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987694331686574e-05,
+      "loss": 0.5866,
+      "step": 8935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987686182561313e-05,
+      "loss": 0.5548,
+      "step": 8936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987678030740335e-05,
+      "loss": 0.68,
+      "step": 8937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998766987622363e-05,
+      "loss": 0.5272,
+      "step": 8938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998766171901121e-05,
+      "loss": 0.5689,
+      "step": 8939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987653559103074e-05,
+      "loss": 0.5844,
+      "step": 8940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998764539649922e-05,
+      "loss": 0.6065,
+      "step": 8941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987637231199658e-05,
+      "loss": 0.5601,
+      "step": 8942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987629063204383e-05,
+      "loss": 0.5835,
+      "step": 8943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.99876208925134e-05,
+      "loss": 0.5355,
+      "step": 8944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998761271912671e-05,
+      "loss": 0.5662,
+      "step": 8945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987604543044324e-05,
+      "loss": 0.6412,
+      "step": 8946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998759636426623e-05,
+      "loss": 0.57,
+      "step": 8947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998758818279244e-05,
+      "loss": 0.6152,
+      "step": 8948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998757999862295e-05,
+      "loss": 0.5929,
+      "step": 8949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998757181175777e-05,
+      "loss": 0.5216,
+      "step": 8950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987563622196898e-05,
+      "loss": 0.5185,
+      "step": 8951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998755542994033e-05,
+      "loss": 0.5107,
+      "step": 8952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987547234988078e-05,
+      "loss": 0.5995,
+      "step": 8953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998753903734014e-05,
+      "loss": 0.5622,
+      "step": 8954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987530836996518e-05,
+      "loss": 0.5182,
+      "step": 8955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987522633957213e-05,
+      "loss": 0.6498,
+      "step": 8956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987514428222236e-05,
+      "loss": 0.5527,
+      "step": 8957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987506219791576e-05,
+      "loss": 0.5412,
+      "step": 8958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998749800866524e-05,
+      "loss": 0.4928,
+      "step": 8959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987489794843237e-05,
+      "loss": 0.5633,
+      "step": 8960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998748157832556e-05,
+      "loss": 0.6648,
+      "step": 8961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987473359112216e-05,
+      "loss": 0.5694,
+      "step": 8962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987465137203205e-05,
+      "loss": 0.5516,
+      "step": 8963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987456912598533e-05,
+      "loss": 0.6921,
+      "step": 8964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.99874486852982e-05,
+      "loss": 0.5986,
+      "step": 8965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987440455302205e-05,
+      "loss": 0.5144,
+      "step": 8966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987432222610554e-05,
+      "loss": 0.6609,
+      "step": 8967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998742398722325e-05,
+      "loss": 0.5693,
+      "step": 8968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987415749140295e-05,
+      "loss": 0.625,
+      "step": 8969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987407508361684e-05,
+      "loss": 0.6312,
+      "step": 8970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987399264887428e-05,
+      "loss": 0.6261,
+      "step": 8971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987391018717527e-05,
+      "loss": 0.5671,
+      "step": 8972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998738276985198e-05,
+      "loss": 0.5328,
+      "step": 8973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987374518290794e-05,
+      "loss": 0.5021,
+      "step": 8974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998736626403397e-05,
+      "loss": 0.7636,
+      "step": 8975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987358007081507e-05,
+      "loss": 0.5234,
+      "step": 8976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998734974743341e-05,
+      "loss": 0.688,
+      "step": 8977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987341485089684e-05,
+      "loss": 0.5598,
+      "step": 8978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998733322005032e-05,
+      "loss": 0.5906,
+      "step": 8979
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987324952315338e-05,
+      "loss": 0.5459,
+      "step": 8980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987316681884722e-05,
+      "loss": 0.6285,
+      "step": 8981
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987308408758486e-05,
+      "loss": 0.5507,
+      "step": 8982
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987300132936628e-05,
+      "loss": 0.6206,
+      "step": 8983
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987291854419152e-05,
+      "loss": 0.6594,
+      "step": 8984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987283573206058e-05,
+      "loss": 0.5257,
+      "step": 8985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998727528929735e-05,
+      "loss": 0.6782,
+      "step": 8986
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998726700269303e-05,
+      "loss": 0.5066,
+      "step": 8987
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.99872587133931e-05,
+      "loss": 0.6207,
+      "step": 8988
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998725042139756e-05,
+      "loss": 0.5986,
+      "step": 8989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987242126706417e-05,
+      "loss": 0.5332,
+      "step": 8990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987233829319673e-05,
+      "loss": 0.6323,
+      "step": 8991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998722552923732e-05,
+      "loss": 0.5073,
+      "step": 8992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987217226459375e-05,
+      "loss": 0.5097,
+      "step": 8993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998720892098583e-05,
+      "loss": 0.5541,
+      "step": 8994
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987200612816696e-05,
+      "loss": 0.5673,
+      "step": 8995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987192301951964e-05,
+      "loss": 0.547,
+      "step": 8996
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987183988391644e-05,
+      "loss": 0.5106,
+      "step": 8997
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987175672135737e-05,
+      "loss": 0.5447,
+      "step": 8998
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987167353184245e-05,
+      "loss": 0.6412,
+      "step": 8999
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987159031537166e-05,
+      "loss": 0.5996,
+      "step": 9000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998715070719451e-05,
+      "loss": 0.5944,
+      "step": 9001
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987142380156276e-05,
+      "loss": 0.6144,
+      "step": 9002
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987134050422462e-05,
+      "loss": 0.5935,
+      "step": 9003
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987125717993077e-05,
+      "loss": 0.4981,
+      "step": 9004
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998711738286812e-05,
+      "loss": 0.6015,
+      "step": 9005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987109045047592e-05,
+      "loss": 0.5712,
+      "step": 9006
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987100704531496e-05,
+      "loss": 0.5322,
+      "step": 9007
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987092361319836e-05,
+      "loss": 0.5469,
+      "step": 9008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998708401541261e-05,
+      "loss": 0.5043,
+      "step": 9009
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987075666809827e-05,
+      "loss": 0.5981,
+      "step": 9010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987067315511486e-05,
+      "loss": 0.5706,
+      "step": 9011
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987058961517587e-05,
+      "loss": 0.6467,
+      "step": 9012
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987050604828132e-05,
+      "loss": 0.6102,
+      "step": 9013
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998704224544313e-05,
+      "loss": 0.5996,
+      "step": 9014
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987033883362577e-05,
+      "loss": 0.5458,
+      "step": 9015
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987025518586474e-05,
+      "loss": 0.5734,
+      "step": 9016
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987017151114832e-05,
+      "loss": 0.463,
+      "step": 9017
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987008780947642e-05,
+      "loss": 0.5474,
+      "step": 9018
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9987000408084913e-05,
+      "loss": 0.5467,
+      "step": 9019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998699203252665e-05,
+      "loss": 0.6184,
+      "step": 9020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986983654272843e-05,
+      "loss": 0.6163,
+      "step": 9021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986975273323507e-05,
+      "loss": 0.5893,
+      "step": 9022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998696688967864e-05,
+      "loss": 0.5163,
+      "step": 9023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986958503338245e-05,
+      "loss": 0.592,
+      "step": 9024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986950114302322e-05,
+      "loss": 0.7406,
+      "step": 9025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986941722570873e-05,
+      "loss": 0.5752,
+      "step": 9026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986933328143905e-05,
+      "loss": 0.5355,
+      "step": 9027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986924931021416e-05,
+      "loss": 0.6527,
+      "step": 9028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986916531203405e-05,
+      "loss": 0.5836,
+      "step": 9029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986908128689887e-05,
+      "loss": 0.5466,
+      "step": 9030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998689972348085e-05,
+      "loss": 0.6264,
+      "step": 9031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.99868913155763e-05,
+      "loss": 0.6532,
+      "step": 9032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986882904976243e-05,
+      "loss": 0.583,
+      "step": 9033
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986874491680683e-05,
+      "loss": 0.5517,
+      "step": 9034
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986866075689618e-05,
+      "loss": 0.6345,
+      "step": 9035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986857657003046e-05,
+      "loss": 0.5961,
+      "step": 9036
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986849235620978e-05,
+      "loss": 0.531,
+      "step": 9037
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986840811543415e-05,
+      "loss": 0.4851,
+      "step": 9038
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986832384770355e-05,
+      "loss": 0.5232,
+      "step": 9039
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986823955301802e-05,
+      "loss": 0.6085,
+      "step": 9040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998681552313776e-05,
+      "loss": 0.5847,
+      "step": 9041
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986807088278227e-05,
+      "loss": 0.6479,
+      "step": 9042
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998679865072321e-05,
+      "loss": 0.6308,
+      "step": 9043
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998679021047271e-05,
+      "loss": 0.6146,
+      "step": 9044
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998678176752673e-05,
+      "loss": 0.6311,
+      "step": 9045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986773321885265e-05,
+      "loss": 0.6478,
+      "step": 9046
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998676487354833e-05,
+      "loss": 0.5964,
+      "step": 9047
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986756422515917e-05,
+      "loss": 0.5748,
+      "step": 9048
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998674796878803e-05,
+      "loss": 0.52,
+      "step": 9049
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986739512364676e-05,
+      "loss": 0.6385,
+      "step": 9050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986731053245854e-05,
+      "loss": 0.6434,
+      "step": 9051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986722591431563e-05,
+      "loss": 0.5497,
+      "step": 9052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986714126921813e-05,
+      "loss": 0.6047,
+      "step": 9053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.99867056597166e-05,
+      "loss": 0.6539,
+      "step": 9054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998669718981593e-05,
+      "loss": 0.5588,
+      "step": 9055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986688717219805e-05,
+      "loss": 0.6033,
+      "step": 9056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986680241928224e-05,
+      "loss": 0.5372,
+      "step": 9057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998667176394119e-05,
+      "loss": 0.526,
+      "step": 9058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986663283258712e-05,
+      "loss": 0.5544,
+      "step": 9059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986654799880782e-05,
+      "loss": 0.6518,
+      "step": 9060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986646313807406e-05,
+      "loss": 0.5924,
+      "step": 9061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986637825038592e-05,
+      "loss": 0.5548,
+      "step": 9062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986629333574333e-05,
+      "loss": 0.5223,
+      "step": 9063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986620839414642e-05,
+      "loss": 0.5138,
+      "step": 9064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998661234255951e-05,
+      "loss": 0.5064,
+      "step": 9065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998660384300895e-05,
+      "loss": 0.545,
+      "step": 9066
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986595340762953e-05,
+      "loss": 0.4828,
+      "step": 9067
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998658683582153e-05,
+      "loss": 0.6524,
+      "step": 9068
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998657832818468e-05,
+      "loss": 0.5912,
+      "step": 9069
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986569817852406e-05,
+      "loss": 0.5852,
+      "step": 9070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998656130482471e-05,
+      "loss": 0.586,
+      "step": 9071
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986552789101594e-05,
+      "loss": 0.5225,
+      "step": 9072
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986544270683062e-05,
+      "loss": 0.6147,
+      "step": 9073
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986535749569116e-05,
+      "loss": 0.5724,
+      "step": 9074
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986527225759755e-05,
+      "loss": 0.646,
+      "step": 9075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986518699254983e-05,
+      "loss": 0.5538,
+      "step": 9076
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986510170054803e-05,
+      "loss": 0.6205,
+      "step": 9077
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998650163815922e-05,
+      "loss": 0.5583,
+      "step": 9078
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986493103568233e-05,
+      "loss": 0.6195,
+      "step": 9079
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986484566281843e-05,
+      "loss": 0.5904,
+      "step": 9080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986476026300058e-05,
+      "loss": 0.5386,
+      "step": 9081
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986467483622872e-05,
+      "loss": 0.5289,
+      "step": 9082
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998645893825029e-05,
+      "loss": 0.5985,
+      "step": 9083
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998645039018232e-05,
+      "loss": 0.6253,
+      "step": 9084
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986441839418962e-05,
+      "loss": 0.7042,
+      "step": 9085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986433285960213e-05,
+      "loss": 0.4736,
+      "step": 9086
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998642472980608e-05,
+      "loss": 0.5021,
+      "step": 9087
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986416170956568e-05,
+      "loss": 0.6145,
+      "step": 9088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998640760941167e-05,
+      "loss": 0.6135,
+      "step": 9089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986399045171392e-05,
+      "loss": 0.5481,
+      "step": 9090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986390478235746e-05,
+      "loss": 0.5325,
+      "step": 9091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998638190860472e-05,
+      "loss": 0.5376,
+      "step": 9092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986373336278326e-05,
+      "loss": 0.6168,
+      "step": 9093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986364761256564e-05,
+      "loss": 0.5553,
+      "step": 9094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986356183539434e-05,
+      "loss": 0.5348,
+      "step": 9095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998634760312694e-05,
+      "loss": 0.5718,
+      "step": 9096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986339020019084e-05,
+      "loss": 0.645,
+      "step": 9097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998633043421587e-05,
+      "loss": 0.5583,
+      "step": 9098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986321845717298e-05,
+      "loss": 0.5002,
+      "step": 9099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998631325452337e-05,
+      "loss": 0.6355,
+      "step": 9100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986304660634087e-05,
+      "loss": 0.6605,
+      "step": 9101
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986296064049458e-05,
+      "loss": 0.4978,
+      "step": 9102
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986287464769482e-05,
+      "loss": 0.5737,
+      "step": 9103
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986278862794155e-05,
+      "loss": 0.5878,
+      "step": 9104
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998627025812349e-05,
+      "loss": 0.4827,
+      "step": 9105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998626165075748e-05,
+      "loss": 0.5095,
+      "step": 9106
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986253040696132e-05,
+      "loss": 0.5518,
+      "step": 9107
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986244427939448e-05,
+      "loss": 0.5487,
+      "step": 9108
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986235812487428e-05,
+      "loss": 0.6942,
+      "step": 9109
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998622719434008e-05,
+      "loss": 0.5663,
+      "step": 9110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986218573497404e-05,
+      "loss": 0.5226,
+      "step": 9111
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986209949959393e-05,
+      "loss": 0.7016,
+      "step": 9112
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9986201323726066e-05,
+      "loss": 0.5478,
+      "step": 9113
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998619269479741e-05,
+      "loss": 0.6208,
+      "step": 9114
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998618406317344e-05,
+      "loss": 0.6359,
+      "step": 9115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998617542885415e-05,
+      "loss": 0.5471,
+      "step": 9116
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.998616679183954e-05,
+      "loss": 0.6066,
+      "step": 9117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998615815212962e-05,
+      "loss": 0.5885,
+      "step": 9118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998614950972439e-05,
+      "loss": 0.6182,
+      "step": 9119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9986140864623854e-05,
+      "loss": 0.5323,
+      "step": 9120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998613221682801e-05,
+      "loss": 0.6145,
+      "step": 9121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998612356633686e-05,
+      "loss": 0.5411,
+      "step": 9122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998611491315041e-05,
+      "loss": 0.5476,
+      "step": 9123
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998610625726866e-05,
+      "loss": 0.5932,
+      "step": 9124
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9986097598691615e-05,
+      "loss": 0.6249,
+      "step": 9125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9986088937419276e-05,
+      "loss": 0.6279,
+      "step": 9126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998608027345164e-05,
+      "loss": 0.6547,
+      "step": 9127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998607160678872e-05,
+      "loss": 0.523,
+      "step": 9128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998606293743051e-05,
+      "loss": 0.5699,
+      "step": 9129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9986054265377014e-05,
+      "loss": 0.6854,
+      "step": 9130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998604559062824e-05,
+      "loss": 0.6477,
+      "step": 9131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998603691318418e-05,
+      "loss": 0.5802,
+      "step": 9132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9986028233044844e-05,
+      "loss": 0.6788,
+      "step": 9133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9986019550210234e-05,
+      "loss": 0.665,
+      "step": 9134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998601086468035e-05,
+      "loss": 0.6434,
+      "step": 9135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998600217645519e-05,
+      "loss": 0.4489,
+      "step": 9136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985993485534768e-05,
+      "loss": 0.5857,
+      "step": 9137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985984791919076e-05,
+      "loss": 0.6289,
+      "step": 9138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985976095608125e-05,
+      "loss": 0.6778,
+      "step": 9139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985967396601908e-05,
+      "loss": 0.5249,
+      "step": 9140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985958694900432e-05,
+      "loss": 0.5135,
+      "step": 9141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985949990503698e-05,
+      "loss": 0.5348,
+      "step": 9142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985941283411708e-05,
+      "loss": 0.6986,
+      "step": 9143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998593257362447e-05,
+      "loss": 0.5551,
+      "step": 9144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985923861141983e-05,
+      "loss": 0.5936,
+      "step": 9145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985915145964247e-05,
+      "loss": 0.5534,
+      "step": 9146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985906428091266e-05,
+      "loss": 0.6185,
+      "step": 9147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998589770752304e-05,
+      "loss": 0.5562,
+      "step": 9148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985888984259574e-05,
+      "loss": 0.6081,
+      "step": 9149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998588025830087e-05,
+      "loss": 0.5281,
+      "step": 9150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985871529646932e-05,
+      "loss": 0.5344,
+      "step": 9151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998586279829776e-05,
+      "loss": 0.6396,
+      "step": 9152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985854064253357e-05,
+      "loss": 0.5167,
+      "step": 9153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985845327513723e-05,
+      "loss": 0.5279,
+      "step": 9154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985836588078864e-05,
+      "loss": 0.6188,
+      "step": 9155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998582784594878e-05,
+      "loss": 0.6423,
+      "step": 9156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985819101123475e-05,
+      "loss": 0.6476,
+      "step": 9157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985810353602955e-05,
+      "loss": 0.5194,
+      "step": 9158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998580160338721e-05,
+      "loss": 0.5355,
+      "step": 9159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985792850476257e-05,
+      "loss": 0.582,
+      "step": 9160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998578409487009e-05,
+      "loss": 0.6261,
+      "step": 9161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985775336568714e-05,
+      "loss": 0.5981,
+      "step": 9162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998576657557213e-05,
+      "loss": 0.556,
+      "step": 9163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998575781188034e-05,
+      "loss": 0.6248,
+      "step": 9164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985749045493348e-05,
+      "loss": 0.6528,
+      "step": 9165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985740276411154e-05,
+      "loss": 0.5634,
+      "step": 9166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985731504633765e-05,
+      "loss": 0.6451,
+      "step": 9167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998572273016118e-05,
+      "loss": 0.6155,
+      "step": 9168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.99857139529934e-05,
+      "loss": 0.5985,
+      "step": 9169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998570517313043e-05,
+      "loss": 0.5739,
+      "step": 9170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985696390572272e-05,
+      "loss": 0.5259,
+      "step": 9171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985687605318928e-05,
+      "loss": 0.5867,
+      "step": 9172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985678817370396e-05,
+      "loss": 0.6004,
+      "step": 9173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998567002672669e-05,
+      "loss": 0.5681,
+      "step": 9174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.99856612333878e-05,
+      "loss": 0.5265,
+      "step": 9175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985652437353732e-05,
+      "loss": 0.5253,
+      "step": 9176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985643638624493e-05,
+      "loss": 0.6627,
+      "step": 9177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998563483720008e-05,
+      "loss": 0.6889,
+      "step": 9178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.99856260330805e-05,
+      "loss": 0.6053,
+      "step": 9179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985617226265753e-05,
+      "loss": 0.6596,
+      "step": 9180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998560841675584e-05,
+      "loss": 0.6014,
+      "step": 9181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985599604550763e-05,
+      "loss": 0.5612,
+      "step": 9182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985590789650527e-05,
+      "loss": 0.5607,
+      "step": 9183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985581972055134e-05,
+      "loss": 0.5459,
+      "step": 9184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985573151764587e-05,
+      "loss": 0.5793,
+      "step": 9185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985564328778883e-05,
+      "loss": 0.6622,
+      "step": 9186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985555503098032e-05,
+      "loss": 0.6293,
+      "step": 9187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998554667472203e-05,
+      "loss": 0.5625,
+      "step": 9188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985537843650885e-05,
+      "loss": 0.5339,
+      "step": 9189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985529009884596e-05,
+      "loss": 0.5352,
+      "step": 9190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985520173423167e-05,
+      "loss": 0.5506,
+      "step": 9191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985511334266598e-05,
+      "loss": 0.5508,
+      "step": 9192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985502492414895e-05,
+      "loss": 0.4953,
+      "step": 9193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985493647868055e-05,
+      "loss": 0.5243,
+      "step": 9194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985484800626086e-05,
+      "loss": 0.6257,
+      "step": 9195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985475950688986e-05,
+      "loss": 0.5257,
+      "step": 9196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985467098056762e-05,
+      "loss": 0.5076,
+      "step": 9197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985458242729413e-05,
+      "loss": 0.5278,
+      "step": 9198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998544938470694e-05,
+      "loss": 0.5944,
+      "step": 9199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985440523989346e-05,
+      "loss": 0.6216,
+      "step": 9200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998543166057664e-05,
+      "loss": 0.4682,
+      "step": 9201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985422794468817e-05,
+      "loss": 0.6109,
+      "step": 9202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998541392566588e-05,
+      "loss": 0.5228,
+      "step": 9203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985405054167835e-05,
+      "loss": 0.5639,
+      "step": 9204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985396179974687e-05,
+      "loss": 0.5812,
+      "step": 9205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985387303086425e-05,
+      "loss": 0.5955,
+      "step": 9206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985378423503067e-05,
+      "loss": 0.5767,
+      "step": 9207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985369541224606e-05,
+      "loss": 0.6732,
+      "step": 9208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998536065625105e-05,
+      "loss": 0.5264,
+      "step": 9209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998535176858239e-05,
+      "loss": 0.6366,
+      "step": 9210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985342878218645e-05,
+      "loss": 0.6022,
+      "step": 9211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985333985159806e-05,
+      "loss": 0.6404,
+      "step": 9212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.998532508940588e-05,
+      "loss": 0.6525,
+      "step": 9213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985316190956866e-05,
+      "loss": 0.6606,
+      "step": 9214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985307289812773e-05,
+      "loss": 0.5766,
+      "step": 9215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9985298385973596e-05,
+      "loss": 0.5914,
+      "step": 9216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998528947943934e-05,
+      "loss": 0.5574,
+      "step": 9217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998528057021001e-05,
+      "loss": 0.5727,
+      "step": 9218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.99852716582856e-05,
+      "loss": 0.5488,
+      "step": 9219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985262743666124e-05,
+      "loss": 0.6183,
+      "step": 9220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998525382635158e-05,
+      "loss": 0.5826,
+      "step": 9221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985244906341964e-05,
+      "loss": 0.6983,
+      "step": 9222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985235983637287e-05,
+      "loss": 0.5071,
+      "step": 9223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985227058237547e-05,
+      "loss": 0.4913,
+      "step": 9224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998521813014275e-05,
+      "loss": 0.6318,
+      "step": 9225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985209199352895e-05,
+      "loss": 0.6447,
+      "step": 9226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985200265867982e-05,
+      "loss": 0.6658,
+      "step": 9227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998519132968802e-05,
+      "loss": 0.6023,
+      "step": 9228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998518239081301e-05,
+      "loss": 0.5583,
+      "step": 9229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985173449242947e-05,
+      "loss": 0.5602,
+      "step": 9230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985164504977846e-05,
+      "loss": 0.596,
+      "step": 9231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985155558017696e-05,
+      "loss": 0.52,
+      "step": 9232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998514660836251e-05,
+      "loss": 0.5319,
+      "step": 9233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985137656012283e-05,
+      "loss": 0.5885,
+      "step": 9234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998512870096702e-05,
+      "loss": 0.5937,
+      "step": 9235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998511974322673e-05,
+      "loss": 0.4775,
+      "step": 9236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985110782791407e-05,
+      "loss": 0.7073,
+      "step": 9237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985101819661056e-05,
+      "loss": 0.4939,
+      "step": 9238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985092853835676e-05,
+      "loss": 0.5771,
+      "step": 9239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985083885315275e-05,
+      "loss": 0.4835,
+      "step": 9240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985074914099852e-05,
+      "loss": 0.647,
+      "step": 9241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998506594018941e-05,
+      "loss": 0.5619,
+      "step": 9242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985056963583958e-05,
+      "loss": 0.6166,
+      "step": 9243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985047984283484e-05,
+      "loss": 0.5709,
+      "step": 9244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985039002288005e-05,
+      "loss": 0.5209,
+      "step": 9245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985030017597514e-05,
+      "loss": 0.7662,
+      "step": 9246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9985021030212022e-05,
+      "loss": 0.5663,
+      "step": 9247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998501204013152e-05,
+      "loss": 0.5617,
+      "step": 9248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998500304735602e-05,
+      "loss": 0.534,
+      "step": 9249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998499405188552e-05,
+      "loss": 0.6191,
+      "step": 9250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984985053720024e-05,
+      "loss": 0.6415,
+      "step": 9251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984976052859532e-05,
+      "loss": 0.5403,
+      "step": 9252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998496704930405e-05,
+      "loss": 0.6492,
+      "step": 9253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984958043053578e-05,
+      "loss": 0.5239,
+      "step": 9254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998494903410812e-05,
+      "loss": 0.579,
+      "step": 9255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984940022467676e-05,
+      "loss": 0.6391,
+      "step": 9256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984931008132252e-05,
+      "loss": 0.5559,
+      "step": 9257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984921991101847e-05,
+      "loss": 0.6074,
+      "step": 9258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984912971376464e-05,
+      "loss": 0.5751,
+      "step": 9259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998490394895611e-05,
+      "loss": 0.6664,
+      "step": 9260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998489492384078e-05,
+      "loss": 0.5846,
+      "step": 9261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998488589603048e-05,
+      "loss": 0.6309,
+      "step": 9262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984876865525215e-05,
+      "loss": 0.6582,
+      "step": 9263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984867832324985e-05,
+      "loss": 0.524,
+      "step": 9264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998485879642979e-05,
+      "loss": 0.5543,
+      "step": 9265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998484975783964e-05,
+      "loss": 0.6168,
+      "step": 9266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984840716554527e-05,
+      "loss": 0.7065,
+      "step": 9267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998483167257446e-05,
+      "loss": 0.6175,
+      "step": 9268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984822625899444e-05,
+      "loss": 0.5692,
+      "step": 9269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998481357652947e-05,
+      "loss": 0.583,
+      "step": 9270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984804524464552e-05,
+      "loss": 0.5562,
+      "step": 9271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998479546970469e-05,
+      "loss": 0.6263,
+      "step": 9272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998478641224989e-05,
+      "loss": 0.5561,
+      "step": 9273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984777352100144e-05,
+      "loss": 0.5284,
+      "step": 9274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984768289255458e-05,
+      "loss": 0.5413,
+      "step": 9275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984759223715838e-05,
+      "loss": 0.52,
+      "step": 9276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984750155481282e-05,
+      "loss": 0.5601,
+      "step": 9277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.99847410845518e-05,
+      "loss": 0.663,
+      "step": 9278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998473201092739e-05,
+      "loss": 0.5914,
+      "step": 9279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984722934608052e-05,
+      "loss": 0.6025,
+      "step": 9280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984713855593788e-05,
+      "loss": 0.5102,
+      "step": 9281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984704773884608e-05,
+      "loss": 0.4843,
+      "step": 9282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984695689480508e-05,
+      "loss": 0.5658,
+      "step": 9283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998468660238149e-05,
+      "loss": 0.5436,
+      "step": 9284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998467751258756e-05,
+      "loss": 0.4466,
+      "step": 9285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984668420098717e-05,
+      "loss": 0.5496,
+      "step": 9286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998465932491497e-05,
+      "loss": 0.589,
+      "step": 9287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998465022703631e-05,
+      "loss": 0.5747,
+      "step": 9288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998464112646275e-05,
+      "loss": 0.6655,
+      "step": 9289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998463202319429e-05,
+      "loss": 0.5663,
+      "step": 9290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998462291723093e-05,
+      "loss": 0.4689,
+      "step": 9291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984613808572675e-05,
+      "loss": 0.5342,
+      "step": 9292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984604697219523e-05,
+      "loss": 0.5739,
+      "step": 9293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984595583171482e-05,
+      "loss": 0.5671,
+      "step": 9294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984586466428552e-05,
+      "loss": 0.6003,
+      "step": 9295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984577346990732e-05,
+      "loss": 0.4958,
+      "step": 9296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984568224858033e-05,
+      "loss": 0.5291,
+      "step": 9297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984559100030447e-05,
+      "loss": 0.5718,
+      "step": 9298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984549972507985e-05,
+      "loss": 0.5165,
+      "step": 9299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984540842290645e-05,
+      "loss": 0.6198,
+      "step": 9300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998453170937843e-05,
+      "loss": 0.5345,
+      "step": 9301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998452257377135e-05,
+      "loss": 0.6507,
+      "step": 9302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984513435469393e-05,
+      "loss": 0.5806,
+      "step": 9303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984504294472572e-05,
+      "loss": 0.5225,
+      "step": 9304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984495150780885e-05,
+      "loss": 0.461,
+      "step": 9305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984486004394336e-05,
+      "loss": 0.6545,
+      "step": 9306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984476855312927e-05,
+      "loss": 0.5708,
+      "step": 9307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984467703536667e-05,
+      "loss": 0.576,
+      "step": 9308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984458549065547e-05,
+      "loss": 0.5076,
+      "step": 9309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984449391899575e-05,
+      "loss": 0.5887,
+      "step": 9310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984440232038754e-05,
+      "loss": 0.5976,
+      "step": 9311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.998443106948309e-05,
+      "loss": 0.5767,
+      "step": 9312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984421904232576e-05,
+      "loss": 0.6786,
+      "step": 9313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984412736287222e-05,
+      "loss": 0.5595,
+      "step": 9314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998440356564703e-05,
+      "loss": 0.5811,
+      "step": 9315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984394392311998e-05,
+      "loss": 0.6098,
+      "step": 9316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984385216282135e-05,
+      "loss": 0.5997,
+      "step": 9317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984376037557433e-05,
+      "loss": 0.5769,
+      "step": 9318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984366856137906e-05,
+      "loss": 0.6049,
+      "step": 9319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984357672023554e-05,
+      "loss": 0.5499,
+      "step": 9320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984348485214373e-05,
+      "loss": 0.5314,
+      "step": 9321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998433929571037e-05,
+      "loss": 0.4718,
+      "step": 9322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984330103511546e-05,
+      "loss": 0.5926,
+      "step": 9323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998432090861791e-05,
+      "loss": 0.5707,
+      "step": 9324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984311711029452e-05,
+      "loss": 0.5292,
+      "step": 9325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998430251074619e-05,
+      "loss": 0.5478,
+      "step": 9326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984293307768112e-05,
+      "loss": 0.4989,
+      "step": 9327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984284102095226e-05,
+      "loss": 0.6055,
+      "step": 9328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998427489372754e-05,
+      "loss": 0.4956,
+      "step": 9329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984265682665047e-05,
+      "loss": 0.5862,
+      "step": 9330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984256468907754e-05,
+      "loss": 0.4781,
+      "step": 9331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998424725245567e-05,
+      "loss": 0.6141,
+      "step": 9332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984238033308783e-05,
+      "loss": 0.5655,
+      "step": 9333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984228811467105e-05,
+      "loss": 0.5185,
+      "step": 9334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984219586930643e-05,
+      "loss": 0.5722,
+      "step": 9335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984210359699387e-05,
+      "loss": 0.5652,
+      "step": 9336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984201129773346e-05,
+      "loss": 0.6342,
+      "step": 9337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984191897152527e-05,
+      "loss": 0.6744,
+      "step": 9338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984182661836923e-05,
+      "loss": 0.5636,
+      "step": 9339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984173423826546e-05,
+      "loss": 0.4995,
+      "step": 9340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998416418312139e-05,
+      "loss": 0.6364,
+      "step": 9341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984154939721464e-05,
+      "loss": 0.5601,
+      "step": 9342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984145693626767e-05,
+      "loss": 0.5183,
+      "step": 9343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.99841364448373e-05,
+      "loss": 0.6955,
+      "step": 9344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984127193353073e-05,
+      "loss": 0.5128,
+      "step": 9345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998411793917408e-05,
+      "loss": 0.5838,
+      "step": 9346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998410868230033e-05,
+      "loss": 0.6225,
+      "step": 9347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998409942273182e-05,
+      "loss": 0.5142,
+      "step": 9348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984090160468552e-05,
+      "loss": 0.6658,
+      "step": 9349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984080895510536e-05,
+      "loss": 0.5986,
+      "step": 9350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998407162785777e-05,
+      "loss": 0.5457,
+      "step": 9351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984062357510252e-05,
+      "loss": 0.5258,
+      "step": 9352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984053084467994e-05,
+      "loss": 0.5832,
+      "step": 9353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998404380873099e-05,
+      "loss": 0.5123,
+      "step": 9354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984034530299244e-05,
+      "loss": 0.524,
+      "step": 9355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984025249172765e-05,
+      "loss": 0.6043,
+      "step": 9356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984015965351552e-05,
+      "loss": 0.5397,
+      "step": 9357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9984006678835603e-05,
+      "loss": 0.5731,
+      "step": 9358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983997389624922e-05,
+      "loss": 0.5826,
+      "step": 9359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998398809771952e-05,
+      "loss": 0.5967,
+      "step": 9360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983978803119385e-05,
+      "loss": 0.5898,
+      "step": 9361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998396950582453e-05,
+      "loss": 0.6104,
+      "step": 9362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983960205834957e-05,
+      "loss": 0.63,
+      "step": 9363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983950903150666e-05,
+      "loss": 0.5143,
+      "step": 9364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998394159777166e-05,
+      "loss": 0.6425,
+      "step": 9365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983932289697944e-05,
+      "loss": 0.5691,
+      "step": 9366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983922978929513e-05,
+      "loss": 0.5987,
+      "step": 9367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983913665466376e-05,
+      "loss": 0.5159,
+      "step": 9368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983904349308535e-05,
+      "loss": 0.5524,
+      "step": 9369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998389503045599e-05,
+      "loss": 0.69,
+      "step": 9370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998388570890875e-05,
+      "loss": 0.5779,
+      "step": 9371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983876384666807e-05,
+      "loss": 0.5791,
+      "step": 9372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998386705773017e-05,
+      "loss": 0.5691,
+      "step": 9373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983857728098845e-05,
+      "loss": 0.6487,
+      "step": 9374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983848395772826e-05,
+      "loss": 0.5604,
+      "step": 9375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998383906075212e-05,
+      "loss": 0.5948,
+      "step": 9376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998382972303673e-05,
+      "loss": 0.587,
+      "step": 9377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983820382626657e-05,
+      "loss": 0.6565,
+      "step": 9378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983811039521903e-05,
+      "loss": 0.7047,
+      "step": 9379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983801693722477e-05,
+      "loss": 0.5681,
+      "step": 9380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998379234522837e-05,
+      "loss": 0.6104,
+      "step": 9381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983782994039592e-05,
+      "loss": 0.5993,
+      "step": 9382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983773640156147e-05,
+      "loss": 0.5759,
+      "step": 9383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983764283578037e-05,
+      "loss": 0.551,
+      "step": 9384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983754924305257e-05,
+      "loss": 0.5205,
+      "step": 9385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983745562337814e-05,
+      "loss": 0.6619,
+      "step": 9386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983736197675716e-05,
+      "loss": 0.5938,
+      "step": 9387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998372683031896e-05,
+      "loss": 0.6482,
+      "step": 9388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983717460267546e-05,
+      "loss": 0.5684,
+      "step": 9389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983708087521483e-05,
+      "loss": 0.5961,
+      "step": 9390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998369871208077e-05,
+      "loss": 0.5611,
+      "step": 9391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983689333945412e-05,
+      "loss": 0.6264,
+      "step": 9392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983679953115403e-05,
+      "loss": 0.678,
+      "step": 9393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998367056959076e-05,
+      "loss": 0.624,
+      "step": 9394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983661183371476e-05,
+      "loss": 0.5679,
+      "step": 9395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998365179445755e-05,
+      "loss": 0.6365,
+      "step": 9396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983642402848996e-05,
+      "loss": 0.548,
+      "step": 9397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983633008545807e-05,
+      "loss": 0.628,
+      "step": 9398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998362361154799e-05,
+      "loss": 0.6435,
+      "step": 9399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983614211855545e-05,
+      "loss": 0.6249,
+      "step": 9400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983604809468477e-05,
+      "loss": 0.6117,
+      "step": 9401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983595404386786e-05,
+      "loss": 0.7448,
+      "step": 9402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983585996610477e-05,
+      "loss": 0.5691,
+      "step": 9403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998357658613955e-05,
+      "loss": 0.4969,
+      "step": 9404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998356717297401e-05,
+      "loss": 0.5693,
+      "step": 9405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998355775711386e-05,
+      "loss": 0.6202,
+      "step": 9406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983548338559097e-05,
+      "loss": 0.6421,
+      "step": 9407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983538917309733e-05,
+      "loss": 0.5087,
+      "step": 9408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983529493365764e-05,
+      "loss": 0.5934,
+      "step": 9409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998352006672719e-05,
+      "loss": 0.5964,
+      "step": 9410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.998351063739402e-05,
+      "loss": 0.5807,
+      "step": 9411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983501205366257e-05,
+      "loss": 0.5554,
+      "step": 9412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9983491770643896e-05,
+      "loss": 0.6072,
+      "step": 9413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983482333226943e-05,
+      "loss": 0.5294,
+      "step": 9414
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983472893115405e-05,
+      "loss": 0.4617,
+      "step": 9415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983463450309276e-05,
+      "loss": 0.5463,
+      "step": 9416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983454004808566e-05,
+      "loss": 0.5642,
+      "step": 9417
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983444556613277e-05,
+      "loss": 0.6384,
+      "step": 9418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998343510572341e-05,
+      "loss": 0.7143,
+      "step": 9419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983425652138964e-05,
+      "loss": 0.5987,
+      "step": 9420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998341619585995e-05,
+      "loss": 0.6607,
+      "step": 9421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998340673688636e-05,
+      "loss": 0.6292,
+      "step": 9422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983397275218204e-05,
+      "loss": 0.6316,
+      "step": 9423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983387810855483e-05,
+      "loss": 0.5997,
+      "step": 9424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983378343798197e-05,
+      "loss": 0.5082,
+      "step": 9425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998336887404635e-05,
+      "loss": 0.5883,
+      "step": 9426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998335940159995e-05,
+      "loss": 0.5819,
+      "step": 9427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998334992645899e-05,
+      "loss": 0.6333,
+      "step": 9428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983340448623477e-05,
+      "loss": 0.497,
+      "step": 9429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983330968093414e-05,
+      "loss": 0.6091,
+      "step": 9430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983321484868804e-05,
+      "loss": 0.5349,
+      "step": 9431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998331199894965e-05,
+      "loss": 0.5179,
+      "step": 9432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983302510335952e-05,
+      "loss": 0.677,
+      "step": 9433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983293019027717e-05,
+      "loss": 0.5733,
+      "step": 9434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983283525024942e-05,
+      "loss": 0.5353,
+      "step": 9435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998327402832763e-05,
+      "loss": 0.5525,
+      "step": 9436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998326452893579e-05,
+      "loss": 0.637,
+      "step": 9437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998325502684942e-05,
+      "loss": 0.5607,
+      "step": 9438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998324552206852e-05,
+      "loss": 0.5608,
+      "step": 9439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983236014593097e-05,
+      "loss": 0.6605,
+      "step": 9440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998322650442315e-05,
+      "loss": 0.5757,
+      "step": 9441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983216991558685e-05,
+      "loss": 0.6074,
+      "step": 9442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983207475999704e-05,
+      "loss": 0.621,
+      "step": 9443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983197957746205e-05,
+      "loss": 0.5741,
+      "step": 9444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983188436798197e-05,
+      "loss": 0.5268,
+      "step": 9445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983178913155682e-05,
+      "loss": 0.5254,
+      "step": 9446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983169386818657e-05,
+      "loss": 0.4988,
+      "step": 9447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983159857787128e-05,
+      "loss": 0.5615,
+      "step": 9448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.99831503260611e-05,
+      "loss": 0.6595,
+      "step": 9449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983140791640568e-05,
+      "loss": 0.5098,
+      "step": 9450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983131254525547e-05,
+      "loss": 0.5775,
+      "step": 9451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983121714716026e-05,
+      "loss": 0.5521,
+      "step": 9452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983112172212018e-05,
+      "loss": 0.6365,
+      "step": 9453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998310262701352e-05,
+      "loss": 0.5382,
+      "step": 9454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983093079120534e-05,
+      "loss": 0.5213,
+      "step": 9455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983083528533064e-05,
+      "loss": 0.608,
+      "step": 9456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983073975251118e-05,
+      "loss": 0.5198,
+      "step": 9457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998306441927469e-05,
+      "loss": 0.6131,
+      "step": 9458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983054860603787e-05,
+      "loss": 0.5286,
+      "step": 9459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983045299238412e-05,
+      "loss": 0.5567,
+      "step": 9460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998303573517856e-05,
+      "loss": 0.5999,
+      "step": 9461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983026168424247e-05,
+      "loss": 0.6435,
+      "step": 9462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983016598975468e-05,
+      "loss": 0.5924,
+      "step": 9463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9983007026832222e-05,
+      "loss": 0.6264,
+      "step": 9464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982997451994518e-05,
+      "loss": 0.5396,
+      "step": 9465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998298787446236e-05,
+      "loss": 0.5525,
+      "step": 9466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982978294235744e-05,
+      "loss": 0.5501,
+      "step": 9467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982968711314675e-05,
+      "loss": 0.5443,
+      "step": 9468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982959125699153e-05,
+      "loss": 0.5722,
+      "step": 9469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998294953738919e-05,
+      "loss": 0.6076,
+      "step": 9470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982939946384777e-05,
+      "loss": 0.5434,
+      "step": 9471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982930352685925e-05,
+      "loss": 0.6944,
+      "step": 9472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998292075629263e-05,
+      "loss": 0.5163,
+      "step": 9473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982911157204902e-05,
+      "loss": 0.5094,
+      "step": 9474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998290155542274e-05,
+      "loss": 0.5492,
+      "step": 9475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982891950946146e-05,
+      "loss": 0.5824,
+      "step": 9476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982882343775117e-05,
+      "loss": 0.6347,
+      "step": 9477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998287273390967e-05,
+      "loss": 0.5137,
+      "step": 9478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982863121349793e-05,
+      "loss": 0.5112,
+      "step": 9479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982853506095497e-05,
+      "loss": 0.5816,
+      "step": 9480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998284388814678e-05,
+      "loss": 0.6119,
+      "step": 9481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982834267503648e-05,
+      "loss": 0.5614,
+      "step": 9482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982824644166104e-05,
+      "loss": 0.5997,
+      "step": 9483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982815018134144e-05,
+      "loss": 0.5406,
+      "step": 9484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982805389407783e-05,
+      "loss": 0.6271,
+      "step": 9485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998279575798701e-05,
+      "loss": 0.6014,
+      "step": 9486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982786123871836e-05,
+      "loss": 0.5869,
+      "step": 9487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998277648706226e-05,
+      "loss": 0.6411,
+      "step": 9488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998276684755829e-05,
+      "loss": 0.4786,
+      "step": 9489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998275720535992e-05,
+      "loss": 0.603,
+      "step": 9490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998274756046716e-05,
+      "loss": 0.547,
+      "step": 9491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982737912880007e-05,
+      "loss": 0.6979,
+      "step": 9492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998272826259847e-05,
+      "loss": 0.5429,
+      "step": 9493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982718609622545e-05,
+      "loss": 0.5168,
+      "step": 9494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982708953952235e-05,
+      "loss": 0.496,
+      "step": 9495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998269929558755e-05,
+      "loss": 0.5989,
+      "step": 9496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998268963452849e-05,
+      "loss": 0.5122,
+      "step": 9497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998267997077505e-05,
+      "loss": 0.508,
+      "step": 9498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998267030432724e-05,
+      "loss": 0.5385,
+      "step": 9499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982660635185062e-05,
+      "loss": 0.5699,
+      "step": 9500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982650963348515e-05,
+      "loss": 0.6428,
+      "step": 9501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982641288817604e-05,
+      "loss": 0.5485,
+      "step": 9502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982631611592332e-05,
+      "loss": 0.565,
+      "step": 9503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982621931672702e-05,
+      "loss": 0.6598,
+      "step": 9504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982612249058715e-05,
+      "loss": 0.5755,
+      "step": 9505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982602563750374e-05,
+      "loss": 0.5815,
+      "step": 9506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998259287574768e-05,
+      "loss": 0.4948,
+      "step": 9507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982583185050642e-05,
+      "loss": 0.5667,
+      "step": 9508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982573491659255e-05,
+      "loss": 0.6795,
+      "step": 9509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982563795573527e-05,
+      "loss": 0.5476,
+      "step": 9510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9982554096793455e-05,
+      "loss": 0.5817,
+      "step": 9511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998254439531905e-05,
+      "loss": 0.5635,
+      "step": 9512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982534691150306e-05,
+      "loss": 0.6529,
+      "step": 9513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998252498428723e-05,
+      "loss": 0.6183,
+      "step": 9514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982515274729822e-05,
+      "loss": 0.6321,
+      "step": 9515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998250556247809e-05,
+      "loss": 0.5165,
+      "step": 9516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998249584753203e-05,
+      "loss": 0.5025,
+      "step": 9517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982486129891648e-05,
+      "loss": 0.5268,
+      "step": 9518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982476409556945e-05,
+      "loss": 0.6196,
+      "step": 9519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998246668652793e-05,
+      "loss": 0.6387,
+      "step": 9520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982456960804597e-05,
+      "loss": 0.5875,
+      "step": 9521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998244723238695e-05,
+      "loss": 0.5999,
+      "step": 9522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982437501274998e-05,
+      "loss": 0.5105,
+      "step": 9523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982427767468738e-05,
+      "loss": 0.5908,
+      "step": 9524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982418030968175e-05,
+      "loss": 0.5532,
+      "step": 9525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998240829177331e-05,
+      "loss": 0.538,
+      "step": 9526
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982398549884148e-05,
+      "loss": 0.5335,
+      "step": 9527
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998238880530069e-05,
+      "loss": 0.5657,
+      "step": 9528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982379058022932e-05,
+      "loss": 0.5808,
+      "step": 9529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998236930805089e-05,
+      "loss": 0.5684,
+      "step": 9530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998235955538456e-05,
+      "loss": 0.5628,
+      "step": 9531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998234980002394e-05,
+      "loss": 0.6248,
+      "step": 9532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998234004196904e-05,
+      "loss": 0.5379,
+      "step": 9533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998233028121986e-05,
+      "loss": 0.5921,
+      "step": 9534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982320517776402e-05,
+      "loss": 0.5861,
+      "step": 9535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982310751638668e-05,
+      "loss": 0.4839,
+      "step": 9536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998230098280666e-05,
+      "loss": 0.5664,
+      "step": 9537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982291211280387e-05,
+      "loss": 0.5141,
+      "step": 9538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982281437059844e-05,
+      "loss": 0.5617,
+      "step": 9539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982271660145035e-05,
+      "loss": 0.6383,
+      "step": 9540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998226188053597e-05,
+      "loss": 0.5967,
+      "step": 9541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982252098232642e-05,
+      "loss": 0.6656,
+      "step": 9542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998224231323506e-05,
+      "loss": 0.6154,
+      "step": 9543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998223252554322e-05,
+      "loss": 0.619,
+      "step": 9544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998222273515713e-05,
+      "loss": 0.5263,
+      "step": 9545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982212942076792e-05,
+      "loss": 0.566,
+      "step": 9546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998220314630221e-05,
+      "loss": 0.4843,
+      "step": 9547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998219334783338e-05,
+      "loss": 0.5033,
+      "step": 9548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982183546670314e-05,
+      "loss": 0.5877,
+      "step": 9549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982173742813006e-05,
+      "loss": 0.5817,
+      "step": 9550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982163936261467e-05,
+      "loss": 0.572,
+      "step": 9551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998215412701569e-05,
+      "loss": 0.5151,
+      "step": 9552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998214431507569e-05,
+      "loss": 0.6208,
+      "step": 9553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998213450044146e-05,
+      "loss": 0.53,
+      "step": 9554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982124683113002e-05,
+      "loss": 0.5636,
+      "step": 9555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998211486309032e-05,
+      "loss": 0.5679,
+      "step": 9556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982105040373426e-05,
+      "loss": 0.6244,
+      "step": 9557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998209521496231e-05,
+      "loss": 0.5518,
+      "step": 9558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982085386856985e-05,
+      "loss": 0.5691,
+      "step": 9559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982075556057443e-05,
+      "loss": 0.6051,
+      "step": 9560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982065722563692e-05,
+      "loss": 0.5372,
+      "step": 9561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998205588637574e-05,
+      "loss": 0.5971,
+      "step": 9562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998204604749358e-05,
+      "loss": 0.6255,
+      "step": 9563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982036205917217e-05,
+      "loss": 0.6447,
+      "step": 9564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982026361646664e-05,
+      "loss": 0.5539,
+      "step": 9565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998201651468191e-05,
+      "loss": 0.5822,
+      "step": 9566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9982006665022964e-05,
+      "loss": 0.5015,
+      "step": 9567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981996812669825e-05,
+      "loss": 0.6093,
+      "step": 9568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981986957622502e-05,
+      "loss": 0.6045,
+      "step": 9569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981977099880993e-05,
+      "loss": 0.5644,
+      "step": 9570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.99819672394453e-05,
+      "loss": 0.648,
+      "step": 9571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998195737631543e-05,
+      "loss": 0.579,
+      "step": 9572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981947510491383e-05,
+      "loss": 0.5306,
+      "step": 9573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998193764197316e-05,
+      "loss": 0.6191,
+      "step": 9574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981927770760764e-05,
+      "loss": 0.5315,
+      "step": 9575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981917896854205e-05,
+      "loss": 0.5262,
+      "step": 9576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981908020253474e-05,
+      "loss": 0.54,
+      "step": 9577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981898140958582e-05,
+      "loss": 0.6185,
+      "step": 9578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998188825896953e-05,
+      "loss": 0.5408,
+      "step": 9579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981878374286318e-05,
+      "loss": 0.6798,
+      "step": 9580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998186848690895e-05,
+      "loss": 0.5665,
+      "step": 9581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998185859683743e-05,
+      "loss": 0.4752,
+      "step": 9582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981848704071757e-05,
+      "loss": 0.5093,
+      "step": 9583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998183880861194e-05,
+      "loss": 0.63,
+      "step": 9584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981828910457974e-05,
+      "loss": 0.5521,
+      "step": 9585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981819009609868e-05,
+      "loss": 0.699,
+      "step": 9586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998180910606762e-05,
+      "loss": 0.4709,
+      "step": 9587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998179919983124e-05,
+      "loss": 0.6582,
+      "step": 9588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981789290900725e-05,
+      "loss": 0.5669,
+      "step": 9589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981779379276077e-05,
+      "loss": 0.5262,
+      "step": 9590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.99817694649573e-05,
+      "loss": 0.5282,
+      "step": 9591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981759547944394e-05,
+      "loss": 0.5494,
+      "step": 9592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998174962823737e-05,
+      "loss": 0.5866,
+      "step": 9593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998173970583622e-05,
+      "loss": 0.6252,
+      "step": 9594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981729780740955e-05,
+      "loss": 0.4715,
+      "step": 9595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981719852951574e-05,
+      "loss": 0.5074,
+      "step": 9596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998170992246808e-05,
+      "loss": 0.6137,
+      "step": 9597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981699989290477e-05,
+      "loss": 0.7769,
+      "step": 9598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981690053418764e-05,
+      "loss": 0.5834,
+      "step": 9599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981680114852945e-05,
+      "loss": 0.5871,
+      "step": 9600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981670173593026e-05,
+      "loss": 0.5093,
+      "step": 9601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981660229639007e-05,
+      "loss": 0.5053,
+      "step": 9602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998165028299089e-05,
+      "loss": 0.5904,
+      "step": 9603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981640333648685e-05,
+      "loss": 0.6571,
+      "step": 9604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981630381612382e-05,
+      "loss": 0.514,
+      "step": 9605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981620426881993e-05,
+      "loss": 0.5459,
+      "step": 9606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998161046945752e-05,
+      "loss": 0.5483,
+      "step": 9607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.998160050933896e-05,
+      "loss": 0.6041,
+      "step": 9608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981590546526323e-05,
+      "loss": 0.5356,
+      "step": 9609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9981580581019605e-05,
+      "loss": 0.5917,
+      "step": 9610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998157061281881e-05,
+      "loss": 0.4835,
+      "step": 9611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981560641923945e-05,
+      "loss": 0.5464,
+      "step": 9612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998155066833501e-05,
+      "loss": 0.5624,
+      "step": 9613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981540692052008e-05,
+      "loss": 0.6016,
+      "step": 9614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998153071307494e-05,
+      "loss": 0.5869,
+      "step": 9615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981520731403812e-05,
+      "loss": 0.6809,
+      "step": 9616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981510747038627e-05,
+      "loss": 0.5226,
+      "step": 9617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981500759979382e-05,
+      "loss": 0.6677,
+      "step": 9618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981490770226084e-05,
+      "loss": 0.6928,
+      "step": 9619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981480777778738e-05,
+      "loss": 0.5863,
+      "step": 9620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981470782637338e-05,
+      "loss": 0.5566,
+      "step": 9621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981460784801896e-05,
+      "loss": 0.5869,
+      "step": 9622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998145078427241e-05,
+      "loss": 0.6207,
+      "step": 9623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981440781048885e-05,
+      "loss": 0.5886,
+      "step": 9624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981430775131322e-05,
+      "loss": 0.4586,
+      "step": 9625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981420766519724e-05,
+      "loss": 0.6048,
+      "step": 9626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981410755214093e-05,
+      "loss": 0.4989,
+      "step": 9627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981400741214434e-05,
+      "loss": 0.5524,
+      "step": 9628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998139072452075e-05,
+      "loss": 0.6424,
+      "step": 9629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981380705133038e-05,
+      "loss": 0.7196,
+      "step": 9630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981370683051305e-05,
+      "loss": 0.5981,
+      "step": 9631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981360658275553e-05,
+      "loss": 0.6082,
+      "step": 9632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981350630805786e-05,
+      "loss": 0.5306,
+      "step": 9633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998134060064201e-05,
+      "loss": 0.6367,
+      "step": 9634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998133056778422e-05,
+      "loss": 0.5922,
+      "step": 9635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998132053223242e-05,
+      "loss": 0.6091,
+      "step": 9636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998131049398662e-05,
+      "loss": 0.5968,
+      "step": 9637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998130045304681e-05,
+      "loss": 0.5164,
+      "step": 9638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981290409413006e-05,
+      "loss": 0.4995,
+      "step": 9639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981280363085206e-05,
+      "loss": 0.6109,
+      "step": 9640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998127031406341e-05,
+      "loss": 0.5433,
+      "step": 9641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981260262347624e-05,
+      "loss": 0.5085,
+      "step": 9642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998125020793785e-05,
+      "loss": 0.6405,
+      "step": 9643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981240150834088e-05,
+      "loss": 0.5482,
+      "step": 9644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998123009103634e-05,
+      "loss": 0.5939,
+      "step": 9645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981220028544613e-05,
+      "loss": 0.5862,
+      "step": 9646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981209963358908e-05,
+      "loss": 0.6291,
+      "step": 9647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998119989547923e-05,
+      "loss": 0.5274,
+      "step": 9648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981189824905578e-05,
+      "loss": 0.5635,
+      "step": 9649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981179751637956e-05,
+      "loss": 0.5656,
+      "step": 9650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998116967567637e-05,
+      "loss": 0.6137,
+      "step": 9651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981159597020814e-05,
+      "loss": 0.4869,
+      "step": 9652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.99811495156713e-05,
+      "loss": 0.5465,
+      "step": 9653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981139431627825e-05,
+      "loss": 0.5844,
+      "step": 9654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981129344890397e-05,
+      "loss": 0.6333,
+      "step": 9655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998111925545901e-05,
+      "loss": 0.5825,
+      "step": 9656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981109163333677e-05,
+      "loss": 0.5653,
+      "step": 9657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981099068514396e-05,
+      "loss": 0.5988,
+      "step": 9658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981088971001167e-05,
+      "loss": 0.5984,
+      "step": 9659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981078870793997e-05,
+      "loss": 0.6162,
+      "step": 9660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981068767892887e-05,
+      "loss": 0.5797,
+      "step": 9661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981058662297842e-05,
+      "loss": 0.5178,
+      "step": 9662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998104855400886e-05,
+      "loss": 0.5386,
+      "step": 9663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981038443025948e-05,
+      "loss": 0.5861,
+      "step": 9664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981028329349104e-05,
+      "loss": 0.6802,
+      "step": 9665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981018212978337e-05,
+      "loss": 0.4794,
+      "step": 9666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9981008093913642e-05,
+      "loss": 0.5828,
+      "step": 9667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998099797215503e-05,
+      "loss": 0.6447,
+      "step": 9668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.99809878477025e-05,
+      "loss": 0.5948,
+      "step": 9669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980977720556055e-05,
+      "loss": 0.5862,
+      "step": 9670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980967590715696e-05,
+      "loss": 0.5946,
+      "step": 9671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980957458181432e-05,
+      "loss": 0.6098,
+      "step": 9672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980947322953255e-05,
+      "loss": 0.5494,
+      "step": 9673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980937185031174e-05,
+      "loss": 0.6927,
+      "step": 9674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980927044415193e-05,
+      "loss": 0.6023,
+      "step": 9675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980916901105312e-05,
+      "loss": 0.5862,
+      "step": 9676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980906755101538e-05,
+      "loss": 0.5331,
+      "step": 9677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980896606403866e-05,
+      "loss": 0.7193,
+      "step": 9678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998088645501231e-05,
+      "loss": 0.4753,
+      "step": 9679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998087630092686e-05,
+      "loss": 0.5692,
+      "step": 9680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998086614414753e-05,
+      "loss": 0.6224,
+      "step": 9681
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998085598467431e-05,
+      "loss": 0.5816,
+      "step": 9682
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980845822507217e-05,
+      "loss": 0.6418,
+      "step": 9683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980835657646246e-05,
+      "loss": 0.5525,
+      "step": 9684
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.99808254900914e-05,
+      "loss": 0.6305,
+      "step": 9685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980815319842678e-05,
+      "loss": 0.5511,
+      "step": 9686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980805146900094e-05,
+      "loss": 0.543,
+      "step": 9687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980794971263638e-05,
+      "loss": 0.6109,
+      "step": 9688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980784792933325e-05,
+      "loss": 0.6047,
+      "step": 9689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998077461190915e-05,
+      "loss": 0.4848,
+      "step": 9690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980764428191115e-05,
+      "loss": 0.4834,
+      "step": 9691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980754241779227e-05,
+      "loss": 0.4676,
+      "step": 9692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980744052673483e-05,
+      "loss": 0.5399,
+      "step": 9693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980733860873896e-05,
+      "loss": 0.5776,
+      "step": 9694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980723666380457e-05,
+      "loss": 0.6718,
+      "step": 9695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980713469193175e-05,
+      "loss": 0.5465,
+      "step": 9696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998070326931205e-05,
+      "loss": 0.6786,
+      "step": 9697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998069306673709e-05,
+      "loss": 0.5239,
+      "step": 9698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980682861468294e-05,
+      "loss": 0.7141,
+      "step": 9699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980672653505665e-05,
+      "loss": 0.6053,
+      "step": 9700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980662442849207e-05,
+      "loss": 0.5584,
+      "step": 9701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980652229498917e-05,
+      "loss": 0.5535,
+      "step": 9702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980642013454808e-05,
+      "loss": 0.5697,
+      "step": 9703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998063179471687e-05,
+      "loss": 0.627,
+      "step": 9704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980621573285116e-05,
+      "loss": 0.55,
+      "step": 9705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980611349159547e-05,
+      "loss": 0.6332,
+      "step": 9706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980601122340166e-05,
+      "loss": 0.5895,
+      "step": 9707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980590892826973e-05,
+      "loss": 0.611,
+      "step": 9708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9980580660619967e-05,
+      "loss": 0.5615,
+      "step": 9709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998057042571916e-05,
+      "loss": 0.5106,
+      "step": 9710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998056018812455e-05,
+      "loss": 0.5542,
+      "step": 9711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998054994783614e-05,
+      "loss": 0.4909,
+      "step": 9712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980539704853933e-05,
+      "loss": 0.4914,
+      "step": 9713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980529459177932e-05,
+      "loss": 0.4846,
+      "step": 9714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980519210808142e-05,
+      "loss": 0.5759,
+      "step": 9715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980508959744557e-05,
+      "loss": 0.5738,
+      "step": 9716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998049870598719e-05,
+      "loss": 0.6534,
+      "step": 9717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998048844953604e-05,
+      "loss": 0.6358,
+      "step": 9718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980478190391105e-05,
+      "loss": 0.6366,
+      "step": 9719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980467928552397e-05,
+      "loss": 0.5609,
+      "step": 9720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980457664019914e-05,
+      "loss": 0.5648,
+      "step": 9721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980447396793656e-05,
+      "loss": 0.5159,
+      "step": 9722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998043712687363e-05,
+      "loss": 0.5412,
+      "step": 9723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998042685425984e-05,
+      "loss": 0.6026,
+      "step": 9724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980416578952284e-05,
+      "loss": 0.5613,
+      "step": 9725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980406300950964e-05,
+      "loss": 0.4969,
+      "step": 9726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998039602025589e-05,
+      "loss": 0.6587,
+      "step": 9727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980385736867058e-05,
+      "loss": 0.4821,
+      "step": 9728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980375450784474e-05,
+      "loss": 0.5521,
+      "step": 9729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998036516200814e-05,
+      "loss": 0.5952,
+      "step": 9730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980354870538057e-05,
+      "loss": 0.6182,
+      "step": 9731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980344576374234e-05,
+      "loss": 0.6056,
+      "step": 9732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980334279516667e-05,
+      "loss": 0.4881,
+      "step": 9733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980323979965358e-05,
+      "loss": 0.6147,
+      "step": 9734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998031367772032e-05,
+      "loss": 0.5959,
+      "step": 9735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980303372781542e-05,
+      "loss": 0.5885,
+      "step": 9736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980293065149038e-05,
+      "loss": 0.5873,
+      "step": 9737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980282754822803e-05,
+      "loss": 0.5664,
+      "step": 9738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998027244180284e-05,
+      "loss": 0.5791,
+      "step": 9739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980262126089164e-05,
+      "loss": 0.608,
+      "step": 9740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998025180768176e-05,
+      "loss": 0.5566,
+      "step": 9741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980241486580642e-05,
+      "loss": 0.6146,
+      "step": 9742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980231162785814e-05,
+      "loss": 0.6532,
+      "step": 9743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998022083629727e-05,
+      "loss": 0.5649,
+      "step": 9744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998021050711502e-05,
+      "loss": 0.607,
+      "step": 9745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980200175239064e-05,
+      "loss": 0.4859,
+      "step": 9746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980189840669408e-05,
+      "loss": 0.596,
+      "step": 9747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998017950340605e-05,
+      "loss": 0.5659,
+      "step": 9748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980169163448994e-05,
+      "loss": 0.5069,
+      "step": 9749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980158820798244e-05,
+      "loss": 0.6101,
+      "step": 9750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980148475453803e-05,
+      "loss": 0.7373,
+      "step": 9751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980138127415672e-05,
+      "loss": 0.5789,
+      "step": 9752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980127776683855e-05,
+      "loss": 0.4602,
+      "step": 9753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980117423258357e-05,
+      "loss": 0.5233,
+      "step": 9754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980107067139177e-05,
+      "loss": 0.6127,
+      "step": 9755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980096708326323e-05,
+      "loss": 0.5249,
+      "step": 9756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980086346819792e-05,
+      "loss": 0.5505,
+      "step": 9757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998007598261959e-05,
+      "loss": 0.5528,
+      "step": 9758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980065615725715e-05,
+      "loss": 0.6313,
+      "step": 9759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980055246138175e-05,
+      "loss": 0.5167,
+      "step": 9760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980044873856972e-05,
+      "loss": 0.5315,
+      "step": 9761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998003449888211e-05,
+      "loss": 0.5209,
+      "step": 9762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980024121213587e-05,
+      "loss": 0.6082,
+      "step": 9763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.998001374085141e-05,
+      "loss": 0.6488,
+      "step": 9764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980003357795584e-05,
+      "loss": 0.5315,
+      "step": 9765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979992972046103e-05,
+      "loss": 0.5107,
+      "step": 9766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979982583602976e-05,
+      "loss": 0.6033,
+      "step": 9767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997997219246621e-05,
+      "loss": 0.5711,
+      "step": 9768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979961798635797e-05,
+      "loss": 0.559,
+      "step": 9769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997995140211175e-05,
+      "loss": 0.6038,
+      "step": 9770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979941002894065e-05,
+      "loss": 0.5049,
+      "step": 9771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979930600982748e-05,
+      "loss": 0.511,
+      "step": 9772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979920196377802e-05,
+      "loss": 0.6073,
+      "step": 9773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979909789079227e-05,
+      "loss": 0.5655,
+      "step": 9774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997989937908703e-05,
+      "loss": 0.5734,
+      "step": 9775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997988896640121e-05,
+      "loss": 0.6433,
+      "step": 9776
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997987855102177e-05,
+      "loss": 0.4993,
+      "step": 9777
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979868132948717e-05,
+      "loss": 0.4608,
+      "step": 9778
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997985771218205e-05,
+      "loss": 0.4535,
+      "step": 9779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997984728872177e-05,
+      "loss": 0.6296,
+      "step": 9780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979836862567887e-05,
+      "loss": 0.6019,
+      "step": 9781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979826433720397e-05,
+      "loss": 0.5706,
+      "step": 9782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979816002179306e-05,
+      "loss": 0.5452,
+      "step": 9783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979805567944615e-05,
+      "loss": 0.4717,
+      "step": 9784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997979513101633e-05,
+      "loss": 0.4971,
+      "step": 9785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997978469139445e-05,
+      "loss": 0.5726,
+      "step": 9786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997977424907898e-05,
+      "loss": 0.5693,
+      "step": 9787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979763804069922e-05,
+      "loss": 0.5512,
+      "step": 9788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997975335636728e-05,
+      "loss": 0.5137,
+      "step": 9789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997974290597105e-05,
+      "loss": 0.5488,
+      "step": 9790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979732452881248e-05,
+      "loss": 0.6125,
+      "step": 9791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979721997097867e-05,
+      "loss": 0.5173,
+      "step": 9792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997971153862091e-05,
+      "loss": 0.5406,
+      "step": 9793
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979701077450386e-05,
+      "loss": 0.5594,
+      "step": 9794
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979690613586293e-05,
+      "loss": 0.6797,
+      "step": 9795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979680147028636e-05,
+      "loss": 0.6485,
+      "step": 9796
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979669677777414e-05,
+      "loss": 0.5753,
+      "step": 9797
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979659205832634e-05,
+      "loss": 0.5869,
+      "step": 9798
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979648731194296e-05,
+      "loss": 0.5751,
+      "step": 9799
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979638253862406e-05,
+      "loss": 0.5388,
+      "step": 9800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979627773836966e-05,
+      "loss": 0.5972,
+      "step": 9801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979617291117974e-05,
+      "loss": 0.6103,
+      "step": 9802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997960680570544e-05,
+      "loss": 0.4974,
+      "step": 9803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997959631759936e-05,
+      "loss": 0.5206,
+      "step": 9804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9979585826799742e-05,
+      "loss": 0.5737,
+      "step": 9805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.997957533330659e-05,
+      "loss": 0.4905,
+      "step": 9806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.99795648371199e-05,
+      "loss": 0.5437,
+      "step": 9807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979554338239682e-05,
+      "loss": 0.7054,
+      "step": 9808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979543836665933e-05,
+      "loss": 0.5434,
+      "step": 9809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997953333239866e-05,
+      "loss": 0.5648,
+      "step": 9810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979522825437864e-05,
+      "loss": 0.4959,
+      "step": 9811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979512315783546e-05,
+      "loss": 0.6917,
+      "step": 9812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997950180343571e-05,
+      "loss": 0.6167,
+      "step": 9813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979491288394365e-05,
+      "loss": 0.5152,
+      "step": 9814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979480770659506e-05,
+      "loss": 0.594,
+      "step": 9815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997947025023114e-05,
+      "loss": 0.6117,
+      "step": 9816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979459727109266e-05,
+      "loss": 0.5197,
+      "step": 9817
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997944920129389e-05,
+      "loss": 0.5167,
+      "step": 9818
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979438672785017e-05,
+      "loss": 0.7168,
+      "step": 9819
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979428141582642e-05,
+      "loss": 0.5268,
+      "step": 9820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979417607686777e-05,
+      "loss": 0.6327,
+      "step": 9821
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979407071097418e-05,
+      "loss": 0.6561,
+      "step": 9822
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979396531814572e-05,
+      "loss": 0.5906,
+      "step": 9823
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997938598983824e-05,
+      "loss": 0.6905,
+      "step": 9824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979375445168423e-05,
+      "loss": 0.5756,
+      "step": 9825
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979364897805124e-05,
+      "loss": 0.5763,
+      "step": 9826
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997935434774835e-05,
+      "loss": 0.5623,
+      "step": 9827
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979343794998105e-05,
+      "loss": 0.6264,
+      "step": 9828
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979333239554386e-05,
+      "loss": 0.6933,
+      "step": 9829
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979322681417197e-05,
+      "loss": 0.7303,
+      "step": 9830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979312120586544e-05,
+      "loss": 0.6765,
+      "step": 9831
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997930155706243e-05,
+      "loss": 0.5554,
+      "step": 9832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997929099084485e-05,
+      "loss": 0.555,
+      "step": 9833
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997928042193382e-05,
+      "loss": 0.5652,
+      "step": 9834
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979269850329327e-05,
+      "loss": 0.613,
+      "step": 9835
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997925927603139e-05,
+      "loss": 0.5505,
+      "step": 9836
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997924869904e-05,
+      "loss": 0.5895,
+      "step": 9837
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979238119355166e-05,
+      "loss": 0.5755,
+      "step": 9838
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997922753697689e-05,
+      "loss": 0.5369,
+      "step": 9839
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979216951905168e-05,
+      "loss": 0.5929,
+      "step": 9840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979206364140013e-05,
+      "loss": 0.5693,
+      "step": 9841
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979195773681425e-05,
+      "loss": 0.5648,
+      "step": 9842
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979185180529405e-05,
+      "loss": 0.6227,
+      "step": 9843
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979174584683952e-05,
+      "loss": 0.6001,
+      "step": 9844
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979163986145077e-05,
+      "loss": 0.5829,
+      "step": 9845
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979153384912776e-05,
+      "loss": 0.5201,
+      "step": 9846
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997914278098706e-05,
+      "loss": 0.6148,
+      "step": 9847
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979132174367923e-05,
+      "loss": 0.478,
+      "step": 9848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997912156505537e-05,
+      "loss": 0.5517,
+      "step": 9849
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997911095304941e-05,
+      "loss": 0.6224,
+      "step": 9850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979100338350038e-05,
+      "loss": 0.5428,
+      "step": 9851
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979089720957263e-05,
+      "loss": 0.5544,
+      "step": 9852
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979079100871083e-05,
+      "loss": 0.4776,
+      "step": 9853
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979068478091503e-05,
+      "loss": 0.5436,
+      "step": 9854
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979057852618525e-05,
+      "loss": 0.6884,
+      "step": 9855
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979047224452155e-05,
+      "loss": 0.5196,
+      "step": 9856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979036593592393e-05,
+      "loss": 0.5815,
+      "step": 9857
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997902596003924e-05,
+      "loss": 0.4471,
+      "step": 9858
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979015323792704e-05,
+      "loss": 0.5024,
+      "step": 9859
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9979004684852783e-05,
+      "loss": 0.5214,
+      "step": 9860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978994043219483e-05,
+      "loss": 0.4948,
+      "step": 9861
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978983398892806e-05,
+      "loss": 0.4677,
+      "step": 9862
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978972751872753e-05,
+      "loss": 0.4384,
+      "step": 9863
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997896210215933e-05,
+      "loss": 0.5905,
+      "step": 9864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997895144975254e-05,
+      "loss": 0.4869,
+      "step": 9865
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997894079465238e-05,
+      "loss": 0.5071,
+      "step": 9866
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978930136858862e-05,
+      "loss": 0.4781,
+      "step": 9867
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997891947637198e-05,
+      "loss": 0.4736,
+      "step": 9868
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978908813191746e-05,
+      "loss": 0.4289,
+      "step": 9869
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978898147318156e-05,
+      "loss": 0.5272,
+      "step": 9870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997888747875121e-05,
+      "loss": 0.4901,
+      "step": 9871
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978876807490922e-05,
+      "loss": 0.4732,
+      "step": 9872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997886613353729e-05,
+      "loss": 0.4759,
+      "step": 9873
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978855456890307e-05,
+      "loss": 0.5143,
+      "step": 9874
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978844777549987e-05,
+      "loss": 0.4955,
+      "step": 9875
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978834095516337e-05,
+      "loss": 0.4617,
+      "step": 9876
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978823410789345e-05,
+      "loss": 0.5495,
+      "step": 9877
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978812723369026e-05,
+      "loss": 0.5385,
+      "step": 9878
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978802033255375e-05,
+      "loss": 0.4425,
+      "step": 9879
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.99787913404484e-05,
+      "loss": 0.5178,
+      "step": 9880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978780644948106e-05,
+      "loss": 0.4033,
+      "step": 9881
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997876994675449e-05,
+      "loss": 0.5431,
+      "step": 9882
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997875924586756e-05,
+      "loss": 0.4966,
+      "step": 9883
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978748542287312e-05,
+      "loss": 0.4273,
+      "step": 9884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978737836013754e-05,
+      "loss": 0.4722,
+      "step": 9885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978727127046888e-05,
+      "loss": 0.5057,
+      "step": 9886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978716415386722e-05,
+      "loss": 0.4535,
+      "step": 9887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978705701033245e-05,
+      "loss": 0.4844,
+      "step": 9888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978694983986475e-05,
+      "loss": 0.4759,
+      "step": 9889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978684264246407e-05,
+      "loss": 0.4444,
+      "step": 9890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978673541813046e-05,
+      "loss": 0.4225,
+      "step": 9891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978662816686394e-05,
+      "loss": 0.6116,
+      "step": 9892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978652088866456e-05,
+      "loss": 0.4819,
+      "step": 9893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978641358353227e-05,
+      "loss": 0.4858,
+      "step": 9894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997863062514672e-05,
+      "loss": 0.436,
+      "step": 9895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978619889246936e-05,
+      "loss": 0.5445,
+      "step": 9896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978609150653873e-05,
+      "loss": 0.5304,
+      "step": 9897
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978598409367537e-05,
+      "loss": 0.4306,
+      "step": 9898
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978587665387934e-05,
+      "loss": 0.4147,
+      "step": 9899
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978576918715062e-05,
+      "loss": 0.3998,
+      "step": 9900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978566169348923e-05,
+      "loss": 0.494,
+      "step": 9901
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978555417289523e-05,
+      "loss": 0.4756,
+      "step": 9902
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978544662536865e-05,
+      "loss": 0.4601,
+      "step": 9903
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978533905090953e-05,
+      "loss": 0.4137,
+      "step": 9904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9978523144951784e-05,
+      "loss": 0.4594,
+      "step": 9905
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997851238211937e-05,
+      "loss": 0.4438,
+      "step": 9906
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.99785016165937e-05,
+      "loss": 0.577,
+      "step": 9907
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978490848374794e-05,
+      "loss": 0.5185,
+      "step": 9908
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978480077462644e-05,
+      "loss": 0.4417,
+      "step": 9909
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978469303857258e-05,
+      "loss": 0.4403,
+      "step": 9910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978458527558632e-05,
+      "loss": 0.4042,
+      "step": 9911
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978447748566777e-05,
+      "loss": 0.4626,
+      "step": 9912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978436966881692e-05,
+      "loss": 0.4881,
+      "step": 9913
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978426182503382e-05,
+      "loss": 0.6276,
+      "step": 9914
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978415395431842e-05,
+      "loss": 0.5051,
+      "step": 9915
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978404605667086e-05,
+      "loss": 0.4844,
+      "step": 9916
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978393813209108e-05,
+      "loss": 0.4663,
+      "step": 9917
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978383018057918e-05,
+      "loss": 0.5001,
+      "step": 9918
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978372220213518e-05,
+      "loss": 0.5074,
+      "step": 9919
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978361419675906e-05,
+      "loss": 0.4683,
+      "step": 9920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978350616445085e-05,
+      "loss": 0.5247,
+      "step": 9921
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978339810521065e-05,
+      "loss": 0.5424,
+      "step": 9922
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997832900190384e-05,
+      "loss": 0.4773,
+      "step": 9923
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978318190593422e-05,
+      "loss": 0.49,
+      "step": 9924
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997830737658981e-05,
+      "loss": 0.4848,
+      "step": 9925
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978296559893e-05,
+      "loss": 0.4279,
+      "step": 9926
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978285740503006e-05,
+      "loss": 0.4905,
+      "step": 9927
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978274918419824e-05,
+      "loss": 0.5148,
+      "step": 9928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997826409364346e-05,
+      "loss": 0.5246,
+      "step": 9929
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978253266173913e-05,
+      "loss": 0.5294,
+      "step": 9930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997824243601119e-05,
+      "loss": 0.4898,
+      "step": 9931
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978231603155295e-05,
+      "loss": 0.5044,
+      "step": 9932
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978220767606227e-05,
+      "loss": 0.442,
+      "step": 9933
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978209929363993e-05,
+      "loss": 0.4964,
+      "step": 9934
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997819908842859e-05,
+      "loss": 0.4685,
+      "step": 9935
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978188244800028e-05,
+      "loss": 0.4815,
+      "step": 9936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978177398478303e-05,
+      "loss": 0.4341,
+      "step": 9937
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978166549463424e-05,
+      "loss": 0.5569,
+      "step": 9938
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997815569775539e-05,
+      "loss": 0.4299,
+      "step": 9939
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978144843354203e-05,
+      "loss": 0.4006,
+      "step": 9940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997813398625987e-05,
+      "loss": 0.4116,
+      "step": 9941
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997812312647239e-05,
+      "loss": 0.4702,
+      "step": 9942
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978112263991774e-05,
+      "loss": 0.4163,
+      "step": 9943
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978101398818013e-05,
+      "loss": 0.4955,
+      "step": 9944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978090530951118e-05,
+      "loss": 0.5215,
+      "step": 9945
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978079660391088e-05,
+      "loss": 0.4262,
+      "step": 9946
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997806878713793e-05,
+      "loss": 0.4653,
+      "step": 9947
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997805791119164e-05,
+      "loss": 0.4424,
+      "step": 9948
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997804703255223e-05,
+      "loss": 0.4998,
+      "step": 9949
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.99780361512197e-05,
+      "loss": 0.5828,
+      "step": 9950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997802526719405e-05,
+      "loss": 0.5045,
+      "step": 9951
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978014380475283e-05,
+      "loss": 0.493,
+      "step": 9952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9978003491063403e-05,
+      "loss": 0.4013,
+      "step": 9953
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977992598958412e-05,
+      "loss": 0.4745,
+      "step": 9954
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977981704160318e-05,
+      "loss": 0.4136,
+      "step": 9955
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997797080666912e-05,
+      "loss": 0.4331,
+      "step": 9956
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977959906484816e-05,
+      "loss": 0.4427,
+      "step": 9957
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977949003607416e-05,
+      "loss": 0.4781,
+      "step": 9958
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977938098036923e-05,
+      "loss": 0.4597,
+      "step": 9959
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977927189773335e-05,
+      "loss": 0.4621,
+      "step": 9960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977916278816664e-05,
+      "loss": 0.3986,
+      "step": 9961
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977905365166902e-05,
+      "loss": 0.3958,
+      "step": 9962
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977894448824054e-05,
+      "loss": 0.5244,
+      "step": 9963
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977883529788128e-05,
+      "loss": 0.4371,
+      "step": 9964
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997787260805913e-05,
+      "loss": 0.3968,
+      "step": 9965
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997786168363705e-05,
+      "loss": 0.5076,
+      "step": 9966
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.99778507565219e-05,
+      "loss": 0.5385,
+      "step": 9967
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977839826713684e-05,
+      "loss": 0.4307,
+      "step": 9968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.99778288942124e-05,
+      "loss": 0.5305,
+      "step": 9969
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977817959018055e-05,
+      "loss": 0.4036,
+      "step": 9970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997780702113065e-05,
+      "loss": 0.4629,
+      "step": 9971
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997779608055019e-05,
+      "loss": 0.4723,
+      "step": 9972
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977785137276674e-05,
+      "loss": 0.4739,
+      "step": 9973
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977774191310105e-05,
+      "loss": 0.4361,
+      "step": 9974
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977763242650493e-05,
+      "loss": 0.5513,
+      "step": 9975
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997775229129783e-05,
+      "loss": 0.4314,
+      "step": 9976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997774133725213e-05,
+      "loss": 0.4471,
+      "step": 9977
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977730380513392e-05,
+      "loss": 0.4628,
+      "step": 9978
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977719421081615e-05,
+      "loss": 0.4577,
+      "step": 9979
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977708458956805e-05,
+      "loss": 0.5113,
+      "step": 9980
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977697494138962e-05,
+      "loss": 0.4978,
+      "step": 9981
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977686526628097e-05,
+      "loss": 0.4129,
+      "step": 9982
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977675556424205e-05,
+      "loss": 0.4045,
+      "step": 9983
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997766458352729e-05,
+      "loss": 0.5124,
+      "step": 9984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997765360793736e-05,
+      "loss": 0.5144,
+      "step": 9985
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977642629654418e-05,
+      "loss": 0.4695,
+      "step": 9986
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977631648678456e-05,
+      "loss": 0.4998,
+      "step": 9987
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997762066500949e-05,
+      "loss": 0.4761,
+      "step": 9988
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977609678647515e-05,
+      "loss": 0.4576,
+      "step": 9989
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997759868959254e-05,
+      "loss": 0.4681,
+      "step": 9990
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997758769784456e-05,
+      "loss": 0.4558,
+      "step": 9991
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977576703403584e-05,
+      "loss": 0.4048,
+      "step": 9992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997756570626961e-05,
+      "loss": 0.4425,
+      "step": 9993
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997755470644265e-05,
+      "loss": 0.5182,
+      "step": 9994
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.99775437039227e-05,
+      "loss": 0.4147,
+      "step": 9995
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977532698709764e-05,
+      "loss": 0.4054,
+      "step": 9996
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977521690803843e-05,
+      "loss": 0.4666,
+      "step": 9997
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977510680204943e-05,
+      "loss": 0.4209,
+      "step": 9998
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997749966691307e-05,
+      "loss": 0.4531,
+      "step": 9999
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977488650928222e-05,
+      "loss": 0.4828,
+      "step": 10000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977477632250396e-05,
+      "loss": 0.4824,
+      "step": 10001
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.997746661087961e-05,
+      "loss": 0.5224,
+      "step": 10002
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977455586815857e-05,
+      "loss": 0.5228,
+      "step": 10003
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9977444560059143e-05,
+      "loss": 0.3871,
+      "step": 10004
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977433530609468e-05,
+      "loss": 0.469,
+      "step": 10005
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977422498466838e-05,
+      "loss": 0.4428,
+      "step": 10006
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977411463631256e-05,
+      "loss": 0.44,
+      "step": 10007
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977400426102722e-05,
+      "loss": 0.4844,
+      "step": 10008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977389385881244e-05,
+      "loss": 0.4207,
+      "step": 10009
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997737834296682e-05,
+      "loss": 0.4524,
+      "step": 10010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977367297359453e-05,
+      "loss": 0.4536,
+      "step": 10011
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997735624905915e-05,
+      "loss": 0.4907,
+      "step": 10012
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997734519806591e-05,
+      "loss": 0.3807,
+      "step": 10013
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997733414437974e-05,
+      "loss": 0.4747,
+      "step": 10014
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977323088000644e-05,
+      "loss": 0.484,
+      "step": 10015
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977312028928617e-05,
+      "loss": 0.5036,
+      "step": 10016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997730096716367e-05,
+      "loss": 0.4525,
+      "step": 10017
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.99772899027058e-05,
+      "loss": 0.4825,
+      "step": 10018
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977278835555013e-05,
+      "loss": 0.4892,
+      "step": 10019
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977267765711312e-05,
+      "loss": 0.4206,
+      "step": 10020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.99772566931747e-05,
+      "loss": 0.3757,
+      "step": 10021
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997724561794518e-05,
+      "loss": 0.4692,
+      "step": 10022
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977234540022758e-05,
+      "loss": 0.4793,
+      "step": 10023
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997722345940743e-05,
+      "loss": 0.4746,
+      "step": 10024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977212376099205e-05,
+      "loss": 0.4805,
+      "step": 10025
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997720129009808e-05,
+      "loss": 0.5115,
+      "step": 10026
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977190201404064e-05,
+      "loss": 0.4795,
+      "step": 10027
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977179110017157e-05,
+      "loss": 0.4896,
+      "step": 10028
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977168015937364e-05,
+      "loss": 0.5153,
+      "step": 10029
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977156919164686e-05,
+      "loss": 0.5154,
+      "step": 10030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977145819699128e-05,
+      "loss": 0.5242,
+      "step": 10031
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997713471754069e-05,
+      "loss": 0.5236,
+      "step": 10032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997712361268938e-05,
+      "loss": 0.4841,
+      "step": 10033
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977112505145193e-05,
+      "loss": 0.5165,
+      "step": 10034
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977101394908137e-05,
+      "loss": 0.4073,
+      "step": 10035
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997709028197822e-05,
+      "loss": 0.4099,
+      "step": 10036
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977079166355436e-05,
+      "loss": 0.5227,
+      "step": 10037
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997706804803979e-05,
+      "loss": 0.5459,
+      "step": 10038
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997705692703129e-05,
+      "loss": 0.482,
+      "step": 10039
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977045803329935e-05,
+      "loss": 0.466,
+      "step": 10040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977034676935726e-05,
+      "loss": 0.3862,
+      "step": 10041
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977023547848674e-05,
+      "loss": 0.5053,
+      "step": 10042
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977012416068772e-05,
+      "loss": 0.4959,
+      "step": 10043
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9977001281596027e-05,
+      "loss": 0.4393,
+      "step": 10044
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976990144430445e-05,
+      "loss": 0.4903,
+      "step": 10045
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976979004572028e-05,
+      "loss": 0.4588,
+      "step": 10046
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976967862020777e-05,
+      "loss": 0.5227,
+      "step": 10047
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976956716776693e-05,
+      "loss": 0.5207,
+      "step": 10048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976945568839783e-05,
+      "loss": 0.4424,
+      "step": 10049
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997693441821005e-05,
+      "loss": 0.5106,
+      "step": 10050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976923264887496e-05,
+      "loss": 0.4648,
+      "step": 10051
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976912108872124e-05,
+      "loss": 0.4283,
+      "step": 10052
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976900950163937e-05,
+      "loss": 0.4877,
+      "step": 10053
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976889788762934e-05,
+      "loss": 0.4522,
+      "step": 10054
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976878624669124e-05,
+      "loss": 0.5023,
+      "step": 10055
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976867457882506e-05,
+      "loss": 0.5032,
+      "step": 10056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976856288403092e-05,
+      "loss": 0.4912,
+      "step": 10057
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976845116230872e-05,
+      "loss": 0.5043,
+      "step": 10058
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976833941365853e-05,
+      "loss": 0.474,
+      "step": 10059
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976822763808045e-05,
+      "loss": 0.4224,
+      "step": 10060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976811583557442e-05,
+      "loss": 0.5305,
+      "step": 10061
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976800400614053e-05,
+      "loss": 0.4445,
+      "step": 10062
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976789214977878e-05,
+      "loss": 0.4269,
+      "step": 10063
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976778026648922e-05,
+      "loss": 0.4521,
+      "step": 10064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976766835627186e-05,
+      "loss": 0.4264,
+      "step": 10065
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976755641912676e-05,
+      "loss": 0.4252,
+      "step": 10066
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976744445505393e-05,
+      "loss": 0.4264,
+      "step": 10067
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976733246405335e-05,
+      "loss": 0.5267,
+      "step": 10068
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976722044612512e-05,
+      "loss": 0.5091,
+      "step": 10069
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997671084012693e-05,
+      "loss": 0.4442,
+      "step": 10070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997669963294858e-05,
+      "loss": 0.5455,
+      "step": 10071
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997668842307748e-05,
+      "loss": 0.4528,
+      "step": 10072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997667721051362e-05,
+      "loss": 0.5199,
+      "step": 10073
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976665995257007e-05,
+      "loss": 0.4633,
+      "step": 10074
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997665477730765e-05,
+      "loss": 0.5429,
+      "step": 10075
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976643556665542e-05,
+      "loss": 0.4173,
+      "step": 10076
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976632333330696e-05,
+      "loss": 0.4987,
+      "step": 10077
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976621107303108e-05,
+      "loss": 0.5055,
+      "step": 10078
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976609878582782e-05,
+      "loss": 0.4839,
+      "step": 10079
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976598647169725e-05,
+      "loss": 0.458,
+      "step": 10080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976587413063933e-05,
+      "loss": 0.5135,
+      "step": 10081
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976576176265417e-05,
+      "loss": 0.446,
+      "step": 10082
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997656493677418e-05,
+      "loss": 0.5473,
+      "step": 10083
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976553694590215e-05,
+      "loss": 0.4512,
+      "step": 10084
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976542449713535e-05,
+      "loss": 0.4989,
+      "step": 10085
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976531202144137e-05,
+      "loss": 0.4717,
+      "step": 10086
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997651995188203e-05,
+      "loss": 0.448,
+      "step": 10087
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976508698927213e-05,
+      "loss": 0.4163,
+      "step": 10088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976497443279686e-05,
+      "loss": 0.4256,
+      "step": 10089
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997648618493946e-05,
+      "loss": 0.4395,
+      "step": 10090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997647492390653e-05,
+      "loss": 0.4576,
+      "step": 10091
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976463660180908e-05,
+      "loss": 0.4997,
+      "step": 10092
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976452393762588e-05,
+      "loss": 0.45,
+      "step": 10093
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976441124651577e-05,
+      "loss": 0.3916,
+      "step": 10094
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976429852847876e-05,
+      "loss": 0.5038,
+      "step": 10095
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976418578351494e-05,
+      "loss": 0.4263,
+      "step": 10096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997640730116243e-05,
+      "loss": 0.6259,
+      "step": 10097
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976396021280683e-05,
+      "loss": 0.5474,
+      "step": 10098
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976384738706263e-05,
+      "loss": 0.5195,
+      "step": 10099
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997637345343917e-05,
+      "loss": 0.4716,
+      "step": 10100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976362165479406e-05,
+      "loss": 0.5198,
+      "step": 10101
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9976350874826975e-05,
+      "loss": 0.4816,
+      "step": 10102
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.997633958148188e-05,
+      "loss": 0.3973,
+      "step": 10103
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976328285444127e-05,
+      "loss": 0.4774,
+      "step": 10104
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976316986713716e-05,
+      "loss": 0.4938,
+      "step": 10105
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976305685290648e-05,
+      "loss": 0.443,
+      "step": 10106
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997629438117493e-05,
+      "loss": 0.5486,
+      "step": 10107
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976283074366566e-05,
+      "loss": 0.4746,
+      "step": 10108
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976271764865552e-05,
+      "loss": 0.4926,
+      "step": 10109
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976260452671898e-05,
+      "loss": 0.4719,
+      "step": 10110
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.99762491377856e-05,
+      "loss": 0.4517,
+      "step": 10111
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976237820206673e-05,
+      "loss": 0.5507,
+      "step": 10112
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997622649993511e-05,
+      "loss": 0.4333,
+      "step": 10113
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976215176970917e-05,
+      "loss": 0.509,
+      "step": 10114
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976203851314094e-05,
+      "loss": 0.4294,
+      "step": 10115
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997619252296465e-05,
+      "loss": 0.5201,
+      "step": 10116
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976181191922587e-05,
+      "loss": 0.4031,
+      "step": 10117
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976169858187902e-05,
+      "loss": 0.5411,
+      "step": 10118
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976158521760602e-05,
+      "loss": 0.4405,
+      "step": 10119
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976147182640692e-05,
+      "loss": 0.4811,
+      "step": 10120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976135840828173e-05,
+      "loss": 0.5207,
+      "step": 10121
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976124496323047e-05,
+      "loss": 0.4822,
+      "step": 10122
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997611314912532e-05,
+      "loss": 0.4838,
+      "step": 10123
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976101799234993e-05,
+      "loss": 0.5408,
+      "step": 10124
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997609044665207e-05,
+      "loss": 0.476,
+      "step": 10125
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976079091376555e-05,
+      "loss": 0.4578,
+      "step": 10126
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976067733408448e-05,
+      "loss": 0.5209,
+      "step": 10127
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997605637274775e-05,
+      "loss": 0.4998,
+      "step": 10128
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976045009394473e-05,
+      "loss": 0.4437,
+      "step": 10129
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976033643348615e-05,
+      "loss": 0.4755,
+      "step": 10130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9976022274610175e-05,
+      "loss": 0.4123,
+      "step": 10131
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997601090317916e-05,
+      "loss": 0.5251,
+      "step": 10132
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975999529055573e-05,
+      "loss": 0.5974,
+      "step": 10133
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975988152239424e-05,
+      "loss": 0.5054,
+      "step": 10134
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.99759767727307e-05,
+      "loss": 0.4339,
+      "step": 10135
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975965390529417e-05,
+      "loss": 0.4164,
+      "step": 10136
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975954005635575e-05,
+      "loss": 0.4558,
+      "step": 10137
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997594261804918e-05,
+      "loss": 0.403,
+      "step": 10138
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975931227770223e-05,
+      "loss": 0.5725,
+      "step": 10139
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975919834798723e-05,
+      "loss": 0.5266,
+      "step": 10140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975908439134672e-05,
+      "loss": 0.466,
+      "step": 10141
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975897040778075e-05,
+      "loss": 0.5408,
+      "step": 10142
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997588563972894e-05,
+      "loss": 0.5227,
+      "step": 10143
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975874235987267e-05,
+      "loss": 0.5134,
+      "step": 10144
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975862829553056e-05,
+      "loss": 0.5298,
+      "step": 10145
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975851420426317e-05,
+      "loss": 0.4673,
+      "step": 10146
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975840008607046e-05,
+      "loss": 0.5079,
+      "step": 10147
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997582859409525e-05,
+      "loss": 0.4185,
+      "step": 10148
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997581717689093e-05,
+      "loss": 0.4457,
+      "step": 10149
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975805756994093e-05,
+      "loss": 0.4903,
+      "step": 10150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975794334404738e-05,
+      "loss": 0.5115,
+      "step": 10151
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975782909122868e-05,
+      "loss": 0.5597,
+      "step": 10152
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997577148114849e-05,
+      "loss": 0.4437,
+      "step": 10153
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975760050481602e-05,
+      "loss": 0.5549,
+      "step": 10154
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997574861712221e-05,
+      "loss": 0.4235,
+      "step": 10155
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997573718107032e-05,
+      "loss": 0.4651,
+      "step": 10156
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997572574232593e-05,
+      "loss": 0.4716,
+      "step": 10157
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975714300889045e-05,
+      "loss": 0.4424,
+      "step": 10158
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975702856759665e-05,
+      "loss": 0.489,
+      "step": 10159
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.99756914099378e-05,
+      "loss": 0.6249,
+      "step": 10160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997567996042345e-05,
+      "loss": 0.5187,
+      "step": 10161
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975668508216616e-05,
+      "loss": 0.4262,
+      "step": 10162
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975657053317304e-05,
+      "loss": 0.4794,
+      "step": 10163
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997564559572551e-05,
+      "loss": 0.5107,
+      "step": 10164
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975634135441248e-05,
+      "loss": 0.4795,
+      "step": 10165
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975622672464518e-05,
+      "loss": 0.5243,
+      "step": 10166
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975611206795314e-05,
+      "loss": 0.4671,
+      "step": 10167
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997559973843365e-05,
+      "loss": 0.4933,
+      "step": 10168
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975588267379522e-05,
+      "loss": 0.4597,
+      "step": 10169
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975576793632938e-05,
+      "loss": 0.4484,
+      "step": 10170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.99755653171939e-05,
+      "loss": 0.4978,
+      "step": 10171
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975553838062408e-05,
+      "loss": 0.4012,
+      "step": 10172
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997554235623847e-05,
+      "loss": 0.5482,
+      "step": 10173
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975530871722085e-05,
+      "loss": 0.3855,
+      "step": 10174
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975519384513257e-05,
+      "loss": 0.4473,
+      "step": 10175
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975507894611992e-05,
+      "loss": 0.4674,
+      "step": 10176
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975496402018287e-05,
+      "loss": 0.457,
+      "step": 10177
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975484906732153e-05,
+      "loss": 0.454,
+      "step": 10178
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997547340875359e-05,
+      "loss": 0.434,
+      "step": 10179
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975461908082595e-05,
+      "loss": 0.4812,
+      "step": 10180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975450404719178e-05,
+      "loss": 0.4478,
+      "step": 10181
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997543889866334e-05,
+      "loss": 0.4723,
+      "step": 10182
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997542738991509e-05,
+      "loss": 0.4867,
+      "step": 10183
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975415878474417e-05,
+      "loss": 0.4839,
+      "step": 10184
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997540436434134e-05,
+      "loss": 0.4995,
+      "step": 10185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975392847515848e-05,
+      "loss": 0.4947,
+      "step": 10186
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975381327997954e-05,
+      "loss": 0.4458,
+      "step": 10187
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975369805787657e-05,
+      "loss": 0.5339,
+      "step": 10188
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975358280884965e-05,
+      "loss": 0.4304,
+      "step": 10189
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975346753289874e-05,
+      "loss": 0.4554,
+      "step": 10190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997533522300239e-05,
+      "loss": 0.483,
+      "step": 10191
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975323690022517e-05,
+      "loss": 0.4635,
+      "step": 10192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975312154350258e-05,
+      "loss": 0.4219,
+      "step": 10193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975300615985614e-05,
+      "loss": 0.4573,
+      "step": 10194
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997528907492859e-05,
+      "loss": 0.4271,
+      "step": 10195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997527753117919e-05,
+      "loss": 0.5087,
+      "step": 10196
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975265984737417e-05,
+      "loss": 0.5161,
+      "step": 10197
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997525443560327e-05,
+      "loss": 0.4968,
+      "step": 10198
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9975242883776755e-05,
+      "loss": 0.4707,
+      "step": 10199
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997523132925788e-05,
+      "loss": 0.5173,
+      "step": 10200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.997521977204664e-05,
+      "loss": 0.5139,
+      "step": 10201
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997520821214304e-05,
+      "loss": 0.4648,
+      "step": 10202
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975196649547087e-05,
+      "loss": 0.4006,
+      "step": 10203
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975185084258782e-05,
+      "loss": 0.4813,
+      "step": 10204
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997517351627813e-05,
+      "loss": 0.5157,
+      "step": 10205
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975161945605127e-05,
+      "loss": 0.5234,
+      "step": 10206
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975150372239787e-05,
+      "loss": 0.4121,
+      "step": 10207
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975138796182102e-05,
+      "loss": 0.4819,
+      "step": 10208
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975127217432082e-05,
+      "loss": 0.4771,
+      "step": 10209
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975115635989727e-05,
+      "loss": 0.3732,
+      "step": 10210
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975104051855044e-05,
+      "loss": 0.5121,
+      "step": 10211
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975092465028036e-05,
+      "loss": 0.4569,
+      "step": 10212
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.99750808755087e-05,
+      "loss": 0.4175,
+      "step": 10213
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975069283297043e-05,
+      "loss": 0.4565,
+      "step": 10214
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975057688393072e-05,
+      "loss": 0.4795,
+      "step": 10215
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975046090796782e-05,
+      "loss": 0.5195,
+      "step": 10216
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975034490508185e-05,
+      "loss": 0.4824,
+      "step": 10217
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975022887527276e-05,
+      "loss": 0.4693,
+      "step": 10218
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9975011281854063e-05,
+      "loss": 0.49,
+      "step": 10219
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974999673488546e-05,
+      "loss": 0.556,
+      "step": 10220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997498806243073e-05,
+      "loss": 0.4303,
+      "step": 10221
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974976448680622e-05,
+      "loss": 0.4721,
+      "step": 10222
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997496483223822e-05,
+      "loss": 0.5211,
+      "step": 10223
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974953213103524e-05,
+      "loss": 0.4361,
+      "step": 10224
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974941591276546e-05,
+      "loss": 0.4872,
+      "step": 10225
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997492996675728e-05,
+      "loss": 0.5316,
+      "step": 10226
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997491833954574e-05,
+      "loss": 0.5093,
+      "step": 10227
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997490670964192e-05,
+      "loss": 0.4553,
+      "step": 10228
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974895077045823e-05,
+      "loss": 0.4251,
+      "step": 10229
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997488344175746e-05,
+      "loss": 0.4732,
+      "step": 10230
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974871803776826e-05,
+      "loss": 0.5232,
+      "step": 10231
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997486016310393e-05,
+      "loss": 0.4716,
+      "step": 10232
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997484851973877e-05,
+      "loss": 0.5212,
+      "step": 10233
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974836873681355e-05,
+      "loss": 0.4145,
+      "step": 10234
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974825224931682e-05,
+      "loss": 0.4655,
+      "step": 10235
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997481357348976e-05,
+      "loss": 0.4623,
+      "step": 10236
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974801919355585e-05,
+      "loss": 0.4968,
+      "step": 10237
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974790262529168e-05,
+      "loss": 0.4766,
+      "step": 10238
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974778603010508e-05,
+      "loss": 0.4799,
+      "step": 10239
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974766940799607e-05,
+      "loss": 0.4738,
+      "step": 10240
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997475527589647e-05,
+      "loss": 0.4536,
+      "step": 10241
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.99747436083011e-05,
+      "loss": 0.4344,
+      "step": 10242
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.99747319380135e-05,
+      "loss": 0.5191,
+      "step": 10243
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974720265033677e-05,
+      "loss": 0.512,
+      "step": 10244
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974708589361627e-05,
+      "loss": 0.4922,
+      "step": 10245
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974696910997355e-05,
+      "loss": 0.4749,
+      "step": 10246
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997468522994087e-05,
+      "loss": 0.4792,
+      "step": 10247
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974673546192167e-05,
+      "loss": 0.4627,
+      "step": 10248
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974661859751256e-05,
+      "loss": 0.442,
+      "step": 10249
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997465017061813e-05,
+      "loss": 0.508,
+      "step": 10250
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974638478792808e-05,
+      "loss": 0.4845,
+      "step": 10251
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974626784275278e-05,
+      "loss": 0.4665,
+      "step": 10252
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974615087065556e-05,
+      "loss": 0.4496,
+      "step": 10253
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974603387163634e-05,
+      "loss": 0.6063,
+      "step": 10254
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974591684569523e-05,
+      "loss": 0.4342,
+      "step": 10255
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997457997928322e-05,
+      "loss": 0.5218,
+      "step": 10256
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974568271304734e-05,
+      "loss": 0.4345,
+      "step": 10257
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974556560634065e-05,
+      "loss": 0.4047,
+      "step": 10258
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974544847271215e-05,
+      "loss": 0.4155,
+      "step": 10259
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974533131216188e-05,
+      "loss": 0.5505,
+      "step": 10260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997452141246899e-05,
+      "loss": 0.5194,
+      "step": 10261
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997450969102962e-05,
+      "loss": 0.4693,
+      "step": 10262
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974497966898086e-05,
+      "loss": 0.3933,
+      "step": 10263
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974486240074387e-05,
+      "loss": 0.4687,
+      "step": 10264
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974474510558527e-05,
+      "loss": 0.4592,
+      "step": 10265
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974462778350508e-05,
+      "loss": 0.4116,
+      "step": 10266
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974451043450336e-05,
+      "loss": 0.4991,
+      "step": 10267
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974439305858017e-05,
+      "loss": 0.4713,
+      "step": 10268
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974427565573545e-05,
+      "loss": 0.444,
+      "step": 10269
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974415822596932e-05,
+      "loss": 0.4998,
+      "step": 10270
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974404076928177e-05,
+      "loss": 0.4756,
+      "step": 10271
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974392328567283e-05,
+      "loss": 0.4762,
+      "step": 10272
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997438057751425e-05,
+      "loss": 0.4795,
+      "step": 10273
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997436882376909e-05,
+      "loss": 0.5084,
+      "step": 10274
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.99743570673318e-05,
+      "loss": 0.5352,
+      "step": 10275
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974345308202387e-05,
+      "loss": 0.479,
+      "step": 10276
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997433354638085e-05,
+      "loss": 0.4173,
+      "step": 10277
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974321781867193e-05,
+      "loss": 0.4747,
+      "step": 10278
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997431001466142e-05,
+      "loss": 0.5325,
+      "step": 10279
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997429824476353e-05,
+      "loss": 0.4586,
+      "step": 10280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974286472173534e-05,
+      "loss": 0.488,
+      "step": 10281
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974274696891433e-05,
+      "loss": 0.4508,
+      "step": 10282
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974262918917227e-05,
+      "loss": 0.5168,
+      "step": 10283
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997425113825092e-05,
+      "loss": 0.4881,
+      "step": 10284
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974239354892516e-05,
+      "loss": 0.4984,
+      "step": 10285
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974227568842018e-05,
+      "loss": 0.4614,
+      "step": 10286
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997421578009943e-05,
+      "loss": 0.4643,
+      "step": 10287
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974203988664758e-05,
+      "loss": 0.4775,
+      "step": 10288
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974192194538e-05,
+      "loss": 0.4615,
+      "step": 10289
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974180397719158e-05,
+      "loss": 0.4885,
+      "step": 10290
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997416859820824e-05,
+      "loss": 0.4371,
+      "step": 10291
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974156796005245e-05,
+      "loss": 0.4368,
+      "step": 10292
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997414499111018e-05,
+      "loss": 0.5157,
+      "step": 10293
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997413318352305e-05,
+      "loss": 0.5103,
+      "step": 10294
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974121373243852e-05,
+      "loss": 0.4044,
+      "step": 10295
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974109560272593e-05,
+      "loss": 0.5023,
+      "step": 10296
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974097744609274e-05,
+      "loss": 0.468,
+      "step": 10297
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.99740859262539e-05,
+      "loss": 0.5454,
+      "step": 10298
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.997407410520647e-05,
+      "loss": 0.4506,
+      "step": 10299
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9974062281466994e-05,
+      "loss": 0.4737,
+      "step": 10300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9974050455035473e-05,
+      "loss": 0.4286,
+      "step": 10301
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997403862591191e-05,
+      "loss": 0.4734,
+      "step": 10302
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9974026794096302e-05,
+      "loss": 0.4041,
+      "step": 10303
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997401495958866e-05,
+      "loss": 0.4758,
+      "step": 10304
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9974003122388987e-05,
+      "loss": 0.5594,
+      "step": 10305
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973991282497286e-05,
+      "loss": 0.5095,
+      "step": 10306
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973979439913552e-05,
+      "loss": 0.4414,
+      "step": 10307
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.99739675946378e-05,
+      "loss": 0.5665,
+      "step": 10308
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997395574667002e-05,
+      "loss": 0.4449,
+      "step": 10309
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997394389601023e-05,
+      "loss": 0.4865,
+      "step": 10310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973932042658425e-05,
+      "loss": 0.4089,
+      "step": 10311
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973920186614606e-05,
+      "loss": 0.4637,
+      "step": 10312
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997390832787878e-05,
+      "loss": 0.4475,
+      "step": 10313
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973896466450952e-05,
+      "loss": 0.4722,
+      "step": 10314
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973884602331117e-05,
+      "loss": 0.4172,
+      "step": 10315
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997387273551929e-05,
+      "loss": 0.4306,
+      "step": 10316
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973860866015466e-05,
+      "loss": 0.5943,
+      "step": 10317
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997384899381965e-05,
+      "loss": 0.4556,
+      "step": 10318
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973837118931845e-05,
+      "loss": 0.4302,
+      "step": 10319
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973825241352057e-05,
+      "loss": 0.4,
+      "step": 10320
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973813361080287e-05,
+      "loss": 0.4238,
+      "step": 10321
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973801478116534e-05,
+      "loss": 0.4415,
+      "step": 10322
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997378959246081e-05,
+      "loss": 0.5291,
+      "step": 10323
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997377770411311e-05,
+      "loss": 0.4186,
+      "step": 10324
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997376581307344e-05,
+      "loss": 0.412,
+      "step": 10325
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997375391934181e-05,
+      "loss": 0.4426,
+      "step": 10326
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997374202291821e-05,
+      "loss": 0.4861,
+      "step": 10327
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973730123802654e-05,
+      "loss": 0.5044,
+      "step": 10328
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973718221995142e-05,
+      "loss": 0.4714,
+      "step": 10329
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997370631749568e-05,
+      "loss": 0.3904,
+      "step": 10330
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997369441030426e-05,
+      "loss": 0.4927,
+      "step": 10331
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.99736825004209e-05,
+      "loss": 0.4908,
+      "step": 10332
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997367058784559e-05,
+      "loss": 0.443,
+      "step": 10333
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997365867257835e-05,
+      "loss": 0.4601,
+      "step": 10334
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973646754619166e-05,
+      "loss": 0.5509,
+      "step": 10335
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973634833968048e-05,
+      "loss": 0.4723,
+      "step": 10336
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973622910624998e-05,
+      "loss": 0.4827,
+      "step": 10337
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973610984590023e-05,
+      "loss": 0.5054,
+      "step": 10338
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973599055863126e-05,
+      "loss": 0.5242,
+      "step": 10339
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973587124444304e-05,
+      "loss": 0.4763,
+      "step": 10340
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973575190333565e-05,
+      "loss": 0.4756,
+      "step": 10341
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973563253530914e-05,
+      "loss": 0.4932,
+      "step": 10342
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997355131403635e-05,
+      "loss": 0.6405,
+      "step": 10343
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973539371849874e-05,
+      "loss": 0.5024,
+      "step": 10344
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.99735274269715e-05,
+      "loss": 0.4669,
+      "step": 10345
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997351547940122e-05,
+      "loss": 0.4884,
+      "step": 10346
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973503529139045e-05,
+      "loss": 0.4107,
+      "step": 10347
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997349157618497e-05,
+      "loss": 0.405,
+      "step": 10348
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973479620539007e-05,
+      "loss": 0.4431,
+      "step": 10349
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973467662201153e-05,
+      "loss": 0.5388,
+      "step": 10350
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973455701171415e-05,
+      "loss": 0.4796,
+      "step": 10351
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973443737449793e-05,
+      "loss": 0.4858,
+      "step": 10352
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997343177103629e-05,
+      "loss": 0.408,
+      "step": 10353
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997341980193092e-05,
+      "loss": 0.5319,
+      "step": 10354
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973407830133668e-05,
+      "loss": 0.4523,
+      "step": 10355
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973395855644552e-05,
+      "loss": 0.4242,
+      "step": 10356
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973383878463566e-05,
+      "loss": 0.4864,
+      "step": 10357
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973371898590716e-05,
+      "loss": 0.4145,
+      "step": 10358
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973359916026012e-05,
+      "loss": 0.4678,
+      "step": 10359
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973347930769448e-05,
+      "loss": 0.4731,
+      "step": 10360
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997333594282103e-05,
+      "loss": 0.4684,
+      "step": 10361
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973323952180765e-05,
+      "loss": 0.4867,
+      "step": 10362
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973311958848653e-05,
+      "loss": 0.459,
+      "step": 10363
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973299962824694e-05,
+      "loss": 0.5326,
+      "step": 10364
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973287964108898e-05,
+      "loss": 0.4712,
+      "step": 10365
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997327596270126e-05,
+      "loss": 0.4374,
+      "step": 10366
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973263958601795e-05,
+      "loss": 0.4528,
+      "step": 10367
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973251951810496e-05,
+      "loss": 0.4186,
+      "step": 10368
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997323994232737e-05,
+      "loss": 0.4772,
+      "step": 10369
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997322793015242e-05,
+      "loss": 0.4992,
+      "step": 10370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973215915285647e-05,
+      "loss": 0.4993,
+      "step": 10371
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997320389772706e-05,
+      "loss": 0.4585,
+      "step": 10372
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997319187747666e-05,
+      "loss": 0.4392,
+      "step": 10373
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973179854534447e-05,
+      "loss": 0.4448,
+      "step": 10374
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973167828900425e-05,
+      "loss": 0.4604,
+      "step": 10375
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973155800574593e-05,
+      "loss": 0.4696,
+      "step": 10376
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997314376955697e-05,
+      "loss": 0.4374,
+      "step": 10377
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973131735847545e-05,
+      "loss": 0.5064,
+      "step": 10378
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973119699446322e-05,
+      "loss": 0.5696,
+      "step": 10379
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997310766035331e-05,
+      "loss": 0.4694,
+      "step": 10380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997309561856851e-05,
+      "loss": 0.4334,
+      "step": 10381
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973083574091922e-05,
+      "loss": 0.5009,
+      "step": 10382
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973071526923554e-05,
+      "loss": 0.4417,
+      "step": 10383
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997305947706341e-05,
+      "loss": 0.4299,
+      "step": 10384
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997304742451149e-05,
+      "loss": 0.4558,
+      "step": 10385
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973035369267794e-05,
+      "loss": 0.4664,
+      "step": 10386
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973023311332333e-05,
+      "loss": 0.4006,
+      "step": 10387
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9973011250705103e-05,
+      "loss": 0.4418,
+      "step": 10388
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997299918738611e-05,
+      "loss": 0.4227,
+      "step": 10389
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9972987121375364e-05,
+      "loss": 0.5308,
+      "step": 10390
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9972975052672857e-05,
+      "loss": 0.4496,
+      "step": 10391
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.99729629812786e-05,
+      "loss": 0.4891,
+      "step": 10392
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997295090719259e-05,
+      "loss": 0.5032,
+      "step": 10393
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9972938830414836e-05,
+      "loss": 0.4964,
+      "step": 10394
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997292675094534e-05,
+      "loss": 0.5371,
+      "step": 10395
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9972914668784104e-05,
+      "loss": 0.3996,
+      "step": 10396
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9972902583931133e-05,
+      "loss": 0.5816,
+      "step": 10397
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997289049638643e-05,
+      "loss": 0.5278,
+      "step": 10398
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.997287840614999e-05,
+      "loss": 0.5014,
+      "step": 10399
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997286631322183e-05,
+      "loss": 0.5303,
+      "step": 10400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972854217601948e-05,
+      "loss": 0.5306,
+      "step": 10401
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997284211929034e-05,
+      "loss": 0.4701,
+      "step": 10402
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972830018287023e-05,
+      "loss": 0.4746,
+      "step": 10403
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972817914591983e-05,
+      "loss": 0.4385,
+      "step": 10404
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997280580820524e-05,
+      "loss": 0.4835,
+      "step": 10405
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972793699126785e-05,
+      "loss": 0.4667,
+      "step": 10406
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997278158735663e-05,
+      "loss": 0.4818,
+      "step": 10407
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972769472894776e-05,
+      "loss": 0.3994,
+      "step": 10408
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972757355741222e-05,
+      "loss": 0.5118,
+      "step": 10409
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972745235895975e-05,
+      "loss": 0.4574,
+      "step": 10410
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997273311335904e-05,
+      "loss": 0.5288,
+      "step": 10411
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972720988130416e-05,
+      "loss": 0.4654,
+      "step": 10412
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972708860210105e-05,
+      "loss": 0.4398,
+      "step": 10413
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997269672959812e-05,
+      "loss": 0.4759,
+      "step": 10414
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997268459629445e-05,
+      "loss": 0.4774,
+      "step": 10415
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972672460299106e-05,
+      "loss": 0.5741,
+      "step": 10416
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972660321612095e-05,
+      "loss": 0.4367,
+      "step": 10417
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972648180233416e-05,
+      "loss": 0.464,
+      "step": 10418
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997263603616307e-05,
+      "loss": 0.5791,
+      "step": 10419
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972623889401066e-05,
+      "loss": 0.5116,
+      "step": 10420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972611739947406e-05,
+      "loss": 0.5292,
+      "step": 10421
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972599587802088e-05,
+      "loss": 0.4785,
+      "step": 10422
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997258743296512e-05,
+      "loss": 0.4248,
+      "step": 10423
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.99725752754365e-05,
+      "loss": 0.4651,
+      "step": 10424
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997256311521624e-05,
+      "loss": 0.5077,
+      "step": 10425
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997255095230434e-05,
+      "loss": 0.4276,
+      "step": 10426
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972538786700796e-05,
+      "loss": 0.5963,
+      "step": 10427
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972526618405622e-05,
+      "loss": 0.4949,
+      "step": 10428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972514447418812e-05,
+      "loss": 0.405,
+      "step": 10429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997250227374038e-05,
+      "loss": 0.5008,
+      "step": 10430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997249009737032e-05,
+      "loss": 0.5122,
+      "step": 10431
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972477918308635e-05,
+      "loss": 0.4857,
+      "step": 10432
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972465736555334e-05,
+      "loss": 0.4341,
+      "step": 10433
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997245355211042e-05,
+      "loss": 0.5109,
+      "step": 10434
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997244136497389e-05,
+      "loss": 0.4666,
+      "step": 10435
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972429175145754e-05,
+      "loss": 0.4337,
+      "step": 10436
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972416982626014e-05,
+      "loss": 0.4668,
+      "step": 10437
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972404787414672e-05,
+      "loss": 0.5122,
+      "step": 10438
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997239258951173e-05,
+      "loss": 0.4796,
+      "step": 10439
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997238038891719e-05,
+      "loss": 0.52,
+      "step": 10440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972368185631063e-05,
+      "loss": 0.5253,
+      "step": 10441
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972355979653345e-05,
+      "loss": 0.4833,
+      "step": 10442
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997234377098404e-05,
+      "loss": 0.4429,
+      "step": 10443
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972331559623155e-05,
+      "loss": 0.4429,
+      "step": 10444
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997231934557069e-05,
+      "loss": 0.4556,
+      "step": 10445
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972307128826648e-05,
+      "loss": 0.5104,
+      "step": 10446
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972294909391037e-05,
+      "loss": 0.4489,
+      "step": 10447
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972282687263854e-05,
+      "loss": 0.51,
+      "step": 10448
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972270462445108e-05,
+      "loss": 0.499,
+      "step": 10449
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972258234934797e-05,
+      "loss": 0.4856,
+      "step": 10450
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972246004732926e-05,
+      "loss": 0.5619,
+      "step": 10451
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972233771839503e-05,
+      "loss": 0.4868,
+      "step": 10452
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997222153625453e-05,
+      "loss": 0.4913,
+      "step": 10453
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972209297978e-05,
+      "loss": 0.4871,
+      "step": 10454
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997219705700993e-05,
+      "loss": 0.5274,
+      "step": 10455
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972184813350314e-05,
+      "loss": 0.4727,
+      "step": 10456
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972172566999157e-05,
+      "loss": 0.5432,
+      "step": 10457
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997216031795647e-05,
+      "loss": 0.4957,
+      "step": 10458
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972148066222245e-05,
+      "loss": 0.4919,
+      "step": 10459
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972135811796492e-05,
+      "loss": 0.48,
+      "step": 10460
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972123554679215e-05,
+      "loss": 0.4461,
+      "step": 10461
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972111294870412e-05,
+      "loss": 0.4461,
+      "step": 10462
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997209903237009e-05,
+      "loss": 0.4998,
+      "step": 10463
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972086767178255e-05,
+      "loss": 0.5067,
+      "step": 10464
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972074499294908e-05,
+      "loss": 0.4801,
+      "step": 10465
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997206222872005e-05,
+      "loss": 0.5182,
+      "step": 10466
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997204995545368e-05,
+      "loss": 0.4534,
+      "step": 10467
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972037679495813e-05,
+      "loss": 0.468,
+      "step": 10468
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972025400846443e-05,
+      "loss": 0.4971,
+      "step": 10469
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997201311950558e-05,
+      "loss": 0.4746,
+      "step": 10470
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9972000835473226e-05,
+      "loss": 0.5063,
+      "step": 10471
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997198854874938e-05,
+      "loss": 0.4411,
+      "step": 10472
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971976259334043e-05,
+      "loss": 0.5057,
+      "step": 10473
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971963967227227e-05,
+      "loss": 0.4223,
+      "step": 10474
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971951672428934e-05,
+      "loss": 0.4617,
+      "step": 10475
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971939374939162e-05,
+      "loss": 0.4357,
+      "step": 10476
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971927074757917e-05,
+      "loss": 0.5084,
+      "step": 10477
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.99719147718852e-05,
+      "loss": 0.554,
+      "step": 10478
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997190246632102e-05,
+      "loss": 0.5283,
+      "step": 10479
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971890158065374e-05,
+      "loss": 0.4634,
+      "step": 10480
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971877847118272e-05,
+      "loss": 0.4647,
+      "step": 10481
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997186553347971e-05,
+      "loss": 0.5086,
+      "step": 10482
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971853217149695e-05,
+      "loss": 0.5053,
+      "step": 10483
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971840898128235e-05,
+      "loss": 0.4893,
+      "step": 10484
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971828576415323e-05,
+      "loss": 0.592,
+      "step": 10485
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997181625201097e-05,
+      "loss": 0.4341,
+      "step": 10486
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997180392491518e-05,
+      "loss": 0.3882,
+      "step": 10487
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971791595127947e-05,
+      "loss": 0.4482,
+      "step": 10488
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971779262649284e-05,
+      "loss": 0.4733,
+      "step": 10489
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997176692747919e-05,
+      "loss": 0.4147,
+      "step": 10490
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971754589617674e-05,
+      "loss": 0.5259,
+      "step": 10491
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997174224906473e-05,
+      "loss": 0.4952,
+      "step": 10492
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997172990582037e-05,
+      "loss": 0.4707,
+      "step": 10493
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997171755988459e-05,
+      "loss": 0.5731,
+      "step": 10494
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.99717052112574e-05,
+      "loss": 0.4355,
+      "step": 10495
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9971692859938797e-05,
+      "loss": 0.4756,
+      "step": 10496
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.997168050592879e-05,
+      "loss": 0.5349,
+      "step": 10497
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971668149227378e-05,
+      "loss": 0.5049,
+      "step": 10498
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971655789834568e-05,
+      "loss": 0.5121,
+      "step": 10499
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997164342775036e-05,
+      "loss": 0.4618,
+      "step": 10500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997163106297476e-05,
+      "loss": 0.4329,
+      "step": 10501
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997161869550777e-05,
+      "loss": 0.4078,
+      "step": 10502
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971606325349394e-05,
+      "loss": 0.387,
+      "step": 10503
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997159395249963e-05,
+      "loss": 0.5087,
+      "step": 10504
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971581576958493e-05,
+      "loss": 0.5592,
+      "step": 10505
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971569198725976e-05,
+      "loss": 0.4891,
+      "step": 10506
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971556817802088e-05,
+      "loss": 0.4619,
+      "step": 10507
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971544434186826e-05,
+      "loss": 0.4321,
+      "step": 10508
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.99715320478802e-05,
+      "loss": 0.4223,
+      "step": 10509
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971519658882212e-05,
+      "loss": 0.4801,
+      "step": 10510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971507267192863e-05,
+      "loss": 0.4889,
+      "step": 10511
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971494872812158e-05,
+      "loss": 0.4205,
+      "step": 10512
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971482475740098e-05,
+      "loss": 0.4123,
+      "step": 10513
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971470075976692e-05,
+      "loss": 0.5515,
+      "step": 10514
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971457673521937e-05,
+      "loss": 0.4825,
+      "step": 10515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997144526837584e-05,
+      "loss": 0.494,
+      "step": 10516
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971432860538404e-05,
+      "loss": 0.4651,
+      "step": 10517
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997142045000963e-05,
+      "loss": 0.5586,
+      "step": 10518
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971408036789523e-05,
+      "loss": 0.416,
+      "step": 10519
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971395620878087e-05,
+      "loss": 0.4784,
+      "step": 10520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971383202275324e-05,
+      "loss": 0.5903,
+      "step": 10521
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971370780981238e-05,
+      "loss": 0.4726,
+      "step": 10522
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997135835699583e-05,
+      "loss": 0.4193,
+      "step": 10523
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997134593031911e-05,
+      "loss": 0.3935,
+      "step": 10524
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971333500951077e-05,
+      "loss": 0.4929,
+      "step": 10525
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971321068891735e-05,
+      "loss": 0.4885,
+      "step": 10526
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971308634141084e-05,
+      "loss": 0.5656,
+      "step": 10527
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997129619669913e-05,
+      "loss": 0.5077,
+      "step": 10528
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997128375656588e-05,
+      "loss": 0.511,
+      "step": 10529
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971271313741333e-05,
+      "loss": 0.46,
+      "step": 10530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971258868225493e-05,
+      "loss": 0.4872,
+      "step": 10531
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971246420018362e-05,
+      "loss": 0.5077,
+      "step": 10532
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971233969119943e-05,
+      "loss": 0.4986,
+      "step": 10533
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971221515530246e-05,
+      "loss": 0.5616,
+      "step": 10534
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971209059249268e-05,
+      "loss": 0.5258,
+      "step": 10535
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971196600277012e-05,
+      "loss": 0.4346,
+      "step": 10536
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971184138613485e-05,
+      "loss": 0.444,
+      "step": 10537
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997117167425869e-05,
+      "loss": 0.4797,
+      "step": 10538
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997115920721263e-05,
+      "loss": 0.4734,
+      "step": 10539
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971146737475305e-05,
+      "loss": 0.553,
+      "step": 10540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997113426504672e-05,
+      "loss": 0.4495,
+      "step": 10541
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971121789926884e-05,
+      "loss": 0.423,
+      "step": 10542
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971109312115792e-05,
+      "loss": 0.4524,
+      "step": 10543
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997109683161345e-05,
+      "loss": 0.4665,
+      "step": 10544
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971084348419866e-05,
+      "loss": 0.513,
+      "step": 10545
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971071862535036e-05,
+      "loss": 0.4854,
+      "step": 10546
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997105937395897e-05,
+      "loss": 0.551,
+      "step": 10547
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971046882691664e-05,
+      "loss": 0.4097,
+      "step": 10548
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997103438873313e-05,
+      "loss": 0.5175,
+      "step": 10549
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971021892083367e-05,
+      "loss": 0.4764,
+      "step": 10550
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9971009392742378e-05,
+      "loss": 0.4263,
+      "step": 10551
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997099689071017e-05,
+      "loss": 0.5051,
+      "step": 10552
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997098438598674e-05,
+      "loss": 0.4345,
+      "step": 10553
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970971878572093e-05,
+      "loss": 0.4711,
+      "step": 10554
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970959368466234e-05,
+      "loss": 0.5242,
+      "step": 10555
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970946855669168e-05,
+      "loss": 0.4473,
+      "step": 10556
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970934340180895e-05,
+      "loss": 0.4214,
+      "step": 10557
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970921822001426e-05,
+      "loss": 0.4996,
+      "step": 10558
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970909301130753e-05,
+      "loss": 0.4898,
+      "step": 10559
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970896777568888e-05,
+      "loss": 0.4256,
+      "step": 10560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970884251315833e-05,
+      "loss": 0.4046,
+      "step": 10561
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970871722371584e-05,
+      "loss": 0.5079,
+      "step": 10562
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970859190736153e-05,
+      "loss": 0.4852,
+      "step": 10563
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970846656409542e-05,
+      "loss": 0.4426,
+      "step": 10564
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970834119391752e-05,
+      "loss": 0.4986,
+      "step": 10565
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970821579682786e-05,
+      "loss": 0.4702,
+      "step": 10566
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970809037282646e-05,
+      "loss": 0.4821,
+      "step": 10567
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970796492191345e-05,
+      "loss": 0.5192,
+      "step": 10568
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970783944408873e-05,
+      "loss": 0.5429,
+      "step": 10569
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970771393935243e-05,
+      "loss": 0.4328,
+      "step": 10570
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970758840770457e-05,
+      "loss": 0.4985,
+      "step": 10571
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970746284914515e-05,
+      "loss": 0.4815,
+      "step": 10572
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997073372636742e-05,
+      "loss": 0.4949,
+      "step": 10573
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970721165129177e-05,
+      "loss": 0.425,
+      "step": 10574
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997070860119979e-05,
+      "loss": 0.4549,
+      "step": 10575
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970696034579267e-05,
+      "loss": 0.4232,
+      "step": 10576
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970683465267604e-05,
+      "loss": 0.4867,
+      "step": 10577
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970670893264805e-05,
+      "loss": 0.4591,
+      "step": 10578
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970658318570877e-05,
+      "loss": 0.4709,
+      "step": 10579
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997064574118582e-05,
+      "loss": 0.4108,
+      "step": 10580
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997063316110964e-05,
+      "loss": 0.4563,
+      "step": 10581
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970620578342343e-05,
+      "loss": 0.4932,
+      "step": 10582
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970607992883924e-05,
+      "loss": 0.4397,
+      "step": 10583
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970595404734393e-05,
+      "loss": 0.3881,
+      "step": 10584
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997058281389375e-05,
+      "loss": 0.4533,
+      "step": 10585
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970570220362004e-05,
+      "loss": 0.3957,
+      "step": 10586
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970557624139152e-05,
+      "loss": 0.4737,
+      "step": 10587
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.99705450252252e-05,
+      "loss": 0.4574,
+      "step": 10588
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970532423620153e-05,
+      "loss": 0.4857,
+      "step": 10589
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997051981932401e-05,
+      "loss": 0.5094,
+      "step": 10590
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970507212336778e-05,
+      "loss": 0.4863,
+      "step": 10591
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970494602658458e-05,
+      "loss": 0.5068,
+      "step": 10592
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970481990289057e-05,
+      "loss": 0.494,
+      "step": 10593
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970469375228575e-05,
+      "loss": 0.5947,
+      "step": 10594
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997045675747702e-05,
+      "loss": 0.5668,
+      "step": 10595
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997044413703439e-05,
+      "loss": 0.422,
+      "step": 10596
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.997043151390069e-05,
+      "loss": 0.3912,
+      "step": 10597
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970418888075926e-05,
+      "loss": 0.4861,
+      "step": 10598
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970406259560095e-05,
+      "loss": 0.5189,
+      "step": 10599
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970393628353206e-05,
+      "loss": 0.4903,
+      "step": 10600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970380994455264e-05,
+      "loss": 0.4372,
+      "step": 10601
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970368357866267e-05,
+      "loss": 0.4202,
+      "step": 10602
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970355718586223e-05,
+      "loss": 0.4148,
+      "step": 10603
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.997034307661513e-05,
+      "loss": 0.3993,
+      "step": 10604
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970330431953e-05,
+      "loss": 0.4307,
+      "step": 10605
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970317784599827e-05,
+      "loss": 0.461,
+      "step": 10606
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.997030513455562e-05,
+      "loss": 0.4039,
+      "step": 10607
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970292481820383e-05,
+      "loss": 0.5114,
+      "step": 10608
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970279826394115e-05,
+      "loss": 0.5515,
+      "step": 10609
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.997026716827682e-05,
+      "loss": 0.4773,
+      "step": 10610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970254507468506e-05,
+      "loss": 0.3859,
+      "step": 10611
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970241843969174e-05,
+      "loss": 0.4139,
+      "step": 10612
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970229177778826e-05,
+      "loss": 0.53,
+      "step": 10613
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970216508897467e-05,
+      "loss": 0.4239,
+      "step": 10614
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970203837325098e-05,
+      "loss": 0.4614,
+      "step": 10615
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970191163061726e-05,
+      "loss": 0.4632,
+      "step": 10616
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970178486107357e-05,
+      "loss": 0.5305,
+      "step": 10617
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.997016580646198e-05,
+      "loss": 0.5024,
+      "step": 10618
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.997015312412562e-05,
+      "loss": 0.4474,
+      "step": 10619
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.997014043909826e-05,
+      "loss": 0.536,
+      "step": 10620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.997012775137992e-05,
+      "loss": 0.542,
+      "step": 10621
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970115060970592e-05,
+      "loss": 0.5768,
+      "step": 10622
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970102367870284e-05,
+      "loss": 0.4648,
+      "step": 10623
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970089672078997e-05,
+      "loss": 0.6049,
+      "step": 10624
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.997007697359674e-05,
+      "loss": 0.4445,
+      "step": 10625
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970064272423508e-05,
+      "loss": 0.395,
+      "step": 10626
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970051568559312e-05,
+      "loss": 0.5281,
+      "step": 10627
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970038862004154e-05,
+      "loss": 0.4022,
+      "step": 10628
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970026152758033e-05,
+      "loss": 0.4378,
+      "step": 10629
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970013440820956e-05,
+      "loss": 0.4291,
+      "step": 10630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9970000726192927e-05,
+      "loss": 0.532,
+      "step": 10631
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969988008873945e-05,
+      "loss": 0.3778,
+      "step": 10632
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969975288864017e-05,
+      "loss": 0.5604,
+      "step": 10633
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996996256616315e-05,
+      "loss": 0.4994,
+      "step": 10634
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996994984077134e-05,
+      "loss": 0.4269,
+      "step": 10635
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969937112688597e-05,
+      "loss": 0.4389,
+      "step": 10636
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969924381914917e-05,
+      "loss": 0.4932,
+      "step": 10637
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996991164845031e-05,
+      "loss": 0.5415,
+      "step": 10638
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996989891229478e-05,
+      "loss": 0.4893,
+      "step": 10639
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969886173448327e-05,
+      "loss": 0.4675,
+      "step": 10640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969873431910952e-05,
+      "loss": 0.4164,
+      "step": 10641
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969860687682662e-05,
+      "loss": 0.4918,
+      "step": 10642
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969847940763463e-05,
+      "loss": 0.4151,
+      "step": 10643
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969835191153353e-05,
+      "loss": 0.4538,
+      "step": 10644
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969822438852337e-05,
+      "loss": 0.4048,
+      "step": 10645
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969809683860423e-05,
+      "loss": 0.4606,
+      "step": 10646
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969796926177607e-05,
+      "loss": 0.5043,
+      "step": 10647
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969784165803897e-05,
+      "loss": 0.5833,
+      "step": 10648
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969771402739298e-05,
+      "loss": 0.4437,
+      "step": 10649
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969758636983808e-05,
+      "loss": 0.4497,
+      "step": 10650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969745868537433e-05,
+      "loss": 0.4581,
+      "step": 10651
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996973309740018e-05,
+      "loss": 0.5247,
+      "step": 10652
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969720323572052e-05,
+      "loss": 0.568,
+      "step": 10653
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969707547053043e-05,
+      "loss": 0.5118,
+      "step": 10654
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969694767843166e-05,
+      "loss": 0.6195,
+      "step": 10655
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969681985942425e-05,
+      "loss": 0.4139,
+      "step": 10656
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969669201350816e-05,
+      "loss": 0.4523,
+      "step": 10657
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996965641406835e-05,
+      "loss": 0.4563,
+      "step": 10658
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969643624095024e-05,
+      "loss": 0.4876,
+      "step": 10659
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969630831430846e-05,
+      "loss": 0.4824,
+      "step": 10660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969618036075816e-05,
+      "loss": 0.5682,
+      "step": 10661
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969605238029943e-05,
+      "loss": 0.4976,
+      "step": 10662
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969592437293225e-05,
+      "loss": 0.576,
+      "step": 10663
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969579633865666e-05,
+      "loss": 0.5196,
+      "step": 10664
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969566827747274e-05,
+      "loss": 0.4915,
+      "step": 10665
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969554018938048e-05,
+      "loss": 0.4208,
+      "step": 10666
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969541207437994e-05,
+      "loss": 0.4523,
+      "step": 10667
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996952839324711e-05,
+      "loss": 0.4915,
+      "step": 10668
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996951557636541e-05,
+      "loss": 0.4686,
+      "step": 10669
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969502756792884e-05,
+      "loss": 0.4959,
+      "step": 10670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996948993452955e-05,
+      "loss": 0.4164,
+      "step": 10671
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969477109575397e-05,
+      "loss": 0.4273,
+      "step": 10672
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996946428193044e-05,
+      "loss": 0.4775,
+      "step": 10673
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969451451594676e-05,
+      "loss": 0.4798,
+      "step": 10674
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969438618568113e-05,
+      "loss": 0.4847,
+      "step": 10675
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969425782850748e-05,
+      "loss": 0.5319,
+      "step": 10676
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969412944442592e-05,
+      "loss": 0.5374,
+      "step": 10677
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969400103343646e-05,
+      "loss": 0.4142,
+      "step": 10678
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969387259553907e-05,
+      "loss": 0.4602,
+      "step": 10679
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996937441307339e-05,
+      "loss": 0.4594,
+      "step": 10680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969361563902087e-05,
+      "loss": 0.4972,
+      "step": 10681
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969348712040005e-05,
+      "loss": 0.4504,
+      "step": 10682
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969335857487154e-05,
+      "loss": 0.4352,
+      "step": 10683
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969323000243534e-05,
+      "loss": 0.3985,
+      "step": 10684
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969310140309144e-05,
+      "loss": 0.5129,
+      "step": 10685
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996929727768399e-05,
+      "loss": 0.4378,
+      "step": 10686
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.996928441236808e-05,
+      "loss": 0.4808,
+      "step": 10687
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969271544361407e-05,
+      "loss": 0.5294,
+      "step": 10688
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969258673663986e-05,
+      "loss": 0.4172,
+      "step": 10689
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969245800275815e-05,
+      "loss": 0.4753,
+      "step": 10690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969232924196896e-05,
+      "loss": 0.4365,
+      "step": 10691
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969220045427237e-05,
+      "loss": 0.5188,
+      "step": 10692
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9969207163966836e-05,
+      "loss": 0.4119,
+      "step": 10693
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.99691942798157e-05,
+      "loss": 0.4825,
+      "step": 10694
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9969181392973834e-05,
+      "loss": 0.4303,
+      "step": 10695
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996916850344124e-05,
+      "loss": 0.5028,
+      "step": 10696
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996915561121792e-05,
+      "loss": 0.5006,
+      "step": 10697
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9969142716303874e-05,
+      "loss": 0.4658,
+      "step": 10698
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9969129818699113e-05,
+      "loss": 0.4797,
+      "step": 10699
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9969116918403637e-05,
+      "loss": 0.423,
+      "step": 10700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9969104015417452e-05,
+      "loss": 0.4779,
+      "step": 10701
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9969091109740556e-05,
+      "loss": 0.4242,
+      "step": 10702
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9969078201372957e-05,
+      "loss": 0.4142,
+      "step": 10703
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9969065290314657e-05,
+      "loss": 0.4901,
+      "step": 10704
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996905237656566e-05,
+      "loss": 0.4596,
+      "step": 10705
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996903946012597e-05,
+      "loss": 0.4106,
+      "step": 10706
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9969026540995585e-05,
+      "loss": 0.4831,
+      "step": 10707
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996901361917452e-05,
+      "loss": 0.4415,
+      "step": 10708
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9969000694662767e-05,
+      "loss": 0.4846,
+      "step": 10709
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968987767460337e-05,
+      "loss": 0.5344,
+      "step": 10710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996897483756723e-05,
+      "loss": 0.5003,
+      "step": 10711
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996896190498345e-05,
+      "loss": 0.4395,
+      "step": 10712
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968948969708996e-05,
+      "loss": 0.425,
+      "step": 10713
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996893603174388e-05,
+      "loss": 0.4514,
+      "step": 10714
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.99689230910881e-05,
+      "loss": 0.4606,
+      "step": 10715
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968910147741664e-05,
+      "loss": 0.5171,
+      "step": 10716
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968897201704572e-05,
+      "loss": 0.4683,
+      "step": 10717
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996888425297683e-05,
+      "loss": 0.4957,
+      "step": 10718
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968871301558432e-05,
+      "loss": 0.453,
+      "step": 10719
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968858347449394e-05,
+      "loss": 0.4453,
+      "step": 10720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968845390649716e-05,
+      "loss": 0.4648,
+      "step": 10721
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968832431159396e-05,
+      "loss": 0.5083,
+      "step": 10722
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968819468978446e-05,
+      "loss": 0.5009,
+      "step": 10723
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996880650410686e-05,
+      "loss": 0.4367,
+      "step": 10724
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996879353654465e-05,
+      "loss": 0.455,
+      "step": 10725
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968780566291818e-05,
+      "loss": 0.5967,
+      "step": 10726
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996876759334836e-05,
+      "loss": 0.4564,
+      "step": 10727
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996875461771429e-05,
+      "loss": 0.4794,
+      "step": 10728
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968741639389603e-05,
+      "loss": 0.4425,
+      "step": 10729
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996872865837431e-05,
+      "loss": 0.4073,
+      "step": 10730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968715674668405e-05,
+      "loss": 0.427,
+      "step": 10731
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968702688271898e-05,
+      "loss": 0.3915,
+      "step": 10732
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968689699184797e-05,
+      "loss": 0.4052,
+      "step": 10733
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968676707407096e-05,
+      "loss": 0.4397,
+      "step": 10734
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968663712938802e-05,
+      "loss": 0.4709,
+      "step": 10735
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968650715779924e-05,
+      "loss": 0.4257,
+      "step": 10736
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968637715930453e-05,
+      "loss": 0.463,
+      "step": 10737
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968624713390406e-05,
+      "loss": 0.4682,
+      "step": 10738
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996861170815978e-05,
+      "loss": 0.505,
+      "step": 10739
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968598700238575e-05,
+      "loss": 0.4552,
+      "step": 10740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.99685856896268e-05,
+      "loss": 0.52,
+      "step": 10741
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996857267632446e-05,
+      "loss": 0.4074,
+      "step": 10742
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968559660331556e-05,
+      "loss": 0.4774,
+      "step": 10743
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968546641648084e-05,
+      "loss": 0.4348,
+      "step": 10744
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968533620274062e-05,
+      "loss": 0.525,
+      "step": 10745
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968520596209485e-05,
+      "loss": 0.5316,
+      "step": 10746
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968507569454355e-05,
+      "loss": 0.5524,
+      "step": 10747
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996849454000868e-05,
+      "loss": 0.5249,
+      "step": 10748
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968481507872463e-05,
+      "loss": 0.456,
+      "step": 10749
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968468473045705e-05,
+      "loss": 0.4438,
+      "step": 10750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968455435528408e-05,
+      "loss": 0.4996,
+      "step": 10751
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996844239532058e-05,
+      "loss": 0.5141,
+      "step": 10752
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968429352422226e-05,
+      "loss": 0.4752,
+      "step": 10753
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968416306833345e-05,
+      "loss": 0.4706,
+      "step": 10754
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996840325855394e-05,
+      "loss": 0.4846,
+      "step": 10755
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968390207584017e-05,
+      "loss": 0.6183,
+      "step": 10756
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968377153923576e-05,
+      "loss": 0.4657,
+      "step": 10757
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996836409757263e-05,
+      "loss": 0.4641,
+      "step": 10758
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968351038531172e-05,
+      "loss": 0.456,
+      "step": 10759
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996833797679921e-05,
+      "loss": 0.5076,
+      "step": 10760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968324912376747e-05,
+      "loss": 0.4638,
+      "step": 10761
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968311845263787e-05,
+      "loss": 0.5146,
+      "step": 10762
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996829877546033e-05,
+      "loss": 0.4125,
+      "step": 10763
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968285702966384e-05,
+      "loss": 0.431,
+      "step": 10764
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968272627781955e-05,
+      "loss": 0.4965,
+      "step": 10765
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968259549907038e-05,
+      "loss": 0.5818,
+      "step": 10766
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968246469341643e-05,
+      "loss": 0.4566,
+      "step": 10767
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968233386085773e-05,
+      "loss": 0.5255,
+      "step": 10768
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968220300139428e-05,
+      "loss": 0.4106,
+      "step": 10769
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968207211502615e-05,
+      "loss": 0.5113,
+      "step": 10770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968194120175334e-05,
+      "loss": 0.5249,
+      "step": 10771
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968181026157595e-05,
+      "loss": 0.4744,
+      "step": 10772
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996816792944939e-05,
+      "loss": 0.5074,
+      "step": 10773
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968154830050737e-05,
+      "loss": 0.531,
+      "step": 10774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996814172796163e-05,
+      "loss": 0.6041,
+      "step": 10775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968128623182074e-05,
+      "loss": 0.5233,
+      "step": 10776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968115515712077e-05,
+      "loss": 0.4973,
+      "step": 10777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968102405551635e-05,
+      "loss": 0.4968,
+      "step": 10778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996808929270076e-05,
+      "loss": 0.4522,
+      "step": 10779
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996807617715945e-05,
+      "loss": 0.4807,
+      "step": 10780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968063058927704e-05,
+      "loss": 0.4507,
+      "step": 10781
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968049938005536e-05,
+      "loss": 0.5064,
+      "step": 10782
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968036814392944e-05,
+      "loss": 0.4716,
+      "step": 10783
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996802368808993e-05,
+      "loss": 0.4484,
+      "step": 10784
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9968010559096504e-05,
+      "loss": 0.4435,
+      "step": 10785
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9967997427412664e-05,
+      "loss": 0.4274,
+      "step": 10786
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9967984293038413e-05,
+      "loss": 0.41,
+      "step": 10787
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9967971155973758e-05,
+      "loss": 0.4513,
+      "step": 10788
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.99679580162187e-05,
+      "loss": 0.4273,
+      "step": 10789
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9967944873773245e-05,
+      "loss": 0.499,
+      "step": 10790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9967931728637393e-05,
+      "loss": 0.4692,
+      "step": 10791
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.996791858081115e-05,
+      "loss": 0.4224,
+      "step": 10792
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9967905430294522e-05,
+      "loss": 0.527,
+      "step": 10793
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967892277087507e-05,
+      "loss": 0.4492,
+      "step": 10794
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996787912119011e-05,
+      "loss": 0.5081,
+      "step": 10795
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996786596260234e-05,
+      "loss": 0.5133,
+      "step": 10796
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967852801324194e-05,
+      "loss": 0.4782,
+      "step": 10797
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996783963735568e-05,
+      "loss": 0.5298,
+      "step": 10798
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967826470696795e-05,
+      "loss": 0.5009,
+      "step": 10799
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996781330134755e-05,
+      "loss": 0.5407,
+      "step": 10800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967800129307947e-05,
+      "loss": 0.4245,
+      "step": 10801
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967786954577987e-05,
+      "loss": 0.5018,
+      "step": 10802
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967773777157674e-05,
+      "loss": 0.4644,
+      "step": 10803
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967760597047012e-05,
+      "loss": 0.4365,
+      "step": 10804
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967747414246003e-05,
+      "loss": 0.4699,
+      "step": 10805
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967734228754656e-05,
+      "loss": 0.4145,
+      "step": 10806
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967721040572972e-05,
+      "loss": 0.5481,
+      "step": 10807
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996770784970095e-05,
+      "loss": 0.5955,
+      "step": 10808
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967694656138598e-05,
+      "loss": 0.4791,
+      "step": 10809
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996768145988592e-05,
+      "loss": 0.5246,
+      "step": 10810
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967668260942914e-05,
+      "loss": 0.5675,
+      "step": 10811
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996765505930959e-05,
+      "loss": 0.459,
+      "step": 10812
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996764185498595e-05,
+      "loss": 0.4183,
+      "step": 10813
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967628647971997e-05,
+      "loss": 0.4535,
+      "step": 10814
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967615438267735e-05,
+      "loss": 0.4032,
+      "step": 10815
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967602225873167e-05,
+      "loss": 0.4742,
+      "step": 10816
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967589010788298e-05,
+      "loss": 0.4998,
+      "step": 10817
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967575793013126e-05,
+      "loss": 0.5162,
+      "step": 10818
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996756257254766e-05,
+      "loss": 0.4823,
+      "step": 10819
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967549349391904e-05,
+      "loss": 0.5709,
+      "step": 10820
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967536123545857e-05,
+      "loss": 0.4658,
+      "step": 10821
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996752289500953e-05,
+      "loss": 0.5,
+      "step": 10822
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996750966378292e-05,
+      "loss": 0.4395,
+      "step": 10823
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996749642986603e-05,
+      "loss": 0.498,
+      "step": 10824
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967483193258866e-05,
+      "loss": 0.5234,
+      "step": 10825
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967469953961434e-05,
+      "loss": 0.464,
+      "step": 10826
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967456711973735e-05,
+      "loss": 0.5147,
+      "step": 10827
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967443467295774e-05,
+      "loss": 0.5134,
+      "step": 10828
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996743021992755e-05,
+      "loss": 0.5419,
+      "step": 10829
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967416969869074e-05,
+      "loss": 0.4888,
+      "step": 10830
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967403717120342e-05,
+      "loss": 0.4253,
+      "step": 10831
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967390461681363e-05,
+      "loss": 0.4803,
+      "step": 10832
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996737720355214e-05,
+      "loss": 0.4629,
+      "step": 10833
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967363942732674e-05,
+      "loss": 0.4646,
+      "step": 10834
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996735067922297e-05,
+      "loss": 0.4384,
+      "step": 10835
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967337413023032e-05,
+      "loss": 0.4592,
+      "step": 10836
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996732414413286e-05,
+      "loss": 0.4638,
+      "step": 10837
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967310872552463e-05,
+      "loss": 0.4591,
+      "step": 10838
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967297598281844e-05,
+      "loss": 0.578,
+      "step": 10839
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967284321321e-05,
+      "loss": 0.4538,
+      "step": 10840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967271041669943e-05,
+      "loss": 0.465,
+      "step": 10841
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967257759328673e-05,
+      "loss": 0.518,
+      "step": 10842
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996724447429719e-05,
+      "loss": 0.5035,
+      "step": 10843
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967231186575506e-05,
+      "loss": 0.4808,
+      "step": 10844
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967217896163617e-05,
+      "loss": 0.4854,
+      "step": 10845
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996720460306153e-05,
+      "loss": 0.5588,
+      "step": 10846
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967191307269248e-05,
+      "loss": 0.4505,
+      "step": 10847
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967178008786775e-05,
+      "loss": 0.4849,
+      "step": 10848
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996716470761411e-05,
+      "loss": 0.4817,
+      "step": 10849
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967151403751265e-05,
+      "loss": 0.4109,
+      "step": 10850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967138097198238e-05,
+      "loss": 0.4597,
+      "step": 10851
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967124787955035e-05,
+      "loss": 0.4562,
+      "step": 10852
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967111476021658e-05,
+      "loss": 0.4902,
+      "step": 10853
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996709816139811e-05,
+      "loss": 0.3976,
+      "step": 10854
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967084844084395e-05,
+      "loss": 0.5274,
+      "step": 10855
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996707152408052e-05,
+      "loss": 0.5115,
+      "step": 10856
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967058201386483e-05,
+      "loss": 0.4729,
+      "step": 10857
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996704487600229e-05,
+      "loss": 0.446,
+      "step": 10858
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996703154792795e-05,
+      "loss": 0.5393,
+      "step": 10859
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967018217163455e-05,
+      "loss": 0.4285,
+      "step": 10860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9967004883708818e-05,
+      "loss": 0.4017,
+      "step": 10861
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996699154756404e-05,
+      "loss": 0.5354,
+      "step": 10862
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966978208729125e-05,
+      "loss": 0.4386,
+      "step": 10863
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966964867204077e-05,
+      "loss": 0.4715,
+      "step": 10864
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966951522988895e-05,
+      "loss": 0.4881,
+      "step": 10865
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966938176083592e-05,
+      "loss": 0.477,
+      "step": 10866
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996692482648816e-05,
+      "loss": 0.4223,
+      "step": 10867
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966911474202607e-05,
+      "loss": 0.5739,
+      "step": 10868
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966898119226946e-05,
+      "loss": 0.4366,
+      "step": 10869
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966884761561166e-05,
+      "loss": 0.5308,
+      "step": 10870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966871401205278e-05,
+      "loss": 0.5014,
+      "step": 10871
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966858038159287e-05,
+      "loss": 0.4389,
+      "step": 10872
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966844672423194e-05,
+      "loss": 0.4433,
+      "step": 10873
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966831303997002e-05,
+      "loss": 0.5335,
+      "step": 10874
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966817932880715e-05,
+      "loss": 0.4331,
+      "step": 10875
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966804559074337e-05,
+      "loss": 0.5849,
+      "step": 10876
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966791182577876e-05,
+      "loss": 0.4488,
+      "step": 10877
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996677780339133e-05,
+      "loss": 0.4418,
+      "step": 10878
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966764421514703e-05,
+      "loss": 0.4493,
+      "step": 10879
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966751036947997e-05,
+      "loss": 0.4888,
+      "step": 10880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966737649691222e-05,
+      "loss": 0.5115,
+      "step": 10881
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966724259744373e-05,
+      "loss": 0.5174,
+      "step": 10882
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966710867107463e-05,
+      "loss": 0.4475,
+      "step": 10883
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996669747178049e-05,
+      "loss": 0.5225,
+      "step": 10884
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966684073763464e-05,
+      "loss": 0.5853,
+      "step": 10885
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966670673056376e-05,
+      "loss": 0.4686,
+      "step": 10886
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.996665726965924e-05,
+      "loss": 0.4004,
+      "step": 10887
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966643863572056e-05,
+      "loss": 0.4986,
+      "step": 10888
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966630454794827e-05,
+      "loss": 0.464,
+      "step": 10889
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966617043327557e-05,
+      "loss": 0.5195,
+      "step": 10890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9966603629170257e-05,
+      "loss": 0.4408,
+      "step": 10891
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996659021232292e-05,
+      "loss": 0.4527,
+      "step": 10892
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966576792785553e-05,
+      "loss": 0.4221,
+      "step": 10893
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996656337055816e-05,
+      "loss": 0.4963,
+      "step": 10894
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966549945640745e-05,
+      "loss": 0.4228,
+      "step": 10895
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966536518033314e-05,
+      "loss": 0.4674,
+      "step": 10896
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966523087735865e-05,
+      "loss": 0.4792,
+      "step": 10897
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996650965474841e-05,
+      "loss": 0.5358,
+      "step": 10898
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996649621907094e-05,
+      "loss": 0.5315,
+      "step": 10899
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966482780703468e-05,
+      "loss": 0.4377,
+      "step": 10900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966469339646e-05,
+      "loss": 0.5984,
+      "step": 10901
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996645589589853e-05,
+      "loss": 0.4959,
+      "step": 10902
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966442449461073e-05,
+      "loss": 0.5157,
+      "step": 10903
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996642900033362e-05,
+      "loss": 0.4186,
+      "step": 10904
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996641554851619e-05,
+      "loss": 0.5396,
+      "step": 10905
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996640209400877e-05,
+      "loss": 0.6043,
+      "step": 10906
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966388636811374e-05,
+      "loss": 0.4237,
+      "step": 10907
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966375176924004e-05,
+      "loss": 0.4496,
+      "step": 10908
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996636171434666e-05,
+      "loss": 0.4214,
+      "step": 10909
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966348249079352e-05,
+      "loss": 0.4659,
+      "step": 10910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996633478112208e-05,
+      "loss": 0.4384,
+      "step": 10911
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966321310474844e-05,
+      "loss": 0.5561,
+      "step": 10912
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966307837137652e-05,
+      "loss": 0.4731,
+      "step": 10913
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996629436111051e-05,
+      "loss": 0.5084,
+      "step": 10914
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966280882393416e-05,
+      "loss": 0.4707,
+      "step": 10915
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996626740098638e-05,
+      "loss": 0.4996,
+      "step": 10916
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966253916889398e-05,
+      "loss": 0.4875,
+      "step": 10917
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966240430102478e-05,
+      "loss": 0.4708,
+      "step": 10918
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966226940625622e-05,
+      "loss": 0.5047,
+      "step": 10919
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966213448458837e-05,
+      "loss": 0.5158,
+      "step": 10920
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966199953602122e-05,
+      "loss": 0.5119,
+      "step": 10921
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966186456055485e-05,
+      "loss": 0.4915,
+      "step": 10922
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996617295581893e-05,
+      "loss": 0.4598,
+      "step": 10923
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966159452892454e-05,
+      "loss": 0.4387,
+      "step": 10924
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966145947276066e-05,
+      "loss": 0.4495,
+      "step": 10925
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996613243896977e-05,
+      "loss": 0.4471,
+      "step": 10926
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966118927973567e-05,
+      "loss": 0.4691,
+      "step": 10927
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966105414287466e-05,
+      "loss": 0.4958,
+      "step": 10928
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996609189791146e-05,
+      "loss": 0.4861,
+      "step": 10929
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996607837884556e-05,
+      "loss": 0.4239,
+      "step": 10930
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966064857089776e-05,
+      "loss": 0.5292,
+      "step": 10931
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966051332644097e-05,
+      "loss": 0.4736,
+      "step": 10932
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996603780550854e-05,
+      "loss": 0.5271,
+      "step": 10933
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966024275683096e-05,
+      "loss": 0.4572,
+      "step": 10934
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9966010743167777e-05,
+      "loss": 0.5383,
+      "step": 10935
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996599720796259e-05,
+      "loss": 0.5152,
+      "step": 10936
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965983670067532e-05,
+      "loss": 0.4672,
+      "step": 10937
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965970129482606e-05,
+      "loss": 0.3641,
+      "step": 10938
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996595658620782e-05,
+      "loss": 0.4631,
+      "step": 10939
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965943040243175e-05,
+      "loss": 0.5278,
+      "step": 10940
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965929491588674e-05,
+      "loss": 0.4791,
+      "step": 10941
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965915940244324e-05,
+      "loss": 0.5186,
+      "step": 10942
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965902386210127e-05,
+      "loss": 0.4969,
+      "step": 10943
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965888829486086e-05,
+      "loss": 0.4515,
+      "step": 10944
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965875270072206e-05,
+      "loss": 0.5294,
+      "step": 10945
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965861707968485e-05,
+      "loss": 0.4782,
+      "step": 10946
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965848143174934e-05,
+      "loss": 0.4204,
+      "step": 10947
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965834575691558e-05,
+      "loss": 0.6034,
+      "step": 10948
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996582100551835e-05,
+      "loss": 0.5101,
+      "step": 10949
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965807432655326e-05,
+      "loss": 0.5958,
+      "step": 10950
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965793857102484e-05,
+      "loss": 0.5653,
+      "step": 10951
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965780278859824e-05,
+      "loss": 0.5216,
+      "step": 10952
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965766697927354e-05,
+      "loss": 0.5146,
+      "step": 10953
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965753114305076e-05,
+      "loss": 0.5295,
+      "step": 10954
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965739527992998e-05,
+      "loss": 0.4219,
+      "step": 10955
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996572593899112e-05,
+      "loss": 0.4293,
+      "step": 10956
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965712347299443e-05,
+      "loss": 0.4526,
+      "step": 10957
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965698752917973e-05,
+      "loss": 0.505,
+      "step": 10958
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996568515584672e-05,
+      "loss": 0.4558,
+      "step": 10959
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965671556085676e-05,
+      "loss": 0.4682,
+      "step": 10960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965657953634852e-05,
+      "loss": 0.453,
+      "step": 10961
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996564434849425e-05,
+      "loss": 0.4806,
+      "step": 10962
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996563074066388e-05,
+      "loss": 0.4947,
+      "step": 10963
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965617130143733e-05,
+      "loss": 0.5189,
+      "step": 10964
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965603516933822e-05,
+      "loss": 0.509,
+      "step": 10965
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965589901034148e-05,
+      "loss": 0.445,
+      "step": 10966
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965576282444714e-05,
+      "loss": 0.4756,
+      "step": 10967
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965562661165526e-05,
+      "loss": 0.4803,
+      "step": 10968
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965549037196586e-05,
+      "loss": 0.5121,
+      "step": 10969
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965535410537896e-05,
+      "loss": 0.4823,
+      "step": 10970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965521781189462e-05,
+      "loss": 0.439,
+      "step": 10971
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965508149151286e-05,
+      "loss": 0.4743,
+      "step": 10972
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965494514423374e-05,
+      "loss": 0.4751,
+      "step": 10973
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996548087700573e-05,
+      "loss": 0.5265,
+      "step": 10974
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996546723689835e-05,
+      "loss": 0.4626,
+      "step": 10975
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996545359410125e-05,
+      "loss": 0.4689,
+      "step": 10976
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965439948614427e-05,
+      "loss": 0.4346,
+      "step": 10977
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965426300437882e-05,
+      "loss": 0.477,
+      "step": 10978
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965412649571623e-05,
+      "loss": 0.483,
+      "step": 10979
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965398996015653e-05,
+      "loss": 0.4437,
+      "step": 10980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965385339769976e-05,
+      "loss": 0.4292,
+      "step": 10981
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965371680834594e-05,
+      "loss": 0.4464,
+      "step": 10982
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965358019209513e-05,
+      "loss": 0.5285,
+      "step": 10983
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965344354894733e-05,
+      "loss": 0.511,
+      "step": 10984
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.996533068789026e-05,
+      "loss": 0.4221,
+      "step": 10985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.99653170181961e-05,
+      "loss": 0.4962,
+      "step": 10986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965303345812254e-05,
+      "loss": 0.4057,
+      "step": 10987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965289670738726e-05,
+      "loss": 0.5005,
+      "step": 10988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965275992975516e-05,
+      "loss": 0.4875,
+      "step": 10989
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9965262312522634e-05,
+      "loss": 0.43,
+      "step": 10990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965248629380084e-05,
+      "loss": 0.4194,
+      "step": 10991
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965234943547865e-05,
+      "loss": 0.443,
+      "step": 10992
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996522125502598e-05,
+      "loss": 0.4181,
+      "step": 10993
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965207563814436e-05,
+      "loss": 0.4116,
+      "step": 10994
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996519386991324e-05,
+      "loss": 0.4127,
+      "step": 10995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996518017332239e-05,
+      "loss": 0.4577,
+      "step": 10996
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965166474041887e-05,
+      "loss": 0.4346,
+      "step": 10997
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965152772071743e-05,
+      "loss": 0.4772,
+      "step": 10998
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965139067411957e-05,
+      "loss": 0.4421,
+      "step": 10999
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965125360062533e-05,
+      "loss": 0.435,
+      "step": 11000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965111650023475e-05,
+      "loss": 0.4895,
+      "step": 11001
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965097937294785e-05,
+      "loss": 0.5802,
+      "step": 11002
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996508422187647e-05,
+      "loss": 0.4791,
+      "step": 11003
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965070503768533e-05,
+      "loss": 0.4546,
+      "step": 11004
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965056782970977e-05,
+      "loss": 0.427,
+      "step": 11005
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965043059483804e-05,
+      "loss": 0.4618,
+      "step": 11006
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996502933330702e-05,
+      "loss": 0.4767,
+      "step": 11007
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965015604440628e-05,
+      "loss": 0.5689,
+      "step": 11008
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9965001872884632e-05,
+      "loss": 0.4491,
+      "step": 11009
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964988138639036e-05,
+      "loss": 0.4858,
+      "step": 11010
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964974401703842e-05,
+      "loss": 0.5212,
+      "step": 11011
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964960662079055e-05,
+      "loss": 0.4467,
+      "step": 11012
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996494691976468e-05,
+      "loss": 0.413,
+      "step": 11013
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964933174760718e-05,
+      "loss": 0.5193,
+      "step": 11014
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964919427067173e-05,
+      "loss": 0.4746,
+      "step": 11015
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996490567668405e-05,
+      "loss": 0.4713,
+      "step": 11016
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964891923611354e-05,
+      "loss": 0.5397,
+      "step": 11017
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964878167849088e-05,
+      "loss": 0.4876,
+      "step": 11018
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964864409397252e-05,
+      "loss": 0.4712,
+      "step": 11019
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964850648255852e-05,
+      "loss": 0.4473,
+      "step": 11020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964836884424896e-05,
+      "loss": 0.505,
+      "step": 11021
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996482311790438e-05,
+      "loss": 0.481,
+      "step": 11022
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964809348694313e-05,
+      "loss": 0.4738,
+      "step": 11023
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964795576794697e-05,
+      "loss": 0.4524,
+      "step": 11024
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964781802205538e-05,
+      "loss": 0.4639,
+      "step": 11025
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964768024926836e-05,
+      "loss": 0.4914,
+      "step": 11026
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964754244958598e-05,
+      "loss": 0.4365,
+      "step": 11027
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964740462300823e-05,
+      "loss": 0.5074,
+      "step": 11028
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964726676953522e-05,
+      "loss": 0.4518,
+      "step": 11029
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964712888916692e-05,
+      "loss": 0.466,
+      "step": 11030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964699098190342e-05,
+      "loss": 0.4764,
+      "step": 11031
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996468530477447e-05,
+      "loss": 0.4378,
+      "step": 11032
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964671508669085e-05,
+      "loss": 0.4713,
+      "step": 11033
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964657709874187e-05,
+      "loss": 0.4712,
+      "step": 11034
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964643908389784e-05,
+      "loss": 0.3957,
+      "step": 11035
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964630104215875e-05,
+      "loss": 0.4946,
+      "step": 11036
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964616297352467e-05,
+      "loss": 0.6101,
+      "step": 11037
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996460248779956e-05,
+      "loss": 0.4633,
+      "step": 11038
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996458867555716e-05,
+      "loss": 0.4426,
+      "step": 11039
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964574860625276e-05,
+      "loss": 0.4796,
+      "step": 11040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964561043003902e-05,
+      "loss": 0.5308,
+      "step": 11041
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996454722269305e-05,
+      "loss": 0.4925,
+      "step": 11042
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964533399692715e-05,
+      "loss": 0.5067,
+      "step": 11043
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996451957400291e-05,
+      "loss": 0.4309,
+      "step": 11044
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964505745623634e-05,
+      "loss": 0.406,
+      "step": 11045
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996449191455489e-05,
+      "loss": 0.4928,
+      "step": 11046
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964478080796682e-05,
+      "loss": 0.4541,
+      "step": 11047
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996446424434902e-05,
+      "loss": 0.4864,
+      "step": 11048
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964450405211897e-05,
+      "loss": 0.4433,
+      "step": 11049
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964436563385323e-05,
+      "loss": 0.6069,
+      "step": 11050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964422718869302e-05,
+      "loss": 0.502,
+      "step": 11051
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996440887166384e-05,
+      "loss": 0.453,
+      "step": 11052
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996439502176893e-05,
+      "loss": 0.4813,
+      "step": 11053
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964381169184592e-05,
+      "loss": 0.4628,
+      "step": 11054
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964367313910815e-05,
+      "loss": 0.4738,
+      "step": 11055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996435345594761e-05,
+      "loss": 0.4643,
+      "step": 11056
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996433959529498e-05,
+      "loss": 0.4893,
+      "step": 11057
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964325731952926e-05,
+      "loss": 0.4741,
+      "step": 11058
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964311865921458e-05,
+      "loss": 0.4895,
+      "step": 11059
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964297997200576e-05,
+      "loss": 0.4492,
+      "step": 11060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996428412579028e-05,
+      "loss": 0.494,
+      "step": 11061
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964270251690578e-05,
+      "loss": 0.4446,
+      "step": 11062
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996425637490147e-05,
+      "loss": 0.4151,
+      "step": 11063
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964242495422967e-05,
+      "loss": 0.4473,
+      "step": 11064
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964228613255067e-05,
+      "loss": 0.5175,
+      "step": 11065
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964214728397776e-05,
+      "loss": 0.4055,
+      "step": 11066
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964200840851092e-05,
+      "loss": 0.4334,
+      "step": 11067
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996418695061503e-05,
+      "loss": 0.4284,
+      "step": 11068
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964173057689585e-05,
+      "loss": 0.4647,
+      "step": 11069
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964159162074763e-05,
+      "loss": 0.499,
+      "step": 11070
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964145263770567e-05,
+      "loss": 0.4845,
+      "step": 11071
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964131362777e-05,
+      "loss": 0.5154,
+      "step": 11072
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964117459094074e-05,
+      "loss": 0.482,
+      "step": 11073
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996410355272178e-05,
+      "loss": 0.4533,
+      "step": 11074
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996408964366013e-05,
+      "loss": 0.4482,
+      "step": 11075
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964075731909124e-05,
+      "loss": 0.4108,
+      "step": 11076
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996406181746877e-05,
+      "loss": 0.4683,
+      "step": 11077
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9964047900339068e-05,
+      "loss": 0.4839,
+      "step": 11078
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996403398052002e-05,
+      "loss": 0.4448,
+      "step": 11079
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996402005801164e-05,
+      "loss": 0.4718,
+      "step": 11080
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996400613281392e-05,
+      "loss": 0.52,
+      "step": 11081
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9963992204926864e-05,
+      "loss": 0.4469,
+      "step": 11082
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9963978274350486e-05,
+      "loss": 0.5274,
+      "step": 11083
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996396434108478e-05,
+      "loss": 0.515,
+      "step": 11084
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9963950405129754e-05,
+      "loss": 0.4791,
+      "step": 11085
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9963936466485414e-05,
+      "loss": 0.5493,
+      "step": 11086
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.996392252515176e-05,
+      "loss": 0.4527,
+      "step": 11087
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9963908581128796e-05,
+      "loss": 0.3919,
+      "step": 11088
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9963894634416524e-05,
+      "loss": 0.5434,
+      "step": 11089
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963880685014957e-05,
+      "loss": 0.4317,
+      "step": 11090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963866732924088e-05,
+      "loss": 0.4418,
+      "step": 11091
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963852778143923e-05,
+      "loss": 0.5428,
+      "step": 11092
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963838820674472e-05,
+      "loss": 0.4626,
+      "step": 11093
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996382486051573e-05,
+      "loss": 0.4689,
+      "step": 11094
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963810897667707e-05,
+      "loss": 0.5082,
+      "step": 11095
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963796932130403e-05,
+      "loss": 0.5425,
+      "step": 11096
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963782963903827e-05,
+      "loss": 0.4609,
+      "step": 11097
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963768992987975e-05,
+      "loss": 0.4803,
+      "step": 11098
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996375501938286e-05,
+      "loss": 0.5072,
+      "step": 11099
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963741043088477e-05,
+      "loss": 0.5036,
+      "step": 11100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963727064104833e-05,
+      "loss": 0.4991,
+      "step": 11101
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963713082431937e-05,
+      "loss": 0.4901,
+      "step": 11102
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963699098069784e-05,
+      "loss": 0.4991,
+      "step": 11103
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963685111018385e-05,
+      "loss": 0.5261,
+      "step": 11104
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996367112127774e-05,
+      "loss": 0.4072,
+      "step": 11105
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963657128847856e-05,
+      "loss": 0.5014,
+      "step": 11106
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996364313372873e-05,
+      "loss": 0.5179,
+      "step": 11107
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963629135920374e-05,
+      "loss": 0.4237,
+      "step": 11108
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963615135422785e-05,
+      "loss": 0.5179,
+      "step": 11109
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996360113223597e-05,
+      "loss": 0.4603,
+      "step": 11110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963587126359932e-05,
+      "loss": 0.4833,
+      "step": 11111
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963573117794676e-05,
+      "loss": 0.4444,
+      "step": 11112
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963559106540208e-05,
+      "loss": 0.4737,
+      "step": 11113
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963545092596526e-05,
+      "loss": 0.4746,
+      "step": 11114
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963531075963637e-05,
+      "loss": 0.5952,
+      "step": 11115
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963517056641543e-05,
+      "loss": 0.4444,
+      "step": 11116
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996350303463025e-05,
+      "loss": 0.459,
+      "step": 11117
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963489009929763e-05,
+      "loss": 0.4933,
+      "step": 11118
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996347498254008e-05,
+      "loss": 0.4626,
+      "step": 11119
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996346095246121e-05,
+      "loss": 0.4478,
+      "step": 11120
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963446919693156e-05,
+      "loss": 0.4738,
+      "step": 11121
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963432884235922e-05,
+      "loss": 0.4916,
+      "step": 11122
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963418846089507e-05,
+      "loss": 0.4863,
+      "step": 11123
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963404805253925e-05,
+      "loss": 0.5617,
+      "step": 11124
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996339076172917e-05,
+      "loss": 0.4686,
+      "step": 11125
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963376715515246e-05,
+      "loss": 0.462,
+      "step": 11126
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963362666612167e-05,
+      "loss": 0.6041,
+      "step": 11127
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963348615019927e-05,
+      "loss": 0.4239,
+      "step": 11128
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996333456073853e-05,
+      "loss": 0.4082,
+      "step": 11129
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963320503767984e-05,
+      "loss": 0.5133,
+      "step": 11130
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963306444108294e-05,
+      "loss": 0.4964,
+      "step": 11131
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963292381759458e-05,
+      "loss": 0.4485,
+      "step": 11132
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996327831672148e-05,
+      "loss": 0.4485,
+      "step": 11133
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996326424899437e-05,
+      "loss": 0.4448,
+      "step": 11134
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963250178578132e-05,
+      "loss": 0.5128,
+      "step": 11135
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996323610547276e-05,
+      "loss": 0.4607,
+      "step": 11136
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996322202967827e-05,
+      "loss": 0.4628,
+      "step": 11137
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963207951194654e-05,
+      "loss": 0.6153,
+      "step": 11138
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963193870021923e-05,
+      "loss": 0.4826,
+      "step": 11139
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963179786160082e-05,
+      "loss": 0.4974,
+      "step": 11140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996316569960913e-05,
+      "loss": 0.4889,
+      "step": 11141
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963151610369074e-05,
+      "loss": 0.4373,
+      "step": 11142
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963137518439915e-05,
+      "loss": 0.5327,
+      "step": 11143
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996312342382166e-05,
+      "loss": 0.4182,
+      "step": 11144
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996310932651431e-05,
+      "loss": 0.5098,
+      "step": 11145
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963095226517876e-05,
+      "loss": 0.5037,
+      "step": 11146
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9963081123832348e-05,
+      "loss": 0.472,
+      "step": 11147
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996306701845774e-05,
+      "loss": 0.5014,
+      "step": 11148
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996305291039406e-05,
+      "loss": 0.4984,
+      "step": 11149
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.99630387996413e-05,
+      "loss": 0.4862,
+      "step": 11150
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996302468619947e-05,
+      "loss": 0.5318,
+      "step": 11151
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996301057006857e-05,
+      "loss": 0.5121,
+      "step": 11152
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962996451248614e-05,
+      "loss": 0.552,
+      "step": 11153
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962982329739592e-05,
+      "loss": 0.4223,
+      "step": 11154
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962968205541514e-05,
+      "loss": 0.5382,
+      "step": 11155
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962954078654388e-05,
+      "loss": 0.482,
+      "step": 11156
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962939949078217e-05,
+      "loss": 0.5964,
+      "step": 11157
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962925816812997e-05,
+      "loss": 0.4144,
+      "step": 11158
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962911681858736e-05,
+      "loss": 0.4985,
+      "step": 11159
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962897544215443e-05,
+      "loss": 0.5464,
+      "step": 11160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962883403883112e-05,
+      "loss": 0.4749,
+      "step": 11161
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996286926086176e-05,
+      "loss": 0.5088,
+      "step": 11162
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962855115151376e-05,
+      "loss": 0.4713,
+      "step": 11163
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996284096675197e-05,
+      "loss": 0.5366,
+      "step": 11164
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962826815663556e-05,
+      "loss": 0.5092,
+      "step": 11165
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962812661886118e-05,
+      "loss": 0.5063,
+      "step": 11166
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962798505419677e-05,
+      "loss": 0.4751,
+      "step": 11167
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962784346264227e-05,
+      "loss": 0.5339,
+      "step": 11168
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962770184419777e-05,
+      "loss": 0.4552,
+      "step": 11169
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962756019886325e-05,
+      "loss": 0.52,
+      "step": 11170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996274185266388e-05,
+      "loss": 0.4819,
+      "step": 11171
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962727682752447e-05,
+      "loss": 0.4552,
+      "step": 11172
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962713510152027e-05,
+      "loss": 0.4446,
+      "step": 11173
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962699334862622e-05,
+      "loss": 0.4104,
+      "step": 11174
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962685156884237e-05,
+      "loss": 0.4999,
+      "step": 11175
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996267097621688e-05,
+      "loss": 0.4501,
+      "step": 11176
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996265679286055e-05,
+      "loss": 0.4865,
+      "step": 11177
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996264260681525e-05,
+      "loss": 0.5579,
+      "step": 11178
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962628418080986e-05,
+      "loss": 0.4511,
+      "step": 11179
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962614226657766e-05,
+      "loss": 0.4538,
+      "step": 11180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996260003254559e-05,
+      "loss": 0.4705,
+      "step": 11181
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962585835744458e-05,
+      "loss": 0.448,
+      "step": 11182
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962571636254377e-05,
+      "loss": 0.5424,
+      "step": 11183
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962557434075356e-05,
+      "loss": 0.5645,
+      "step": 11184
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996254322920739e-05,
+      "loss": 0.5057,
+      "step": 11185
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.996252902165049e-05,
+      "loss": 0.4303,
+      "step": 11186
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9962514811404657e-05,
+      "loss": 0.5027,
+      "step": 11187
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962500598469893e-05,
+      "loss": 0.5377,
+      "step": 11188
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962486382846206e-05,
+      "loss": 0.4317,
+      "step": 11189
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962472164533593e-05,
+      "loss": 0.5986,
+      "step": 11190
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962457943532067e-05,
+      "loss": 0.5071,
+      "step": 11191
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996244371984162e-05,
+      "loss": 0.4707,
+      "step": 11192
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996242949346227e-05,
+      "loss": 0.5064,
+      "step": 11193
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996241526439401e-05,
+      "loss": 0.4422,
+      "step": 11194
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962401032636847e-05,
+      "loss": 0.4911,
+      "step": 11195
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996238679819079e-05,
+      "loss": 0.4816,
+      "step": 11196
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962372561055834e-05,
+      "loss": 0.4735,
+      "step": 11197
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996235832123199e-05,
+      "loss": 0.4388,
+      "step": 11198
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962344078719255e-05,
+      "loss": 0.5085,
+      "step": 11199
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996232983351764e-05,
+      "loss": 0.4573,
+      "step": 11200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962315585627143e-05,
+      "loss": 0.4298,
+      "step": 11201
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962301335047775e-05,
+      "loss": 0.4609,
+      "step": 11202
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996228708177953e-05,
+      "loss": 0.4417,
+      "step": 11203
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996227282582242e-05,
+      "loss": 0.5405,
+      "step": 11204
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962258567176445e-05,
+      "loss": 0.4131,
+      "step": 11205
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962244305841612e-05,
+      "loss": 0.5843,
+      "step": 11206
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996223004181792e-05,
+      "loss": 0.5052,
+      "step": 11207
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962215775105377e-05,
+      "loss": 0.6162,
+      "step": 11208
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962201505703985e-05,
+      "loss": 0.4079,
+      "step": 11209
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962187233613748e-05,
+      "loss": 0.4549,
+      "step": 11210
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996217295883467e-05,
+      "loss": 0.5551,
+      "step": 11211
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962158681366756e-05,
+      "loss": 0.5517,
+      "step": 11212
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962144401210007e-05,
+      "loss": 0.5007,
+      "step": 11213
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996213011836443e-05,
+      "loss": 0.4707,
+      "step": 11214
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996211583283003e-05,
+      "loss": 0.4527,
+      "step": 11215
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962101544606802e-05,
+      "loss": 0.4289,
+      "step": 11216
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996208725369476e-05,
+      "loss": 0.4262,
+      "step": 11217
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962072960093902e-05,
+      "loss": 0.4223,
+      "step": 11218
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996205866380424e-05,
+      "loss": 0.448,
+      "step": 11219
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962044364825764e-05,
+      "loss": 0.5261,
+      "step": 11220
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996203006315849e-05,
+      "loss": 0.4689,
+      "step": 11221
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9962015758802418e-05,
+      "loss": 0.4688,
+      "step": 11222
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996200145175755e-05,
+      "loss": 0.5093,
+      "step": 11223
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961987142023892e-05,
+      "loss": 0.4574,
+      "step": 11224
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961972829601446e-05,
+      "loss": 0.4916,
+      "step": 11225
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996195851449022e-05,
+      "loss": 0.5254,
+      "step": 11226
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996194419669021e-05,
+      "loss": 0.4799,
+      "step": 11227
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961929876201427e-05,
+      "loss": 0.5094,
+      "step": 11228
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996191555302387e-05,
+      "loss": 0.493,
+      "step": 11229
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996190122715755e-05,
+      "loss": 0.4464,
+      "step": 11230
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961886898602466e-05,
+      "loss": 0.4595,
+      "step": 11231
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996187256735862e-05,
+      "loss": 0.4075,
+      "step": 11232
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996185823342602e-05,
+      "loss": 0.4631,
+      "step": 11233
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996184389680466e-05,
+      "loss": 0.523,
+      "step": 11234
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961829557494563e-05,
+      "loss": 0.4341,
+      "step": 11235
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961815215495716e-05,
+      "loss": 0.5615,
+      "step": 11236
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961800870808128e-05,
+      "loss": 0.4049,
+      "step": 11237
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961786523431807e-05,
+      "loss": 0.496,
+      "step": 11238
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996177217336675e-05,
+      "loss": 0.5037,
+      "step": 11239
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961757820612965e-05,
+      "loss": 0.4018,
+      "step": 11240
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961743465170454e-05,
+      "loss": 0.4577,
+      "step": 11241
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961729107039223e-05,
+      "loss": 0.4563,
+      "step": 11242
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961714746219276e-05,
+      "loss": 0.4811,
+      "step": 11243
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961700382710613e-05,
+      "loss": 0.4147,
+      "step": 11244
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961686016513243e-05,
+      "loss": 0.4297,
+      "step": 11245
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961671647627164e-05,
+      "loss": 0.5308,
+      "step": 11246
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996165727605239e-05,
+      "loss": 0.4741,
+      "step": 11247
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996164290178891e-05,
+      "loss": 0.5766,
+      "step": 11248
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996162852483674e-05,
+      "loss": 0.4457,
+      "step": 11249
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996161414519588e-05,
+      "loss": 0.4485,
+      "step": 11250
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961599762866332e-05,
+      "loss": 0.4764,
+      "step": 11251
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961585377848104e-05,
+      "loss": 0.4844,
+      "step": 11252
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961570990141193e-05,
+      "loss": 0.5025,
+      "step": 11253
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961556599745614e-05,
+      "loss": 0.5319,
+      "step": 11254
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961542206661362e-05,
+      "loss": 0.523,
+      "step": 11255
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996152781088844e-05,
+      "loss": 0.5131,
+      "step": 11256
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996151341242686e-05,
+      "loss": 0.4612,
+      "step": 11257
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961499011276617e-05,
+      "loss": 0.5208,
+      "step": 11258
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996148460743772e-05,
+      "loss": 0.5059,
+      "step": 11259
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996147020091017e-05,
+      "loss": 0.4571,
+      "step": 11260
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961455791693976e-05,
+      "loss": 0.4481,
+      "step": 11261
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996144137978914e-05,
+      "loss": 0.4867,
+      "step": 11262
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996142696519566e-05,
+      "loss": 0.4911,
+      "step": 11263
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961412547913546e-05,
+      "loss": 0.5933,
+      "step": 11264
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961398127942798e-05,
+      "loss": 0.5072,
+      "step": 11265
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961383705283425e-05,
+      "loss": 0.4034,
+      "step": 11266
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996136927993543e-05,
+      "loss": 0.4968,
+      "step": 11267
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961354851898812e-05,
+      "loss": 0.5136,
+      "step": 11268
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961340421173575e-05,
+      "loss": 0.4633,
+      "step": 11269
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996132598775973e-05,
+      "loss": 0.5039,
+      "step": 11270
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961311551657273e-05,
+      "loss": 0.5043,
+      "step": 11271
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961297112866215e-05,
+      "loss": 0.4735,
+      "step": 11272
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961282671386556e-05,
+      "loss": 0.4455,
+      "step": 11273
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961268227218296e-05,
+      "loss": 0.4219,
+      "step": 11274
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996125378036145e-05,
+      "loss": 0.4465,
+      "step": 11275
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996123933081601e-05,
+      "loss": 0.5063,
+      "step": 11276
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961224878581987e-05,
+      "loss": 0.4569,
+      "step": 11277
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996121042365938e-05,
+      "loss": 0.4899,
+      "step": 11278
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961195966048196e-05,
+      "loss": 0.4381,
+      "step": 11279
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996118150574844e-05,
+      "loss": 0.4129,
+      "step": 11280
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961167042760118e-05,
+      "loss": 0.4482,
+      "step": 11281
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961152577083225e-05,
+      "loss": 0.4832,
+      "step": 11282
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.996113810871777e-05,
+      "loss": 0.4704,
+      "step": 11283
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961123637663764e-05,
+      "loss": 0.4891,
+      "step": 11284
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.99611091639212e-05,
+      "loss": 0.4299,
+      "step": 11285
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9961094687490085e-05,
+      "loss": 0.4199,
+      "step": 11286
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9961080208370423e-05,
+      "loss": 0.4818,
+      "step": 11287
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9961065726562225e-05,
+      "loss": 0.4859,
+      "step": 11288
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9961051242065483e-05,
+      "loss": 0.4499,
+      "step": 11289
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996103675488021e-05,
+      "loss": 0.4823,
+      "step": 11290
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9961022265006405e-05,
+      "loss": 0.5113,
+      "step": 11291
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9961007772444073e-05,
+      "loss": 0.4244,
+      "step": 11292
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996099327719322e-05,
+      "loss": 0.442,
+      "step": 11293
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996097877925385e-05,
+      "loss": 0.5394,
+      "step": 11294
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996096427862596e-05,
+      "loss": 0.4428,
+      "step": 11295
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960949775309563e-05,
+      "loss": 0.4198,
+      "step": 11296
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996093526930466e-05,
+      "loss": 0.4401,
+      "step": 11297
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960920760611248e-05,
+      "loss": 0.3964,
+      "step": 11298
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996090624922934e-05,
+      "loss": 0.4497,
+      "step": 11299
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996089173515894e-05,
+      "loss": 0.4371,
+      "step": 11300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960877218400047e-05,
+      "loss": 0.5513,
+      "step": 11301
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960862698952665e-05,
+      "loss": 0.5442,
+      "step": 11302
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.99608481768168e-05,
+      "loss": 0.4801,
+      "step": 11303
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960833651992458e-05,
+      "loss": 0.5317,
+      "step": 11304
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996081912447964e-05,
+      "loss": 0.513,
+      "step": 11305
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960804594278348e-05,
+      "loss": 0.5474,
+      "step": 11306
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960790061388587e-05,
+      "loss": 0.4893,
+      "step": 11307
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960775525810368e-05,
+      "loss": 0.5229,
+      "step": 11308
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960760987543683e-05,
+      "loss": 0.4839,
+      "step": 11309
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960746446588546e-05,
+      "loss": 0.3931,
+      "step": 11310
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960731902944957e-05,
+      "loss": 0.4489,
+      "step": 11311
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996071735661292e-05,
+      "loss": 0.4411,
+      "step": 11312
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960702807592436e-05,
+      "loss": 0.5084,
+      "step": 11313
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960688255883514e-05,
+      "loss": 0.3814,
+      "step": 11314
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960673701486154e-05,
+      "loss": 0.4451,
+      "step": 11315
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960659144400362e-05,
+      "loss": 0.481,
+      "step": 11316
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960644584626142e-05,
+      "loss": 0.4967,
+      "step": 11317
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.99606300221635e-05,
+      "loss": 0.4579,
+      "step": 11318
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996061545701243e-05,
+      "loss": 0.484,
+      "step": 11319
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960600889172948e-05,
+      "loss": 0.5229,
+      "step": 11320
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960586318645055e-05,
+      "loss": 0.4511,
+      "step": 11321
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960571745428754e-05,
+      "loss": 0.4732,
+      "step": 11322
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960557169524045e-05,
+      "loss": 0.4723,
+      "step": 11323
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960542590930934e-05,
+      "loss": 0.4761,
+      "step": 11324
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960528009649433e-05,
+      "loss": 0.4609,
+      "step": 11325
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996051342567953e-05,
+      "loss": 0.4382,
+      "step": 11326
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960498839021243e-05,
+      "loss": 0.5215,
+      "step": 11327
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960484249674568e-05,
+      "loss": 0.5471,
+      "step": 11328
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960469657639513e-05,
+      "loss": 0.4729,
+      "step": 11329
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960455062916083e-05,
+      "loss": 0.4617,
+      "step": 11330
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960440465504275e-05,
+      "loss": 0.4855,
+      "step": 11331
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960425865404104e-05,
+      "loss": 0.4664,
+      "step": 11332
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996041126261556e-05,
+      "loss": 0.4558,
+      "step": 11333
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996039665713866e-05,
+      "loss": 0.4697,
+      "step": 11334
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960382048973402e-05,
+      "loss": 0.4556,
+      "step": 11335
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960367438119785e-05,
+      "loss": 0.4213,
+      "step": 11336
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996035282457782e-05,
+      "loss": 0.4136,
+      "step": 11337
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960338208347516e-05,
+      "loss": 0.5069,
+      "step": 11338
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996032358942886e-05,
+      "loss": 0.4306,
+      "step": 11339
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960308967821876e-05,
+      "loss": 0.4624,
+      "step": 11340
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960294343526554e-05,
+      "loss": 0.4528,
+      "step": 11341
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960279716542902e-05,
+      "loss": 0.5658,
+      "step": 11342
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960265086870923e-05,
+      "loss": 0.4118,
+      "step": 11343
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996025045451062e-05,
+      "loss": 0.4313,
+      "step": 11344
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960235819462003e-05,
+      "loss": 0.4696,
+      "step": 11345
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996022118172507e-05,
+      "loss": 0.4943,
+      "step": 11346
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960206541299826e-05,
+      "loss": 0.4134,
+      "step": 11347
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960191898186278e-05,
+      "loss": 0.5348,
+      "step": 11348
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960177252384427e-05,
+      "loss": 0.5606,
+      "step": 11349
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960162603894276e-05,
+      "loss": 0.4608,
+      "step": 11350
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996014795271583e-05,
+      "loss": 0.4642,
+      "step": 11351
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960133298849093e-05,
+      "loss": 0.4525,
+      "step": 11352
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960118642294074e-05,
+      "loss": 0.4978,
+      "step": 11353
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960103983050766e-05,
+      "loss": 0.4355,
+      "step": 11354
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960089321119185e-05,
+      "loss": 0.4505,
+      "step": 11355
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960074656499328e-05,
+      "loss": 0.45,
+      "step": 11356
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.99600599891912e-05,
+      "loss": 0.4592,
+      "step": 11357
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960045319194803e-05,
+      "loss": 0.4397,
+      "step": 11358
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960030646510146e-05,
+      "loss": 0.5401,
+      "step": 11359
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.996001597113723e-05,
+      "loss": 0.5632,
+      "step": 11360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9960001293076057e-05,
+      "loss": 0.5219,
+      "step": 11361
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959986612326637e-05,
+      "loss": 0.5507,
+      "step": 11362
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959971928888967e-05,
+      "loss": 0.4339,
+      "step": 11363
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959957242763056e-05,
+      "loss": 0.4612,
+      "step": 11364
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959942553948902e-05,
+      "loss": 0.4414,
+      "step": 11365
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959927862446517e-05,
+      "loss": 0.4663,
+      "step": 11366
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.99599131682559e-05,
+      "loss": 0.454,
+      "step": 11367
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959898471377053e-05,
+      "loss": 0.5268,
+      "step": 11368
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959883771809986e-05,
+      "loss": 0.3943,
+      "step": 11369
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959869069554697e-05,
+      "loss": 0.4702,
+      "step": 11370
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959854364611196e-05,
+      "loss": 0.4362,
+      "step": 11371
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959839656979484e-05,
+      "loss": 0.4964,
+      "step": 11372
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.995982494665956e-05,
+      "loss": 0.551,
+      "step": 11373
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.995981023365144e-05,
+      "loss": 0.4813,
+      "step": 11374
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959795517955112e-05,
+      "loss": 0.583,
+      "step": 11375
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959780799570595e-05,
+      "loss": 0.4869,
+      "step": 11376
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959766078497886e-05,
+      "loss": 0.4916,
+      "step": 11377
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959751354736986e-05,
+      "loss": 0.5022,
+      "step": 11378
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959736628287908e-05,
+      "loss": 0.47,
+      "step": 11379
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959721899150646e-05,
+      "loss": 0.4743,
+      "step": 11380
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.995970716732521e-05,
+      "loss": 0.4337,
+      "step": 11381
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959692432811603e-05,
+      "loss": 0.4948,
+      "step": 11382
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9959677695609828e-05,
+      "loss": 0.5062,
+      "step": 11383
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.995966295571989e-05,
+      "loss": 0.4406,
+      "step": 11384
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995964821314179e-05,
+      "loss": 0.4371,
+      "step": 11385
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959633467875536e-05,
+      "loss": 0.5375,
+      "step": 11386
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959618719921132e-05,
+      "loss": 0.4413,
+      "step": 11387
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959603969278575e-05,
+      "loss": 0.4243,
+      "step": 11388
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995958921594788e-05,
+      "loss": 0.4689,
+      "step": 11389
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959574459929044e-05,
+      "loss": 0.5785,
+      "step": 11390
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959559701222072e-05,
+      "loss": 0.5017,
+      "step": 11391
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995954493982697e-05,
+      "loss": 0.5449,
+      "step": 11392
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959530175743737e-05,
+      "loss": 0.4879,
+      "step": 11393
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995951540897238e-05,
+      "loss": 0.5797,
+      "step": 11394
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959500639512905e-05,
+      "loss": 0.4112,
+      "step": 11395
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959485867365312e-05,
+      "loss": 0.4547,
+      "step": 11396
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959471092529613e-05,
+      "loss": 0.4223,
+      "step": 11397
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959456315005804e-05,
+      "loss": 0.5058,
+      "step": 11398
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959441534793885e-05,
+      "loss": 0.4407,
+      "step": 11399
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959426751893873e-05,
+      "loss": 0.5635,
+      "step": 11400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995941196630576e-05,
+      "loss": 0.4729,
+      "step": 11401
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959397178029564e-05,
+      "loss": 0.4583,
+      "step": 11402
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959382387065273e-05,
+      "loss": 0.55,
+      "step": 11403
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959367593412896e-05,
+      "loss": 0.5758,
+      "step": 11404
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959352797072443e-05,
+      "loss": 0.4622,
+      "step": 11405
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959337998043914e-05,
+      "loss": 0.4522,
+      "step": 11406
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959323196327313e-05,
+      "loss": 0.5117,
+      "step": 11407
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959308391922645e-05,
+      "loss": 0.6061,
+      "step": 11408
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959293584829912e-05,
+      "loss": 0.4851,
+      "step": 11409
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995927877504912e-05,
+      "loss": 0.4297,
+      "step": 11410
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995926396258027e-05,
+      "loss": 0.4589,
+      "step": 11411
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959249147423368e-05,
+      "loss": 0.4765,
+      "step": 11412
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959234329578422e-05,
+      "loss": 0.4175,
+      "step": 11413
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959219509045428e-05,
+      "loss": 0.4631,
+      "step": 11414
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959204685824395e-05,
+      "loss": 0.451,
+      "step": 11415
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959189859915326e-05,
+      "loss": 0.5243,
+      "step": 11416
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995917503131823e-05,
+      "loss": 0.4186,
+      "step": 11417
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.99591602000331e-05,
+      "loss": 0.5103,
+      "step": 11418
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995914536605995e-05,
+      "loss": 0.4695,
+      "step": 11419
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959130529398776e-05,
+      "loss": 0.4626,
+      "step": 11420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959115690049588e-05,
+      "loss": 0.4024,
+      "step": 11421
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995910084801239e-05,
+      "loss": 0.4476,
+      "step": 11422
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995908600328718e-05,
+      "loss": 0.4378,
+      "step": 11423
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995907115587397e-05,
+      "loss": 0.4766,
+      "step": 11424
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959056305772758e-05,
+      "loss": 0.4586,
+      "step": 11425
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959041452983552e-05,
+      "loss": 0.4856,
+      "step": 11426
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959026597506353e-05,
+      "loss": 0.447,
+      "step": 11427
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9959011739341165e-05,
+      "loss": 0.5006,
+      "step": 11428
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958996878487996e-05,
+      "loss": 0.4419,
+      "step": 11429
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958982014946842e-05,
+      "loss": 0.5176,
+      "step": 11430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958967148717718e-05,
+      "loss": 0.6327,
+      "step": 11431
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958952279800622e-05,
+      "loss": 0.4567,
+      "step": 11432
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958937408195553e-05,
+      "loss": 0.4096,
+      "step": 11433
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958922533902525e-05,
+      "loss": 0.5389,
+      "step": 11434
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958907656921534e-05,
+      "loss": 0.5175,
+      "step": 11435
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995889277725259e-05,
+      "loss": 0.4731,
+      "step": 11436
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958877894895693e-05,
+      "loss": 0.5261,
+      "step": 11437
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995886300985085e-05,
+      "loss": 0.532,
+      "step": 11438
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958848122118062e-05,
+      "loss": 0.5009,
+      "step": 11439
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958833231697334e-05,
+      "loss": 0.4067,
+      "step": 11440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995881833858867e-05,
+      "loss": 0.4746,
+      "step": 11441
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958803442792075e-05,
+      "loss": 0.431,
+      "step": 11442
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958788544307556e-05,
+      "loss": 0.4682,
+      "step": 11443
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958773643135107e-05,
+      "loss": 0.5234,
+      "step": 11444
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995875873927474e-05,
+      "loss": 0.5417,
+      "step": 11445
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995874383272646e-05,
+      "loss": 0.4256,
+      "step": 11446
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958728923490267e-05,
+      "loss": 0.4498,
+      "step": 11447
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958714011566168e-05,
+      "loss": 0.4802,
+      "step": 11448
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958699096954162e-05,
+      "loss": 0.4716,
+      "step": 11449
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995868417965426e-05,
+      "loss": 0.4618,
+      "step": 11450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958669259666462e-05,
+      "loss": 0.5313,
+      "step": 11451
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995865433699077e-05,
+      "loss": 0.4335,
+      "step": 11452
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958639411627193e-05,
+      "loss": 0.5756,
+      "step": 11453
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958624483575732e-05,
+      "loss": 0.4757,
+      "step": 11454
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958609552836396e-05,
+      "loss": 0.5484,
+      "step": 11455
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995859461940918e-05,
+      "loss": 0.4611,
+      "step": 11456
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958579683294094e-05,
+      "loss": 0.5165,
+      "step": 11457
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995856474449114e-05,
+      "loss": 0.4029,
+      "step": 11458
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995854980300032e-05,
+      "loss": 0.4661,
+      "step": 11459
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958534858821645e-05,
+      "loss": 0.414,
+      "step": 11460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958519911955113e-05,
+      "loss": 0.4471,
+      "step": 11461
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958504962400732e-05,
+      "loss": 0.5426,
+      "step": 11462
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958490010158505e-05,
+      "loss": 0.4646,
+      "step": 11463
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958475055228432e-05,
+      "loss": 0.4861,
+      "step": 11464
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958460097610518e-05,
+      "loss": 0.4379,
+      "step": 11465
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958445137304774e-05,
+      "loss": 0.489,
+      "step": 11466
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958430174311195e-05,
+      "loss": 0.4143,
+      "step": 11467
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958415208629795e-05,
+      "loss": 0.4568,
+      "step": 11468
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958400240260565e-05,
+      "loss": 0.6318,
+      "step": 11469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995838526920352e-05,
+      "loss": 0.498,
+      "step": 11470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995837029545866e-05,
+      "loss": 0.5121,
+      "step": 11471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995835531902599e-05,
+      "loss": 0.4774,
+      "step": 11472
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958340339905514e-05,
+      "loss": 0.4578,
+      "step": 11473
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958325358097232e-05,
+      "loss": 0.4394,
+      "step": 11474
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958310373601153e-05,
+      "loss": 0.4532,
+      "step": 11475
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995829538641728e-05,
+      "loss": 0.4903,
+      "step": 11476
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958280396545617e-05,
+      "loss": 0.4536,
+      "step": 11477
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958265403986167e-05,
+      "loss": 0.5729,
+      "step": 11478
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958250408738932e-05,
+      "loss": 0.5115,
+      "step": 11479
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9958235410803922e-05,
+      "loss": 0.4639,
+      "step": 11480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995822041018114e-05,
+      "loss": 0.5355,
+      "step": 11481
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995820540687058e-05,
+      "loss": 0.481,
+      "step": 11482
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.995819040087226e-05,
+      "loss": 0.4367,
+      "step": 11483
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958175392186178e-05,
+      "loss": 0.4814,
+      "step": 11484
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958160380812334e-05,
+      "loss": 0.5104,
+      "step": 11485
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958145366750742e-05,
+      "loss": 0.4372,
+      "step": 11486
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958130350001393e-05,
+      "loss": 0.4933,
+      "step": 11487
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958115330564303e-05,
+      "loss": 0.4384,
+      "step": 11488
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958100308439466e-05,
+      "loss": 0.458,
+      "step": 11489
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958085283626895e-05,
+      "loss": 0.5405,
+      "step": 11490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995807025612659e-05,
+      "loss": 0.5011,
+      "step": 11491
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958055225938556e-05,
+      "loss": 0.4767,
+      "step": 11492
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958040193062794e-05,
+      "loss": 0.4372,
+      "step": 11493
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958025157499315e-05,
+      "loss": 0.3826,
+      "step": 11494
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9958010119248116e-05,
+      "loss": 0.477,
+      "step": 11495
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.99579950783092e-05,
+      "loss": 0.4885,
+      "step": 11496
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995798003468258e-05,
+      "loss": 0.5007,
+      "step": 11497
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957964988368254e-05,
+      "loss": 0.5349,
+      "step": 11498
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995794993936622e-05,
+      "loss": 0.4924,
+      "step": 11499
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957934887676496e-05,
+      "loss": 0.4684,
+      "step": 11500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957919833299077e-05,
+      "loss": 0.5231,
+      "step": 11501
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957904776233968e-05,
+      "loss": 0.4361,
+      "step": 11502
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957889716481173e-05,
+      "loss": 0.4687,
+      "step": 11503
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957874654040698e-05,
+      "loss": 0.5159,
+      "step": 11504
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957859588912547e-05,
+      "loss": 0.5164,
+      "step": 11505
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995784452109672e-05,
+      "loss": 0.4425,
+      "step": 11506
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995782945059323e-05,
+      "loss": 0.5232,
+      "step": 11507
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995781437740207e-05,
+      "loss": 0.4143,
+      "step": 11508
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957799301523253e-05,
+      "loss": 0.441,
+      "step": 11509
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957784222956778e-05,
+      "loss": 0.5053,
+      "step": 11510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957769141702648e-05,
+      "loss": 0.4615,
+      "step": 11511
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957754057760872e-05,
+      "loss": 0.4231,
+      "step": 11512
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995773897113145e-05,
+      "loss": 0.4337,
+      "step": 11513
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957723881814393e-05,
+      "loss": 0.4803,
+      "step": 11514
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957708789809694e-05,
+      "loss": 0.4709,
+      "step": 11515
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957693695117366e-05,
+      "loss": 0.4626,
+      "step": 11516
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957678597737405e-05,
+      "loss": 0.375,
+      "step": 11517
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957663497669826e-05,
+      "loss": 0.4943,
+      "step": 11518
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957648394914625e-05,
+      "loss": 0.4599,
+      "step": 11519
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957633289471806e-05,
+      "loss": 0.4539,
+      "step": 11520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957618181341378e-05,
+      "loss": 0.5006,
+      "step": 11521
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995760307052334e-05,
+      "loss": 0.4184,
+      "step": 11522
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957587957017703e-05,
+      "loss": 0.5142,
+      "step": 11523
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995757284082446e-05,
+      "loss": 0.4215,
+      "step": 11524
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995755772194363e-05,
+      "loss": 0.4682,
+      "step": 11525
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.99575426003752e-05,
+      "loss": 0.5038,
+      "step": 11526
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995752747611919e-05,
+      "loss": 0.4629,
+      "step": 11527
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995751234917559e-05,
+      "loss": 0.5167,
+      "step": 11528
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957497219544413e-05,
+      "loss": 0.4154,
+      "step": 11529
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995748208722566e-05,
+      "loss": 0.3999,
+      "step": 11530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995746695221934e-05,
+      "loss": 0.5002,
+      "step": 11531
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995745181452545e-05,
+      "loss": 0.4372,
+      "step": 11532
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957436674144e-05,
+      "loss": 0.5467,
+      "step": 11533
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957421531074988e-05,
+      "loss": 0.4724,
+      "step": 11534
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957406385318425e-05,
+      "loss": 0.545,
+      "step": 11535
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957391236874307e-05,
+      "loss": 0.4767,
+      "step": 11536
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957376085742646e-05,
+      "loss": 0.4096,
+      "step": 11537
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957360931923443e-05,
+      "loss": 0.5679,
+      "step": 11538
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.99573457754167e-05,
+      "loss": 0.4261,
+      "step": 11539
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995733061622242e-05,
+      "loss": 0.5068,
+      "step": 11540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957315454340613e-05,
+      "loss": 0.3996,
+      "step": 11541
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995730028977128e-05,
+      "loss": 0.5015,
+      "step": 11542
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957285122514424e-05,
+      "loss": 0.4707,
+      "step": 11543
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957269952570052e-05,
+      "loss": 0.5308,
+      "step": 11544
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957254779938167e-05,
+      "loss": 0.4994,
+      "step": 11545
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995723960461877e-05,
+      "loss": 0.4695,
+      "step": 11546
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995722442661187e-05,
+      "loss": 0.5078,
+      "step": 11547
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957209245917467e-05,
+      "loss": 0.4933,
+      "step": 11548
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957194062535564e-05,
+      "loss": 0.4483,
+      "step": 11549
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957178876466172e-05,
+      "loss": 0.5193,
+      "step": 11550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995716368770929e-05,
+      "loss": 0.4603,
+      "step": 11551
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957148496264922e-05,
+      "loss": 0.4944,
+      "step": 11552
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957133302133074e-05,
+      "loss": 0.4622,
+      "step": 11553
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957118105313746e-05,
+      "loss": 0.5012,
+      "step": 11554
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957102905806952e-05,
+      "loss": 0.4341,
+      "step": 11555
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995708770361268e-05,
+      "loss": 0.4709,
+      "step": 11556
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957072498730952e-05,
+      "loss": 0.5093,
+      "step": 11557
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995705729116176e-05,
+      "loss": 0.413,
+      "step": 11558
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995704208090511e-05,
+      "loss": 0.4776,
+      "step": 11559
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995702686796101e-05,
+      "loss": 0.4682,
+      "step": 11560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9957011652329464e-05,
+      "loss": 0.4422,
+      "step": 11561
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995699643401047e-05,
+      "loss": 0.4224,
+      "step": 11562
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995698121300404e-05,
+      "loss": 0.5622,
+      "step": 11563
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956965989310172e-05,
+      "loss": 0.4689,
+      "step": 11564
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956950762928873e-05,
+      "loss": 0.5139,
+      "step": 11565
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956935533860145e-05,
+      "loss": 0.4513,
+      "step": 11566
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956920302103994e-05,
+      "loss": 0.4618,
+      "step": 11567
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956905067660425e-05,
+      "loss": 0.5328,
+      "step": 11568
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995688983052944e-05,
+      "loss": 0.507,
+      "step": 11569
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995687459071104e-05,
+      "loss": 0.4727,
+      "step": 11570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995685934820524e-05,
+      "loss": 0.4329,
+      "step": 11571
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956844103012033e-05,
+      "loss": 0.4634,
+      "step": 11572
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995682885513143e-05,
+      "loss": 0.5204,
+      "step": 11573
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995681360456343e-05,
+      "loss": 0.423,
+      "step": 11574
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.995679835130804e-05,
+      "loss": 0.5143,
+      "step": 11575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956783095365263e-05,
+      "loss": 0.4503,
+      "step": 11576
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.99567678367351e-05,
+      "loss": 0.4186,
+      "step": 11577
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956752575417566e-05,
+      "loss": 0.4803,
+      "step": 11578
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956737311412655e-05,
+      "loss": 0.5276,
+      "step": 11579
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956722044720377e-05,
+      "loss": 0.4474,
+      "step": 11580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9956706775340727e-05,
+      "loss": 0.5545,
+      "step": 11581
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995669150327372e-05,
+      "loss": 0.5185,
+      "step": 11582
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995667622851935e-05,
+      "loss": 0.4871,
+      "step": 11583
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995666095107763e-05,
+      "loss": 0.4084,
+      "step": 11584
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956645670948565e-05,
+      "loss": 0.4068,
+      "step": 11585
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995663038813215e-05,
+      "loss": 0.4408,
+      "step": 11586
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956615102628395e-05,
+      "loss": 0.5251,
+      "step": 11587
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956599814437302e-05,
+      "loss": 0.5078,
+      "step": 11588
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995658452355888e-05,
+      "loss": 0.4736,
+      "step": 11589
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956569229993125e-05,
+      "loss": 0.5714,
+      "step": 11590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956553933740047e-05,
+      "loss": 0.4601,
+      "step": 11591
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995653863479965e-05,
+      "loss": 0.4313,
+      "step": 11592
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956523333171933e-05,
+      "loss": 0.6231,
+      "step": 11593
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956508028856904e-05,
+      "loss": 0.4533,
+      "step": 11594
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995649272185457e-05,
+      "loss": 0.487,
+      "step": 11595
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956477412164933e-05,
+      "loss": 0.4323,
+      "step": 11596
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956462099787993e-05,
+      "loss": 0.4894,
+      "step": 11597
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956446784723757e-05,
+      "loss": 0.5052,
+      "step": 11598
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956431466972234e-05,
+      "loss": 0.4229,
+      "step": 11599
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956416146533418e-05,
+      "loss": 0.4666,
+      "step": 11600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995640082340732e-05,
+      "loss": 0.4866,
+      "step": 11601
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956385497593947e-05,
+      "loss": 0.4814,
+      "step": 11602
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956370169093294e-05,
+      "loss": 0.4917,
+      "step": 11603
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956354837905373e-05,
+      "loss": 0.4452,
+      "step": 11604
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956339504030185e-05,
+      "loss": 0.5243,
+      "step": 11605
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956324167467734e-05,
+      "loss": 0.4511,
+      "step": 11606
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956308828218024e-05,
+      "loss": 0.4712,
+      "step": 11607
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956293486281058e-05,
+      "loss": 0.4914,
+      "step": 11608
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956278141656845e-05,
+      "loss": 0.4758,
+      "step": 11609
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956262794345384e-05,
+      "loss": 0.5078,
+      "step": 11610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956247444346683e-05,
+      "loss": 0.4384,
+      "step": 11611
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956232091660747e-05,
+      "loss": 0.4729,
+      "step": 11612
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995621673628757e-05,
+      "loss": 0.4672,
+      "step": 11613
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956201378227167e-05,
+      "loss": 0.4521,
+      "step": 11614
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956186017479544e-05,
+      "loss": 0.4811,
+      "step": 11615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995617065404469e-05,
+      "loss": 0.5591,
+      "step": 11616
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956155287922624e-05,
+      "loss": 0.4077,
+      "step": 11617
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956139919113347e-05,
+      "loss": 0.4238,
+      "step": 11618
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995612454761686e-05,
+      "loss": 0.4456,
+      "step": 11619
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956109173433165e-05,
+      "loss": 0.4596,
+      "step": 11620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956093796562273e-05,
+      "loss": 0.5845,
+      "step": 11621
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956078417004186e-05,
+      "loss": 0.5329,
+      "step": 11622
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956063034758904e-05,
+      "loss": 0.4319,
+      "step": 11623
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956047649826437e-05,
+      "loss": 0.4565,
+      "step": 11624
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956032262206785e-05,
+      "loss": 0.508,
+      "step": 11625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995601687189995e-05,
+      "loss": 0.4988,
+      "step": 11626
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9956001478905943e-05,
+      "loss": 0.5321,
+      "step": 11627
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955986083224763e-05,
+      "loss": 0.4325,
+      "step": 11628
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995597068485642e-05,
+      "loss": 0.4906,
+      "step": 11629
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955955283800907e-05,
+      "loss": 0.43,
+      "step": 11630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995593988005824e-05,
+      "loss": 0.4683,
+      "step": 11631
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995592447362842e-05,
+      "loss": 0.4014,
+      "step": 11632
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955909064511447e-05,
+      "loss": 0.5278,
+      "step": 11633
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955893652707323e-05,
+      "loss": 0.509,
+      "step": 11634
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955878238216063e-05,
+      "loss": 0.5442,
+      "step": 11635
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995586282103766e-05,
+      "loss": 0.4855,
+      "step": 11636
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955847401172125e-05,
+      "loss": 0.4089,
+      "step": 11637
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955831978619466e-05,
+      "loss": 0.4386,
+      "step": 11638
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955816553379676e-05,
+      "loss": 0.512,
+      "step": 11639
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995580112545276e-05,
+      "loss": 0.4521,
+      "step": 11640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955785694838737e-05,
+      "loss": 0.4323,
+      "step": 11641
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995577026153759e-05,
+      "loss": 0.5059,
+      "step": 11642
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955754825549346e-05,
+      "loss": 0.5415,
+      "step": 11643
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995573938687399e-05,
+      "loss": 0.5178,
+      "step": 11644
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955723945511534e-05,
+      "loss": 0.4132,
+      "step": 11645
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995570850146198e-05,
+      "loss": 0.4853,
+      "step": 11646
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995569305472534e-05,
+      "loss": 0.4894,
+      "step": 11647
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955677605301606e-05,
+      "loss": 0.479,
+      "step": 11648
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995566215319079e-05,
+      "loss": 0.5044,
+      "step": 11649
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955646698392893e-05,
+      "loss": 0.5807,
+      "step": 11650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995563124090792e-05,
+      "loss": 0.4959,
+      "step": 11651
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995561578073588e-05,
+      "loss": 0.51,
+      "step": 11652
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995560031787677e-05,
+      "loss": 0.4273,
+      "step": 11653
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955584852330595e-05,
+      "loss": 0.5179,
+      "step": 11654
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955569384097364e-05,
+      "loss": 0.4234,
+      "step": 11655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955553913177076e-05,
+      "loss": 0.4061,
+      "step": 11656
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995553843956974e-05,
+      "loss": 0.5794,
+      "step": 11657
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955522963275357e-05,
+      "loss": 0.4505,
+      "step": 11658
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995550748429393e-05,
+      "loss": 0.509,
+      "step": 11659
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955492002625464e-05,
+      "loss": 0.51,
+      "step": 11660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955476518269967e-05,
+      "loss": 0.5426,
+      "step": 11661
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955461031227437e-05,
+      "loss": 0.4964,
+      "step": 11662
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955445541497885e-05,
+      "loss": 0.4797,
+      "step": 11663
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995543004908131e-05,
+      "loss": 0.4937,
+      "step": 11664
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955414553977722e-05,
+      "loss": 0.566,
+      "step": 11665
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955399056187117e-05,
+      "loss": 0.4084,
+      "step": 11666
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955383555709503e-05,
+      "loss": 0.4768,
+      "step": 11667
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955368052544887e-05,
+      "loss": 0.51,
+      "step": 11668
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995535254669327e-05,
+      "loss": 0.4619,
+      "step": 11669
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955337038154653e-05,
+      "loss": 0.4924,
+      "step": 11670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995532152692905e-05,
+      "loss": 0.5867,
+      "step": 11671
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955306013016455e-05,
+      "loss": 0.4858,
+      "step": 11672
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955290496416877e-05,
+      "loss": 0.4299,
+      "step": 11673
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955274977130317e-05,
+      "loss": 0.5398,
+      "step": 11674
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995525945515679e-05,
+      "loss": 0.4859,
+      "step": 11675
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955243930496283e-05,
+      "loss": 0.4806,
+      "step": 11676
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955228403148816e-05,
+      "loss": 0.4065,
+      "step": 11677
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995521287311438e-05,
+      "loss": 0.5051,
+      "step": 11678
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.995519734039299e-05,
+      "loss": 0.492,
+      "step": 11679
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9955181804984646e-05,
+      "loss": 0.4484,
+      "step": 11680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995516626688935e-05,
+      "loss": 0.409,
+      "step": 11681
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9955150726107107e-05,
+      "loss": 0.4537,
+      "step": 11682
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9955135182637922e-05,
+      "loss": 0.5183,
+      "step": 11683
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.99551196364818e-05,
+      "loss": 0.5078,
+      "step": 11684
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9955104087638747e-05,
+      "loss": 0.4454,
+      "step": 11685
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9955088536108764e-05,
+      "loss": 0.4276,
+      "step": 11686
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9955072981891856e-05,
+      "loss": 0.4413,
+      "step": 11687
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9955057424988028e-05,
+      "loss": 0.553,
+      "step": 11688
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9955041865397282e-05,
+      "loss": 0.4558,
+      "step": 11689
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9955026303119623e-05,
+      "loss": 0.4799,
+      "step": 11690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995501073815506e-05,
+      "loss": 0.5671,
+      "step": 11691
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995499517050359e-05,
+      "loss": 0.4544,
+      "step": 11692
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995497960016522e-05,
+      "loss": 0.4424,
+      "step": 11693
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954964027139955e-05,
+      "loss": 0.4522,
+      "step": 11694
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954948451427797e-05,
+      "loss": 0.519,
+      "step": 11695
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954932873028753e-05,
+      "loss": 0.467,
+      "step": 11696
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995491729194283e-05,
+      "loss": 0.5358,
+      "step": 11697
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954901708170024e-05,
+      "loss": 0.4741,
+      "step": 11698
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954886121710343e-05,
+      "loss": 0.5029,
+      "step": 11699
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954870532563792e-05,
+      "loss": 0.4638,
+      "step": 11700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995485494073038e-05,
+      "loss": 0.5762,
+      "step": 11701
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954839346210098e-05,
+      "loss": 0.4748,
+      "step": 11702
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995482374900296e-05,
+      "loss": 0.3685,
+      "step": 11703
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954808149108972e-05,
+      "loss": 0.5097,
+      "step": 11704
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954792546528133e-05,
+      "loss": 0.4501,
+      "step": 11705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954776941260448e-05,
+      "loss": 0.5569,
+      "step": 11706
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954761333305924e-05,
+      "loss": 0.4328,
+      "step": 11707
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954745722664567e-05,
+      "loss": 0.4299,
+      "step": 11708
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954730109336374e-05,
+      "loss": 0.5099,
+      "step": 11709
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995471449332135e-05,
+      "loss": 0.5237,
+      "step": 11710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954698874619508e-05,
+      "loss": 0.5227,
+      "step": 11711
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995468325323084e-05,
+      "loss": 0.4622,
+      "step": 11712
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995466762915536e-05,
+      "loss": 0.4637,
+      "step": 11713
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954652002393067e-05,
+      "loss": 0.5156,
+      "step": 11714
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954636372943967e-05,
+      "loss": 0.4426,
+      "step": 11715
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954620740808065e-05,
+      "loss": 0.4418,
+      "step": 11716
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954605105985364e-05,
+      "loss": 0.4345,
+      "step": 11717
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954589468475868e-05,
+      "loss": 0.5149,
+      "step": 11718
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954573828279583e-05,
+      "loss": 0.5017,
+      "step": 11719
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954558185396514e-05,
+      "loss": 0.5107,
+      "step": 11720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995454253982666e-05,
+      "loss": 0.4281,
+      "step": 11721
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995452689157003e-05,
+      "loss": 0.4357,
+      "step": 11722
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954511240626628e-05,
+      "loss": 0.4301,
+      "step": 11723
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995449558699645e-05,
+      "loss": 0.4901,
+      "step": 11724
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954479930679514e-05,
+      "loss": 0.4927,
+      "step": 11725
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954464271675814e-05,
+      "loss": 0.469,
+      "step": 11726
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954448609985357e-05,
+      "loss": 0.4864,
+      "step": 11727
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954432945608152e-05,
+      "loss": 0.4728,
+      "step": 11728
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954417278544196e-05,
+      "loss": 0.4734,
+      "step": 11729
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954401608793495e-05,
+      "loss": 0.4939,
+      "step": 11730
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954385936356057e-05,
+      "loss": 0.4663,
+      "step": 11731
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954370261231884e-05,
+      "loss": 0.5273,
+      "step": 11732
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954354583420977e-05,
+      "loss": 0.4366,
+      "step": 11733
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954338902923346e-05,
+      "loss": 0.4873,
+      "step": 11734
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954323219738994e-05,
+      "loss": 0.4835,
+      "step": 11735
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995430753386792e-05,
+      "loss": 0.4965,
+      "step": 11736
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995429184531013e-05,
+      "loss": 0.456,
+      "step": 11737
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954276154065634e-05,
+      "loss": 0.5235,
+      "step": 11738
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995426046013443e-05,
+      "loss": 0.5439,
+      "step": 11739
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995424476351653e-05,
+      "loss": 0.5056,
+      "step": 11740
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954229064211927e-05,
+      "loss": 0.5074,
+      "step": 11741
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954213362220635e-05,
+      "loss": 0.4876,
+      "step": 11742
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954197657542652e-05,
+      "loss": 0.4451,
+      "step": 11743
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954181950177983e-05,
+      "loss": 0.5124,
+      "step": 11744
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954166240126637e-05,
+      "loss": 0.4501,
+      "step": 11745
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995415052738861e-05,
+      "loss": 0.4724,
+      "step": 11746
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954134811963915e-05,
+      "loss": 0.4437,
+      "step": 11747
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954119093852553e-05,
+      "loss": 0.4329,
+      "step": 11748
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954103373054527e-05,
+      "loss": 0.4505,
+      "step": 11749
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954087649569842e-05,
+      "loss": 0.4834,
+      "step": 11750
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.99540719233985e-05,
+      "loss": 0.4472,
+      "step": 11751
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954056194540512e-05,
+      "loss": 0.5104,
+      "step": 11752
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954040462995875e-05,
+      "loss": 0.4217,
+      "step": 11753
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954024728764594e-05,
+      "loss": 0.599,
+      "step": 11754
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9954008991846678e-05,
+      "loss": 0.475,
+      "step": 11755
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995399325224213e-05,
+      "loss": 0.5027,
+      "step": 11756
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953977509950947e-05,
+      "loss": 0.4929,
+      "step": 11757
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953961764973143e-05,
+      "loss": 0.5177,
+      "step": 11758
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995394601730872e-05,
+      "loss": 0.3987,
+      "step": 11759
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953930266957677e-05,
+      "loss": 0.4954,
+      "step": 11760
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953914513920023e-05,
+      "loss": 0.4569,
+      "step": 11761
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995389875819576e-05,
+      "loss": 0.4979,
+      "step": 11762
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953882999784893e-05,
+      "loss": 0.5347,
+      "step": 11763
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995386723868743e-05,
+      "loss": 0.4797,
+      "step": 11764
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995385147490337e-05,
+      "loss": 0.4779,
+      "step": 11765
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953835708432713e-05,
+      "loss": 0.4983,
+      "step": 11766
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953819939275476e-05,
+      "loss": 0.5462,
+      "step": 11767
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953804167431654e-05,
+      "loss": 0.4362,
+      "step": 11768
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953788392901256e-05,
+      "loss": 0.457,
+      "step": 11769
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995377261568428e-05,
+      "loss": 0.5111,
+      "step": 11770
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995375683578074e-05,
+      "loss": 0.5439,
+      "step": 11771
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995374105319063e-05,
+      "loss": 0.5799,
+      "step": 11772
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995372526791396e-05,
+      "loss": 0.5093,
+      "step": 11773
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995370947995073e-05,
+      "loss": 0.5126,
+      "step": 11774
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995369368930095e-05,
+      "loss": 0.4554,
+      "step": 11775
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953677895964627e-05,
+      "loss": 0.4614,
+      "step": 11776
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9953662099941754e-05,
+      "loss": 0.4227,
+      "step": 11777
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995364630123234e-05,
+      "loss": 0.5067,
+      "step": 11778
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.995363049983639e-05,
+      "loss": 0.4108,
+      "step": 11779
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995361469575391e-05,
+      "loss": 0.5056,
+      "step": 11780
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953598888984906e-05,
+      "loss": 0.4748,
+      "step": 11781
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953583079529375e-05,
+      "loss": 0.4996,
+      "step": 11782
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953567267387327e-05,
+      "loss": 0.4902,
+      "step": 11783
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953551452558764e-05,
+      "loss": 0.4525,
+      "step": 11784
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995353563504369e-05,
+      "loss": 0.4445,
+      "step": 11785
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953519814842114e-05,
+      "loss": 0.4654,
+      "step": 11786
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953503991954033e-05,
+      "loss": 0.4368,
+      "step": 11787
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953488166379458e-05,
+      "loss": 0.4351,
+      "step": 11788
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953472338118386e-05,
+      "loss": 0.4399,
+      "step": 11789
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995345650717083e-05,
+      "loss": 0.5136,
+      "step": 11790
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953440673536785e-05,
+      "loss": 0.4672,
+      "step": 11791
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953424837216262e-05,
+      "loss": 0.4798,
+      "step": 11792
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953408998209263e-05,
+      "loss": 0.4597,
+      "step": 11793
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953393156515794e-05,
+      "loss": 0.4486,
+      "step": 11794
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953377312135854e-05,
+      "loss": 0.4796,
+      "step": 11795
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995336146506945e-05,
+      "loss": 0.481,
+      "step": 11796
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953345615316592e-05,
+      "loss": 0.4674,
+      "step": 11797
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953329762877276e-05,
+      "loss": 0.4792,
+      "step": 11798
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995331390775151e-05,
+      "loss": 0.423,
+      "step": 11799
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953298049939303e-05,
+      "loss": 0.4368,
+      "step": 11800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995328218944065e-05,
+      "loss": 0.5999,
+      "step": 11801
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995326632625556e-05,
+      "loss": 0.516,
+      "step": 11802
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953250460384037e-05,
+      "loss": 0.472,
+      "step": 11803
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953234591826085e-05,
+      "loss": 0.4577,
+      "step": 11804
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995321872058171e-05,
+      "loss": 0.4713,
+      "step": 11805
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995320284665091e-05,
+      "loss": 0.4508,
+      "step": 11806
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.99531869700337e-05,
+      "loss": 0.4446,
+      "step": 11807
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953171090730078e-05,
+      "loss": 0.4874,
+      "step": 11808
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953155208740042e-05,
+      "loss": 0.5028,
+      "step": 11809
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995313932406361e-05,
+      "loss": 0.4718,
+      "step": 11810
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953123436700775e-05,
+      "loss": 0.4604,
+      "step": 11811
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995310754665155e-05,
+      "loss": 0.4522,
+      "step": 11812
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995309165391593e-05,
+      "loss": 0.4688,
+      "step": 11813
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953075758493924e-05,
+      "loss": 0.3754,
+      "step": 11814
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995305986038554e-05,
+      "loss": 0.5851,
+      "step": 11815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9953043959590776e-05,
+      "loss": 0.4555,
+      "step": 11816
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995302805610964e-05,
+      "loss": 0.5296,
+      "step": 11817
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995301214994213e-05,
+      "loss": 0.4814,
+      "step": 11818
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952996241088263e-05,
+      "loss": 0.5291,
+      "step": 11819
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952980329548037e-05,
+      "loss": 0.4645,
+      "step": 11820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995296441532145e-05,
+      "loss": 0.5394,
+      "step": 11821
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952948498408513e-05,
+      "loss": 0.4945,
+      "step": 11822
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952932578809225e-05,
+      "loss": 0.5195,
+      "step": 11823
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952916656523596e-05,
+      "loss": 0.5,
+      "step": 11824
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952900731551633e-05,
+      "loss": 0.5065,
+      "step": 11825
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995288480389333e-05,
+      "loss": 0.5072,
+      "step": 11826
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.99528688735487e-05,
+      "loss": 0.4497,
+      "step": 11827
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952852940517743e-05,
+      "loss": 0.4276,
+      "step": 11828
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952837004800462e-05,
+      "loss": 0.5247,
+      "step": 11829
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952821066396867e-05,
+      "loss": 0.493,
+      "step": 11830
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995280512530696e-05,
+      "loss": 0.4146,
+      "step": 11831
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995278918153074e-05,
+      "loss": 0.5226,
+      "step": 11832
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995277323506822e-05,
+      "loss": 0.5566,
+      "step": 11833
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.99527572859194e-05,
+      "loss": 0.3765,
+      "step": 11834
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952741334084284e-05,
+      "loss": 0.5074,
+      "step": 11835
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995272537956287e-05,
+      "loss": 0.4355,
+      "step": 11836
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952709422355175e-05,
+      "loss": 0.5413,
+      "step": 11837
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.99526934624612e-05,
+      "loss": 0.4354,
+      "step": 11838
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952677499880942e-05,
+      "loss": 0.5766,
+      "step": 11839
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995266153461441e-05,
+      "loss": 0.4204,
+      "step": 11840
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995264556666161e-05,
+      "loss": 0.5348,
+      "step": 11841
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995262959602254e-05,
+      "loss": 0.5021,
+      "step": 11842
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995261362269721e-05,
+      "loss": 0.4014,
+      "step": 11843
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952597646685628e-05,
+      "loss": 0.5396,
+      "step": 11844
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995258166798779e-05,
+      "loss": 0.4676,
+      "step": 11845
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952565686603705e-05,
+      "loss": 0.5036,
+      "step": 11846
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952549702533375e-05,
+      "loss": 0.4087,
+      "step": 11847
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952533715776806e-05,
+      "loss": 0.4406,
+      "step": 11848
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952517726334e-05,
+      "loss": 0.5359,
+      "step": 11849
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952501734204962e-05,
+      "loss": 0.5441,
+      "step": 11850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.99524857393897e-05,
+      "loss": 0.3904,
+      "step": 11851
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952469741888218e-05,
+      "loss": 0.5875,
+      "step": 11852
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995245374170051e-05,
+      "loss": 0.5293,
+      "step": 11853
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952437738826596e-05,
+      "loss": 0.5224,
+      "step": 11854
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952421733266468e-05,
+      "loss": 0.4406,
+      "step": 11855
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952405725020137e-05,
+      "loss": 0.5449,
+      "step": 11856
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952389714087605e-05,
+      "loss": 0.4548,
+      "step": 11857
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952373700468873e-05,
+      "loss": 0.4965,
+      "step": 11858
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952357684163956e-05,
+      "loss": 0.4993,
+      "step": 11859
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952341665172847e-05,
+      "loss": 0.5089,
+      "step": 11860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952325643495555e-05,
+      "loss": 0.5229,
+      "step": 11861
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952309619132082e-05,
+      "loss": 0.4921,
+      "step": 11862
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952293592082433e-05,
+      "loss": 0.492,
+      "step": 11863
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952277562346616e-05,
+      "loss": 0.4465,
+      "step": 11864
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952261529924634e-05,
+      "loss": 0.477,
+      "step": 11865
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952245494816487e-05,
+      "loss": 0.4611,
+      "step": 11866
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952229457022184e-05,
+      "loss": 0.5139,
+      "step": 11867
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952213416541727e-05,
+      "loss": 0.3824,
+      "step": 11868
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952197373375122e-05,
+      "loss": 0.4932,
+      "step": 11869
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952181327522372e-05,
+      "loss": 0.4741,
+      "step": 11870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952165278983484e-05,
+      "loss": 0.5099,
+      "step": 11871
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952149227758458e-05,
+      "loss": 0.4492,
+      "step": 11872
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.99521331738473e-05,
+      "loss": 0.4653,
+      "step": 11873
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952117117250015e-05,
+      "loss": 0.438,
+      "step": 11874
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952101057966606e-05,
+      "loss": 0.5403,
+      "step": 11875
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9952084995997082e-05,
+      "loss": 0.4582,
+      "step": 11876
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.995206893134144e-05,
+      "loss": 0.4592,
+      "step": 11877
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995205286399969e-05,
+      "loss": 0.5361,
+      "step": 11878
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9952036793971832e-05,
+      "loss": 0.456,
+      "step": 11879
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9952020721257878e-05,
+      "loss": 0.483,
+      "step": 11880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995200464585782e-05,
+      "loss": 0.4391,
+      "step": 11881
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951988567771675e-05,
+      "loss": 0.5983,
+      "step": 11882
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995197248699944e-05,
+      "loss": 0.502,
+      "step": 11883
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995195640354112e-05,
+      "loss": 0.4491,
+      "step": 11884
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995194031739672e-05,
+      "loss": 0.5401,
+      "step": 11885
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951924228566247e-05,
+      "loss": 0.5297,
+      "step": 11886
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951908137049704e-05,
+      "loss": 0.4936,
+      "step": 11887
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951892042847093e-05,
+      "loss": 0.441,
+      "step": 11888
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995187594595842e-05,
+      "loss": 0.4552,
+      "step": 11889
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995185984638369e-05,
+      "loss": 0.5454,
+      "step": 11890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.99518437441229e-05,
+      "loss": 0.4463,
+      "step": 11891
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951827639176068e-05,
+      "loss": 0.4731,
+      "step": 11892
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995181153154319e-05,
+      "loss": 0.4306,
+      "step": 11893
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995179542122427e-05,
+      "loss": 0.4577,
+      "step": 11894
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951779308219318e-05,
+      "loss": 0.5277,
+      "step": 11895
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951763192528332e-05,
+      "loss": 0.4938,
+      "step": 11896
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951747074151317e-05,
+      "loss": 0.4792,
+      "step": 11897
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995173095308828e-05,
+      "loss": 0.4726,
+      "step": 11898
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995171482933922e-05,
+      "loss": 0.5236,
+      "step": 11899
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995169870290415e-05,
+      "loss": 0.4896,
+      "step": 11900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951682573783068e-05,
+      "loss": 0.4856,
+      "step": 11901
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951666441975982e-05,
+      "loss": 0.4969,
+      "step": 11902
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951650307482898e-05,
+      "loss": 0.572,
+      "step": 11903
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951634170303814e-05,
+      "loss": 0.4996,
+      "step": 11904
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951618030438735e-05,
+      "loss": 0.4988,
+      "step": 11905
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995160188788767e-05,
+      "loss": 0.4592,
+      "step": 11906
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995158574265062e-05,
+      "loss": 0.4618,
+      "step": 11907
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951569594727595e-05,
+      "loss": 0.5255,
+      "step": 11908
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951553444118587e-05,
+      "loss": 0.5262,
+      "step": 11909
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951537290823614e-05,
+      "loss": 0.4642,
+      "step": 11910
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951521134842676e-05,
+      "loss": 0.4443,
+      "step": 11911
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995150497617577e-05,
+      "loss": 0.4791,
+      "step": 11912
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995148881482291e-05,
+      "loss": 0.5226,
+      "step": 11913
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951472650784098e-05,
+      "loss": 0.4452,
+      "step": 11914
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951456484059333e-05,
+      "loss": 0.5371,
+      "step": 11915
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951440314648626e-05,
+      "loss": 0.4679,
+      "step": 11916
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951424142551978e-05,
+      "loss": 0.532,
+      "step": 11917
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951407967769395e-05,
+      "loss": 0.5224,
+      "step": 11918
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951391790300878e-05,
+      "loss": 0.4865,
+      "step": 11919
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951375610146436e-05,
+      "loss": 0.524,
+      "step": 11920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995135942730607e-05,
+      "loss": 0.4598,
+      "step": 11921
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951343241779786e-05,
+      "loss": 0.446,
+      "step": 11922
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951327053567587e-05,
+      "loss": 0.419,
+      "step": 11923
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995131086266948e-05,
+      "loss": 0.4767,
+      "step": 11924
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995129466908547e-05,
+      "loss": 0.4916,
+      "step": 11925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951278472815553e-05,
+      "loss": 0.5191,
+      "step": 11926
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951262273859747e-05,
+      "loss": 0.55,
+      "step": 11927
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995124607221804e-05,
+      "loss": 0.5663,
+      "step": 11928
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995122986789045e-05,
+      "loss": 0.3997,
+      "step": 11929
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951213660876978e-05,
+      "loss": 0.4739,
+      "step": 11930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951197451177627e-05,
+      "loss": 0.5157,
+      "step": 11931
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951181238792396e-05,
+      "loss": 0.4112,
+      "step": 11932
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.99511650237213e-05,
+      "loss": 0.5066,
+      "step": 11933
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951148805964335e-05,
+      "loss": 0.5095,
+      "step": 11934
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995113258552151e-05,
+      "loss": 0.526,
+      "step": 11935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951116362392828e-05,
+      "loss": 0.5099,
+      "step": 11936
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951100136578292e-05,
+      "loss": 0.5416,
+      "step": 11937
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995108390807791e-05,
+      "loss": 0.4613,
+      "step": 11938
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995106767689168e-05,
+      "loss": 0.4326,
+      "step": 11939
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951051443019613e-05,
+      "loss": 0.5159,
+      "step": 11940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995103520646171e-05,
+      "loss": 0.4241,
+      "step": 11941
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951018967217973e-05,
+      "loss": 0.5014,
+      "step": 11942
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9951002725288415e-05,
+      "loss": 0.5341,
+      "step": 11943
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995098648067303e-05,
+      "loss": 0.4825,
+      "step": 11944
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950970233371835e-05,
+      "loss": 0.5239,
+      "step": 11945
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995095398338482e-05,
+      "loss": 0.4709,
+      "step": 11946
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950937730711993e-05,
+      "loss": 0.4479,
+      "step": 11947
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995092147535337e-05,
+      "loss": 0.4682,
+      "step": 11948
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995090521730894e-05,
+      "loss": 0.4736,
+      "step": 11949
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995088895657872e-05,
+      "loss": 0.4431,
+      "step": 11950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950872693162703e-05,
+      "loss": 0.4344,
+      "step": 11951
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950856427060902e-05,
+      "loss": 0.4496,
+      "step": 11952
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950840158273315e-05,
+      "loss": 0.4698,
+      "step": 11953
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950823886799953e-05,
+      "loss": 0.4041,
+      "step": 11954
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950807612640816e-05,
+      "loss": 0.442,
+      "step": 11955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950791335795913e-05,
+      "loss": 0.4462,
+      "step": 11956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950775056265238e-05,
+      "loss": 0.4426,
+      "step": 11957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950758774048808e-05,
+      "loss": 0.5053,
+      "step": 11958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950742489146616e-05,
+      "loss": 0.4638,
+      "step": 11959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995072620155868e-05,
+      "loss": 0.6255,
+      "step": 11960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995070991128499e-05,
+      "loss": 0.4905,
+      "step": 11961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995069361832556e-05,
+      "loss": 0.4132,
+      "step": 11962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995067732268039e-05,
+      "loss": 0.4996,
+      "step": 11963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950661024349485e-05,
+      "loss": 0.4077,
+      "step": 11964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950644723332854e-05,
+      "loss": 0.5213,
+      "step": 11965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950628419630495e-05,
+      "loss": 0.4937,
+      "step": 11966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995061211324241e-05,
+      "loss": 0.4574,
+      "step": 11967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950595804168616e-05,
+      "loss": 0.4565,
+      "step": 11968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950579492409107e-05,
+      "loss": 0.3928,
+      "step": 11969
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995056317796389e-05,
+      "loss": 0.4149,
+      "step": 11970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950546860832968e-05,
+      "loss": 0.4364,
+      "step": 11971
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995053054101635e-05,
+      "loss": 0.4128,
+      "step": 11972
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950514218514036e-05,
+      "loss": 0.5648,
+      "step": 11973
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.995049789332603e-05,
+      "loss": 0.4893,
+      "step": 11974
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950481565452334e-05,
+      "loss": 0.4401,
+      "step": 11975
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9950465234892966e-05,
+      "loss": 0.5094,
+      "step": 11976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950448901647913e-05,
+      "loss": 0.5232,
+      "step": 11977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950432565717194e-05,
+      "loss": 0.4606,
+      "step": 11978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.99504162271008e-05,
+      "loss": 0.4906,
+      "step": 11979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950399885798747e-05,
+      "loss": 0.4267,
+      "step": 11980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950383541811033e-05,
+      "loss": 0.4197,
+      "step": 11981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950367195137666e-05,
+      "loss": 0.4999,
+      "step": 11982
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950350845778645e-05,
+      "loss": 0.514,
+      "step": 11983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.995033449373398e-05,
+      "loss": 0.4567,
+      "step": 11984
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950318139003674e-05,
+      "loss": 0.4804,
+      "step": 11985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950301781587726e-05,
+      "loss": 0.4834,
+      "step": 11986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.995028542148615e-05,
+      "loss": 0.4456,
+      "step": 11987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950269058698942e-05,
+      "loss": 0.5206,
+      "step": 11988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950252693226113e-05,
+      "loss": 0.3991,
+      "step": 11989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950236325067663e-05,
+      "loss": 0.4225,
+      "step": 11990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950219954223597e-05,
+      "loss": 0.429,
+      "step": 11991
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950203580693922e-05,
+      "loss": 0.4804,
+      "step": 11992
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950187204478638e-05,
+      "loss": 0.467,
+      "step": 11993
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950170825577754e-05,
+      "loss": 0.4846,
+      "step": 11994
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.995015444399127e-05,
+      "loss": 0.5572,
+      "step": 11995
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.99501380597192e-05,
+      "loss": 0.4177,
+      "step": 11996
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.995012167276153e-05,
+      "loss": 0.5017,
+      "step": 11997
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950105283118283e-05,
+      "loss": 0.4275,
+      "step": 11998
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950088890789454e-05,
+      "loss": 0.4675,
+      "step": 11999
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.995007249577505e-05,
+      "loss": 0.5353,
+      "step": 12000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950056098075075e-05,
+      "loss": 0.5365,
+      "step": 12001
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950039697689533e-05,
+      "loss": 0.6099,
+      "step": 12002
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.995002329461843e-05,
+      "loss": 0.4313,
+      "step": 12003
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9950006888861765e-05,
+      "loss": 0.4355,
+      "step": 12004
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994999048041955e-05,
+      "loss": 0.444,
+      "step": 12005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949974069291787e-05,
+      "loss": 0.478,
+      "step": 12006
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994995765547848e-05,
+      "loss": 0.4383,
+      "step": 12007
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994994123897963e-05,
+      "loss": 0.492,
+      "step": 12008
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949924819795243e-05,
+      "loss": 0.4739,
+      "step": 12009
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994990839792533e-05,
+      "loss": 0.4606,
+      "step": 12010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949891973369886e-05,
+      "loss": 0.5459,
+      "step": 12011
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949875546128923e-05,
+      "loss": 0.3606,
+      "step": 12012
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994985911620244e-05,
+      "loss": 0.4463,
+      "step": 12013
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994984268359044e-05,
+      "loss": 0.5351,
+      "step": 12014
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949826248292937e-05,
+      "loss": 0.4938,
+      "step": 12015
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949809810309927e-05,
+      "loss": 0.5055,
+      "step": 12016
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949793369641418e-05,
+      "loss": 0.4517,
+      "step": 12017
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949776926287412e-05,
+      "loss": 0.4645,
+      "step": 12018
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949760480247912e-05,
+      "loss": 0.4769,
+      "step": 12019
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994974403152293e-05,
+      "loss": 0.4623,
+      "step": 12020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949727580112463e-05,
+      "loss": 0.3984,
+      "step": 12021
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949711126016517e-05,
+      "loss": 0.4579,
+      "step": 12022
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.99496946692351e-05,
+      "loss": 0.6108,
+      "step": 12023
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949678209768214e-05,
+      "loss": 0.5743,
+      "step": 12024
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949661747615863e-05,
+      "loss": 0.4468,
+      "step": 12025
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949645282778052e-05,
+      "loss": 0.4365,
+      "step": 12026
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994962881525478e-05,
+      "loss": 0.3951,
+      "step": 12027
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949612345046063e-05,
+      "loss": 0.4183,
+      "step": 12028
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949595872151898e-05,
+      "loss": 0.4306,
+      "step": 12029
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994957939657229e-05,
+      "loss": 0.4706,
+      "step": 12030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949562918307245e-05,
+      "loss": 0.4165,
+      "step": 12031
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949546437356763e-05,
+      "loss": 0.468,
+      "step": 12032
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949529953720855e-05,
+      "loss": 0.4953,
+      "step": 12033
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994951346739952e-05,
+      "loss": 0.5351,
+      "step": 12034
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949496978392767e-05,
+      "loss": 0.5232,
+      "step": 12035
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949480486700596e-05,
+      "loss": 0.5366,
+      "step": 12036
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949463992323016e-05,
+      "loss": 0.4224,
+      "step": 12037
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994944749526003e-05,
+      "loss": 0.4776,
+      "step": 12038
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949430995511637e-05,
+      "loss": 0.3752,
+      "step": 12039
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994941449307785e-05,
+      "loss": 0.5321,
+      "step": 12040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949397987958668e-05,
+      "loss": 0.6128,
+      "step": 12041
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949381480154098e-05,
+      "loss": 0.4238,
+      "step": 12042
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949364969664145e-05,
+      "loss": 0.4856,
+      "step": 12043
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949348456488807e-05,
+      "loss": 0.4435,
+      "step": 12044
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949331940628096e-05,
+      "loss": 0.5846,
+      "step": 12045
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949315422082013e-05,
+      "loss": 0.3751,
+      "step": 12046
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949298900850568e-05,
+      "loss": 0.6287,
+      "step": 12047
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949282376933757e-05,
+      "loss": 0.4722,
+      "step": 12048
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949265850331585e-05,
+      "loss": 0.5323,
+      "step": 12049
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949249321044064e-05,
+      "loss": 0.4806,
+      "step": 12050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949232789071194e-05,
+      "loss": 0.4796,
+      "step": 12051
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994921625441298e-05,
+      "loss": 0.5105,
+      "step": 12052
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949199717069422e-05,
+      "loss": 0.4591,
+      "step": 12053
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994918317704053e-05,
+      "loss": 0.4917,
+      "step": 12054
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949166634326307e-05,
+      "loss": 0.4509,
+      "step": 12055
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994915008892676e-05,
+      "loss": 0.5771,
+      "step": 12056
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949133540841886e-05,
+      "loss": 0.4331,
+      "step": 12057
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.99491169900717e-05,
+      "loss": 0.5404,
+      "step": 12058
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.99491004366162e-05,
+      "loss": 0.5756,
+      "step": 12059
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994908388047539e-05,
+      "loss": 0.5099,
+      "step": 12060
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949067321649274e-05,
+      "loss": 0.4415,
+      "step": 12061
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949050760137858e-05,
+      "loss": 0.4718,
+      "step": 12062
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994903419594115e-05,
+      "loss": 0.5075,
+      "step": 12063
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994901762905915e-05,
+      "loss": 0.5306,
+      "step": 12064
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9949001059491865e-05,
+      "loss": 0.4647,
+      "step": 12065
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9948984487239292e-05,
+      "loss": 0.4693,
+      "step": 12066
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994896791230145e-05,
+      "loss": 0.5309,
+      "step": 12067
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9948951334678327e-05,
+      "loss": 0.5942,
+      "step": 12068
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9948934754369943e-05,
+      "loss": 0.6557,
+      "step": 12069
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9948918171376287e-05,
+      "loss": 0.5238,
+      "step": 12070
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9948901585697378e-05,
+      "loss": 0.3794,
+      "step": 12071
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9948884997333212e-05,
+      "loss": 0.4371,
+      "step": 12072
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9948868406283796e-05,
+      "loss": 0.4558,
+      "step": 12073
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994885181254913e-05,
+      "loss": 0.5151,
+      "step": 12074
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994883521612923e-05,
+      "loss": 0.4769,
+      "step": 12075
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948818617024085e-05,
+      "loss": 0.4933,
+      "step": 12076
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994880201523371e-05,
+      "loss": 0.4502,
+      "step": 12077
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948785410758108e-05,
+      "loss": 0.4588,
+      "step": 12078
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994876880359728e-05,
+      "loss": 0.4562,
+      "step": 12079
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948752193751238e-05,
+      "loss": 0.5471,
+      "step": 12080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948735581219978e-05,
+      "loss": 0.4495,
+      "step": 12081
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994871896600351e-05,
+      "loss": 0.4494,
+      "step": 12082
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994870234810183e-05,
+      "loss": 0.4548,
+      "step": 12083
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948685727514955e-05,
+      "loss": 0.4407,
+      "step": 12084
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994866910424288e-05,
+      "loss": 0.5341,
+      "step": 12085
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948652478285613e-05,
+      "loss": 0.4871,
+      "step": 12086
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948635849643157e-05,
+      "loss": 0.4446,
+      "step": 12087
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994861921831552e-05,
+      "loss": 0.4246,
+      "step": 12088
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948602584302707e-05,
+      "loss": 0.4827,
+      "step": 12089
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948585947604716e-05,
+      "loss": 0.4841,
+      "step": 12090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948569308221553e-05,
+      "loss": 0.4192,
+      "step": 12091
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948552666153228e-05,
+      "loss": 0.5251,
+      "step": 12092
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994853602139974e-05,
+      "loss": 0.4736,
+      "step": 12093
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948519373961098e-05,
+      "loss": 0.5125,
+      "step": 12094
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948502723837303e-05,
+      "loss": 0.5322,
+      "step": 12095
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994848607102836e-05,
+      "loss": 0.5601,
+      "step": 12096
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948469415534275e-05,
+      "loss": 0.4052,
+      "step": 12097
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994845275735505e-05,
+      "loss": 0.4509,
+      "step": 12098
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948436096490695e-05,
+      "loss": 0.4966,
+      "step": 12099
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948419432941208e-05,
+      "loss": 0.42,
+      "step": 12100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948402766706593e-05,
+      "loss": 0.4696,
+      "step": 12101
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948386097786862e-05,
+      "loss": 0.4003,
+      "step": 12102
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948369426182014e-05,
+      "loss": 0.4986,
+      "step": 12103
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948352751892054e-05,
+      "loss": 0.4364,
+      "step": 12104
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948336074916986e-05,
+      "loss": 0.4435,
+      "step": 12105
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948319395256817e-05,
+      "loss": 0.5479,
+      "step": 12106
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948302712911554e-05,
+      "loss": 0.4718,
+      "step": 12107
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948286027881192e-05,
+      "loss": 0.4659,
+      "step": 12108
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948269340165743e-05,
+      "loss": 0.5502,
+      "step": 12109
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994825264976521e-05,
+      "loss": 0.4768,
+      "step": 12110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948235956679596e-05,
+      "loss": 0.4375,
+      "step": 12111
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948219260908908e-05,
+      "loss": 0.4713,
+      "step": 12112
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994820256245315e-05,
+      "loss": 0.4213,
+      "step": 12113
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948185861312322e-05,
+      "loss": 0.4399,
+      "step": 12114
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948169157486435e-05,
+      "loss": 0.5843,
+      "step": 12115
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948152450975488e-05,
+      "loss": 0.4186,
+      "step": 12116
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994813574177949e-05,
+      "loss": 0.4061,
+      "step": 12117
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948119029898442e-05,
+      "loss": 0.456,
+      "step": 12118
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994810231533235e-05,
+      "loss": 0.4883,
+      "step": 12119
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994808559808122e-05,
+      "loss": 0.4152,
+      "step": 12120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948068878145056e-05,
+      "loss": 0.4361,
+      "step": 12121
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948052155523863e-05,
+      "loss": 0.4071,
+      "step": 12122
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994803543021764e-05,
+      "loss": 0.482,
+      "step": 12123
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948018702226402e-05,
+      "loss": 0.5409,
+      "step": 12124
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9948001971550144e-05,
+      "loss": 0.5106,
+      "step": 12125
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947985238188872e-05,
+      "loss": 0.4845,
+      "step": 12126
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994796850214259e-05,
+      "loss": 0.4647,
+      "step": 12127
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947951763411312e-05,
+      "loss": 0.4467,
+      "step": 12128
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994793502199503e-05,
+      "loss": 0.5608,
+      "step": 12129
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947918277893758e-05,
+      "loss": 0.5131,
+      "step": 12130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994790153110749e-05,
+      "loss": 0.428,
+      "step": 12131
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947884781636243e-05,
+      "loss": 0.3901,
+      "step": 12132
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947868029480014e-05,
+      "loss": 0.5068,
+      "step": 12133
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947851274638807e-05,
+      "loss": 0.4768,
+      "step": 12134
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994783451711263e-05,
+      "loss": 0.4764,
+      "step": 12135
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947817756901485e-05,
+      "loss": 0.5063,
+      "step": 12136
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947800994005377e-05,
+      "loss": 0.4732,
+      "step": 12137
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947784228424313e-05,
+      "loss": 0.4623,
+      "step": 12138
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947767460158294e-05,
+      "loss": 0.4423,
+      "step": 12139
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947750689207327e-05,
+      "loss": 0.5076,
+      "step": 12140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947733915571413e-05,
+      "loss": 0.4315,
+      "step": 12141
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947717139250562e-05,
+      "loss": 0.4675,
+      "step": 12142
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947700360244775e-05,
+      "loss": 0.5124,
+      "step": 12143
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947683578554056e-05,
+      "loss": 0.5549,
+      "step": 12144
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947666794178412e-05,
+      "loss": 0.5148,
+      "step": 12145
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947650007117847e-05,
+      "loss": 0.4595,
+      "step": 12146
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947633217372363e-05,
+      "loss": 0.47,
+      "step": 12147
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947616424941968e-05,
+      "loss": 0.4374,
+      "step": 12148
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947599629826665e-05,
+      "loss": 0.48,
+      "step": 12149
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947582832026457e-05,
+      "loss": 0.4478,
+      "step": 12150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994756603154135e-05,
+      "loss": 0.5154,
+      "step": 12151
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947549228371347e-05,
+      "loss": 0.5009,
+      "step": 12152
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947532422516456e-05,
+      "loss": 0.4862,
+      "step": 12153
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947515613976676e-05,
+      "loss": 0.4242,
+      "step": 12154
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994749880275202e-05,
+      "loss": 0.5395,
+      "step": 12155
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947481988842488e-05,
+      "loss": 0.4603,
+      "step": 12156
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994746517224808e-05,
+      "loss": 0.6308,
+      "step": 12157
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947448352968807e-05,
+      "loss": 0.4667,
+      "step": 12158
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994743153100467e-05,
+      "loss": 0.4538,
+      "step": 12159
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947414706355675e-05,
+      "loss": 0.4413,
+      "step": 12160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947397879021827e-05,
+      "loss": 0.4529,
+      "step": 12161
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994738104900313e-05,
+      "loss": 0.483,
+      "step": 12162
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947364216299587e-05,
+      "loss": 0.484,
+      "step": 12163
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947347380911204e-05,
+      "loss": 0.4978,
+      "step": 12164
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947330542837985e-05,
+      "loss": 0.556,
+      "step": 12165
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947313702079935e-05,
+      "loss": 0.4649,
+      "step": 12166
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994729685863706e-05,
+      "loss": 0.46,
+      "step": 12167
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947280012509362e-05,
+      "loss": 0.4718,
+      "step": 12168
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.994726316369685e-05,
+      "loss": 0.5622,
+      "step": 12169
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947246312199522e-05,
+      "loss": 0.5549,
+      "step": 12170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947229458017386e-05,
+      "loss": 0.4524,
+      "step": 12171
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947212601150444e-05,
+      "loss": 0.4946,
+      "step": 12172
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9947195741598708e-05,
+      "loss": 0.4117,
+      "step": 12173
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9947178879362173e-05,
+      "loss": 0.6075,
+      "step": 12174
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9947162014440852e-05,
+      "loss": 0.4551,
+      "step": 12175
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994714514683474e-05,
+      "loss": 0.4319,
+      "step": 12176
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994712827654385e-05,
+      "loss": 0.4666,
+      "step": 12177
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9947111403568185e-05,
+      "loss": 0.523,
+      "step": 12178
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9947094527907746e-05,
+      "loss": 0.5767,
+      "step": 12179
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994707764956254e-05,
+      "loss": 0.4313,
+      "step": 12180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9947060768532574e-05,
+      "loss": 0.4552,
+      "step": 12181
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994704388481785e-05,
+      "loss": 0.5378,
+      "step": 12182
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9947026998418367e-05,
+      "loss": 0.4393,
+      "step": 12183
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9947010109334137e-05,
+      "loss": 0.4818,
+      "step": 12184
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946993217565164e-05,
+      "loss": 0.4609,
+      "step": 12185
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994697632311145e-05,
+      "loss": 0.3896,
+      "step": 12186
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946959425973e-05,
+      "loss": 0.5,
+      "step": 12187
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946942526149818e-05,
+      "loss": 0.5216,
+      "step": 12188
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946925623641914e-05,
+      "loss": 0.5177,
+      "step": 12189
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946908718449286e-05,
+      "loss": 0.4967,
+      "step": 12190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994689181057194e-05,
+      "loss": 0.494,
+      "step": 12191
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946874900009884e-05,
+      "loss": 0.5213,
+      "step": 12192
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946857986763117e-05,
+      "loss": 0.4359,
+      "step": 12193
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994684107083165e-05,
+      "loss": 0.4313,
+      "step": 12194
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994682415221548e-05,
+      "loss": 0.5321,
+      "step": 12195
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946807230914618e-05,
+      "loss": 0.5103,
+      "step": 12196
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946790306929066e-05,
+      "loss": 0.533,
+      "step": 12197
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994677338025883e-05,
+      "loss": 0.4508,
+      "step": 12198
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994675645090391e-05,
+      "loss": 0.4947,
+      "step": 12199
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946739518864316e-05,
+      "loss": 0.4604,
+      "step": 12200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994672258414005e-05,
+      "loss": 0.5398,
+      "step": 12201
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946705646731116e-05,
+      "loss": 0.4787,
+      "step": 12202
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994668870663752e-05,
+      "loss": 0.5217,
+      "step": 12203
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946671763859273e-05,
+      "loss": 0.5214,
+      "step": 12204
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946654818396363e-05,
+      "loss": 0.3984,
+      "step": 12205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994663787024881e-05,
+      "loss": 0.4535,
+      "step": 12206
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994662091941661e-05,
+      "loss": 0.4019,
+      "step": 12207
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946603965899772e-05,
+      "loss": 0.455,
+      "step": 12208
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.99465870096983e-05,
+      "loss": 0.4539,
+      "step": 12209
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946570050812196e-05,
+      "loss": 0.4357,
+      "step": 12210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946553089241468e-05,
+      "loss": 0.4935,
+      "step": 12211
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946536124986116e-05,
+      "loss": 0.4157,
+      "step": 12212
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994651915804615e-05,
+      "loss": 0.4665,
+      "step": 12213
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994650218842157e-05,
+      "loss": 0.4911,
+      "step": 12214
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946485216112384e-05,
+      "loss": 0.5207,
+      "step": 12215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946468241118596e-05,
+      "loss": 0.4776,
+      "step": 12216
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994645126344021e-05,
+      "loss": 0.4667,
+      "step": 12217
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946434283077227e-05,
+      "loss": 0.5138,
+      "step": 12218
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946417300029657e-05,
+      "loss": 0.5054,
+      "step": 12219
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946400314297504e-05,
+      "loss": 0.4574,
+      "step": 12220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994638332588077e-05,
+      "loss": 0.4224,
+      "step": 12221
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946366334779455e-05,
+      "loss": 0.5495,
+      "step": 12222
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946349340993577e-05,
+      "loss": 0.4637,
+      "step": 12223
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946332344523133e-05,
+      "loss": 0.411,
+      "step": 12224
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994631534536812e-05,
+      "loss": 0.5157,
+      "step": 12225
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946298343528557e-05,
+      "loss": 0.4903,
+      "step": 12226
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994628133900444e-05,
+      "loss": 0.5434,
+      "step": 12227
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994626433179577e-05,
+      "loss": 0.4475,
+      "step": 12228
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946247321902563e-05,
+      "loss": 0.5569,
+      "step": 12229
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946230309324815e-05,
+      "loss": 0.5126,
+      "step": 12230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946213294062535e-05,
+      "loss": 0.5663,
+      "step": 12231
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946196276115724e-05,
+      "loss": 0.5633,
+      "step": 12232
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946179255484386e-05,
+      "loss": 0.529,
+      "step": 12233
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946162232168527e-05,
+      "loss": 0.5391,
+      "step": 12234
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994614520616816e-05,
+      "loss": 0.4704,
+      "step": 12235
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946128177483274e-05,
+      "loss": 0.4947,
+      "step": 12236
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946111146113883e-05,
+      "loss": 0.5753,
+      "step": 12237
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994609411205999e-05,
+      "loss": 0.5002,
+      "step": 12238
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946077075321604e-05,
+      "loss": 0.4705,
+      "step": 12239
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946060035898723e-05,
+      "loss": 0.4448,
+      "step": 12240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946042993791353e-05,
+      "loss": 0.4827,
+      "step": 12241
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9946025948999502e-05,
+      "loss": 0.4598,
+      "step": 12242
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994600890152317e-05,
+      "loss": 0.5575,
+      "step": 12243
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945991851362364e-05,
+      "loss": 0.558,
+      "step": 12244
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945974798517086e-05,
+      "loss": 0.4254,
+      "step": 12245
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945957742987344e-05,
+      "loss": 0.4314,
+      "step": 12246
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945940684773145e-05,
+      "loss": 0.5373,
+      "step": 12247
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945923623874485e-05,
+      "loss": 0.4266,
+      "step": 12248
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945906560291378e-05,
+      "loss": 0.5288,
+      "step": 12249
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945889494023823e-05,
+      "loss": 0.451,
+      "step": 12250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945872425071828e-05,
+      "loss": 0.39,
+      "step": 12251
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945855353435392e-05,
+      "loss": 0.491,
+      "step": 12252
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945838279114522e-05,
+      "loss": 0.4574,
+      "step": 12253
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945821202109226e-05,
+      "loss": 0.5392,
+      "step": 12254
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945804122419506e-05,
+      "loss": 0.4513,
+      "step": 12255
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994578704004537e-05,
+      "loss": 0.4681,
+      "step": 12256
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945769954986815e-05,
+      "loss": 0.533,
+      "step": 12257
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994575286724385e-05,
+      "loss": 0.4615,
+      "step": 12258
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945735776816484e-05,
+      "loss": 0.4742,
+      "step": 12259
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945718683704714e-05,
+      "loss": 0.5059,
+      "step": 12260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945701587908554e-05,
+      "loss": 0.4474,
+      "step": 12261
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945684489427994e-05,
+      "loss": 0.4545,
+      "step": 12262
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994566738826305e-05,
+      "loss": 0.4276,
+      "step": 12263
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945650284413726e-05,
+      "loss": 0.4895,
+      "step": 12264
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945633177880027e-05,
+      "loss": 0.4982,
+      "step": 12265
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994561606866195e-05,
+      "loss": 0.5133,
+      "step": 12266
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945598956759505e-05,
+      "loss": 0.5085,
+      "step": 12267
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.99455818421727e-05,
+      "loss": 0.4605,
+      "step": 12268
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9945564724901532e-05,
+      "loss": 0.4121,
+      "step": 12269
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994554760494601e-05,
+      "loss": 0.4403,
+      "step": 12270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.994553048230614e-05,
+      "loss": 0.4954,
+      "step": 12271
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945513356981927e-05,
+      "loss": 0.4516,
+      "step": 12272
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994549622897337e-05,
+      "loss": 0.4073,
+      "step": 12273
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994547909828048e-05,
+      "loss": 0.4667,
+      "step": 12274
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945461964903258e-05,
+      "loss": 0.4773,
+      "step": 12275
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945444828841706e-05,
+      "loss": 0.4927,
+      "step": 12276
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945427690095832e-05,
+      "loss": 0.4707,
+      "step": 12277
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945410548665644e-05,
+      "loss": 0.4877,
+      "step": 12278
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994539340455114e-05,
+      "loss": 0.5746,
+      "step": 12279
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994537625775233e-05,
+      "loss": 0.4968,
+      "step": 12280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945359108269217e-05,
+      "loss": 0.4387,
+      "step": 12281
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945341956101804e-05,
+      "loss": 0.4298,
+      "step": 12282
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945324801250097e-05,
+      "loss": 0.487,
+      "step": 12283
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.99453076437141e-05,
+      "loss": 0.4982,
+      "step": 12284
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994529048349382e-05,
+      "loss": 0.4958,
+      "step": 12285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945273320589255e-05,
+      "loss": 0.4806,
+      "step": 12286
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994525615500042e-05,
+      "loss": 0.4439,
+      "step": 12287
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945238986727312e-05,
+      "loss": 0.4111,
+      "step": 12288
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945221815769934e-05,
+      "loss": 0.4342,
+      "step": 12289
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.99452046421283e-05,
+      "loss": 0.5244,
+      "step": 12290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945187465802403e-05,
+      "loss": 0.465,
+      "step": 12291
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945170286792255e-05,
+      "loss": 0.5394,
+      "step": 12292
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945153105097857e-05,
+      "loss": 0.4612,
+      "step": 12293
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945135920719222e-05,
+      "loss": 0.4435,
+      "step": 12294
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945118733656343e-05,
+      "loss": 0.4282,
+      "step": 12295
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945101543909233e-05,
+      "loss": 0.5937,
+      "step": 12296
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945084351477893e-05,
+      "loss": 0.5169,
+      "step": 12297
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9945067156362326e-05,
+      "loss": 0.4795,
+      "step": 12298
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994504995856254e-05,
+      "loss": 0.4763,
+      "step": 12299
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994503275807854e-05,
+      "loss": 0.4954,
+      "step": 12300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994501555491033e-05,
+      "loss": 0.4141,
+      "step": 12301
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994499834905791e-05,
+      "loss": 0.6093,
+      "step": 12302
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994498114052129e-05,
+      "loss": 0.4964,
+      "step": 12303
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944963929300475e-05,
+      "loss": 0.4871,
+      "step": 12304
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944946715395467e-05,
+      "loss": 0.4768,
+      "step": 12305
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994492949880627e-05,
+      "loss": 0.4694,
+      "step": 12306
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944912279532893e-05,
+      "loss": 0.507,
+      "step": 12307
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944895057575332e-05,
+      "loss": 0.4536,
+      "step": 12308
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944877832933603e-05,
+      "loss": 0.4329,
+      "step": 12309
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.99448606056077e-05,
+      "loss": 0.4634,
+      "step": 12310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944843375597638e-05,
+      "loss": 0.4843,
+      "step": 12311
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944826142903413e-05,
+      "loss": 0.5756,
+      "step": 12312
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944808907525033e-05,
+      "loss": 0.4974,
+      "step": 12313
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944791669462503e-05,
+      "loss": 0.5218,
+      "step": 12314
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994477442871583e-05,
+      "loss": 0.4717,
+      "step": 12315
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944757185285013e-05,
+      "loss": 0.5556,
+      "step": 12316
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994473993917006e-05,
+      "loss": 0.5195,
+      "step": 12317
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944722690370974e-05,
+      "loss": 0.4438,
+      "step": 12318
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944705438887763e-05,
+      "loss": 0.4654,
+      "step": 12319
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944688184720426e-05,
+      "loss": 0.4267,
+      "step": 12320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944670927868978e-05,
+      "loss": 0.4893,
+      "step": 12321
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994465366833341e-05,
+      "loss": 0.4879,
+      "step": 12322
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944636406113735e-05,
+      "loss": 0.5093,
+      "step": 12323
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944619141209958e-05,
+      "loss": 0.4376,
+      "step": 12324
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944601873622082e-05,
+      "loss": 0.4036,
+      "step": 12325
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994458460335011e-05,
+      "loss": 0.4624,
+      "step": 12326
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994456733039405e-05,
+      "loss": 0.4281,
+      "step": 12327
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944550054753902e-05,
+      "loss": 0.4647,
+      "step": 12328
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944532776429674e-05,
+      "loss": 0.5307,
+      "step": 12329
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994451549542137e-05,
+      "loss": 0.4631,
+      "step": 12330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944498211728997e-05,
+      "loss": 0.4814,
+      "step": 12331
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944480925352555e-05,
+      "loss": 0.4991,
+      "step": 12332
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944463636292052e-05,
+      "loss": 0.4551,
+      "step": 12333
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944446344547495e-05,
+      "loss": 0.507,
+      "step": 12334
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994442905011888e-05,
+      "loss": 0.5972,
+      "step": 12335
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994441175300622e-05,
+      "loss": 0.5394,
+      "step": 12336
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944394453209517e-05,
+      "loss": 0.4689,
+      "step": 12337
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944377150728776e-05,
+      "loss": 0.5164,
+      "step": 12338
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944359845564e-05,
+      "loss": 0.5283,
+      "step": 12339
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944342537715193e-05,
+      "loss": 0.4053,
+      "step": 12340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944325227182364e-05,
+      "loss": 0.4793,
+      "step": 12341
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944307913965515e-05,
+      "loss": 0.4839,
+      "step": 12342
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944290598064652e-05,
+      "loss": 0.4272,
+      "step": 12343
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944273279479772e-05,
+      "loss": 0.4454,
+      "step": 12344
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944255958210892e-05,
+      "loss": 0.5305,
+      "step": 12345
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944238634258012e-05,
+      "loss": 0.4509,
+      "step": 12346
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944221307621128e-05,
+      "loss": 0.4608,
+      "step": 12347
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994420397830026e-05,
+      "loss": 0.5462,
+      "step": 12348
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.99441866462954e-05,
+      "loss": 0.4181,
+      "step": 12349
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944169311606562e-05,
+      "loss": 0.5243,
+      "step": 12350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944151974233742e-05,
+      "loss": 0.5002,
+      "step": 12351
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944134634176953e-05,
+      "loss": 0.5206,
+      "step": 12352
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944117291436194e-05,
+      "loss": 0.5108,
+      "step": 12353
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994409994601147e-05,
+      "loss": 0.5367,
+      "step": 12354
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994408259790279e-05,
+      "loss": 0.4851,
+      "step": 12355
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994406524711015e-05,
+      "loss": 0.4958,
+      "step": 12356
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944047893633565e-05,
+      "loss": 0.5021,
+      "step": 12357
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944030537473033e-05,
+      "loss": 0.5137,
+      "step": 12358
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9944013178628562e-05,
+      "loss": 0.43,
+      "step": 12359
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9943995817100155e-05,
+      "loss": 0.5113,
+      "step": 12360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9943978452887816e-05,
+      "loss": 0.4672,
+      "step": 12361
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994396108599155e-05,
+      "loss": 0.5228,
+      "step": 12362
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9943943716411364e-05,
+      "loss": 0.5152,
+      "step": 12363
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9943926344147262e-05,
+      "loss": 0.4799,
+      "step": 12364
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9943908969199248e-05,
+      "loss": 0.4543,
+      "step": 12365
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9943891591567325e-05,
+      "loss": 0.4918,
+      "step": 12366
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9943874211251497e-05,
+      "loss": 0.4663,
+      "step": 12367
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9943856828251774e-05,
+      "loss": 0.4713,
+      "step": 12368
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9943839442568156e-05,
+      "loss": 0.473,
+      "step": 12369
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.994382205420065e-05,
+      "loss": 0.4727,
+      "step": 12370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943804663149258e-05,
+      "loss": 0.4375,
+      "step": 12371
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943787269413992e-05,
+      "loss": 0.475,
+      "step": 12372
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943769872994848e-05,
+      "loss": 0.5359,
+      "step": 12373
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943752473891833e-05,
+      "loss": 0.5364,
+      "step": 12374
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943735072104956e-05,
+      "loss": 0.4379,
+      "step": 12375
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943717667634215e-05,
+      "loss": 0.4615,
+      "step": 12376
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994370026047962e-05,
+      "loss": 0.5581,
+      "step": 12377
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994368285064117e-05,
+      "loss": 0.5213,
+      "step": 12378
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994366543811888e-05,
+      "loss": 0.5301,
+      "step": 12379
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943648022912744e-05,
+      "loss": 0.4837,
+      "step": 12380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994363060502277e-05,
+      "loss": 0.4351,
+      "step": 12381
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943613184448965e-05,
+      "loss": 0.5194,
+      "step": 12382
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943595761191335e-05,
+      "loss": 0.4246,
+      "step": 12383
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943578335249877e-05,
+      "loss": 0.5053,
+      "step": 12384
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943560906624606e-05,
+      "loss": 0.4464,
+      "step": 12385
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943543475315517e-05,
+      "loss": 0.4995,
+      "step": 12386
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943526041322622e-05,
+      "loss": 0.4924,
+      "step": 12387
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994350860464592e-05,
+      "loss": 0.5232,
+      "step": 12388
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943491165285424e-05,
+      "loss": 0.491,
+      "step": 12389
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943473723241128e-05,
+      "loss": 0.4901,
+      "step": 12390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943456278513045e-05,
+      "loss": 0.4813,
+      "step": 12391
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943438831101172e-05,
+      "loss": 0.5494,
+      "step": 12392
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943421381005523e-05,
+      "loss": 0.4006,
+      "step": 12393
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943403928226098e-05,
+      "loss": 0.5233,
+      "step": 12394
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.99433864727629e-05,
+      "loss": 0.4538,
+      "step": 12395
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943369014615938e-05,
+      "loss": 0.5639,
+      "step": 12396
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994335155378521e-05,
+      "loss": 0.543,
+      "step": 12397
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943334090270726e-05,
+      "loss": 0.5237,
+      "step": 12398
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994331662407249e-05,
+      "loss": 0.4512,
+      "step": 12399
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994329915519051e-05,
+      "loss": 0.4421,
+      "step": 12400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994328168362478e-05,
+      "loss": 0.4276,
+      "step": 12401
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994326420937532e-05,
+      "loss": 0.4914,
+      "step": 12402
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943246732442124e-05,
+      "loss": 0.4513,
+      "step": 12403
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943229252825194e-05,
+      "loss": 0.4316,
+      "step": 12404
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943211770524543e-05,
+      "loss": 0.4452,
+      "step": 12405
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943194285540176e-05,
+      "loss": 0.4721,
+      "step": 12406
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994317679787209e-05,
+      "loss": 0.5994,
+      "step": 12407
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943159307520296e-05,
+      "loss": 0.4355,
+      "step": 12408
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943141814484796e-05,
+      "loss": 0.4785,
+      "step": 12409
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943124318765598e-05,
+      "loss": 0.7009,
+      "step": 12410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943106820362704e-05,
+      "loss": 0.4695,
+      "step": 12411
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943089319276115e-05,
+      "loss": 0.5258,
+      "step": 12412
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943071815505842e-05,
+      "loss": 0.4953,
+      "step": 12413
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943054309051887e-05,
+      "loss": 0.413,
+      "step": 12414
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994303679991426e-05,
+      "loss": 0.5209,
+      "step": 12415
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943019288092955e-05,
+      "loss": 0.4601,
+      "step": 12416
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9943001773587986e-05,
+      "loss": 0.569,
+      "step": 12417
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994298425639935e-05,
+      "loss": 0.4879,
+      "step": 12418
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994296673652706e-05,
+      "loss": 0.5435,
+      "step": 12419
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942949213971113e-05,
+      "loss": 0.4946,
+      "step": 12420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942931688731522e-05,
+      "loss": 0.4683,
+      "step": 12421
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942914160808287e-05,
+      "loss": 0.4526,
+      "step": 12422
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994289663020141e-05,
+      "loss": 0.4775,
+      "step": 12423
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.99428790969109e-05,
+      "loss": 0.4607,
+      "step": 12424
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942861560936763e-05,
+      "loss": 0.4263,
+      "step": 12425
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942844022278995e-05,
+      "loss": 0.4936,
+      "step": 12426
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942826480937612e-05,
+      "loss": 0.4373,
+      "step": 12427
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942808936912613e-05,
+      "loss": 0.5286,
+      "step": 12428
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942791390204003e-05,
+      "loss": 0.4894,
+      "step": 12429
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942773840811783e-05,
+      "loss": 0.4156,
+      "step": 12430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994275628873597e-05,
+      "loss": 0.4273,
+      "step": 12431
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942738733976555e-05,
+      "loss": 0.5486,
+      "step": 12432
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942721176533547e-05,
+      "loss": 0.448,
+      "step": 12433
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942703616406956e-05,
+      "loss": 0.4545,
+      "step": 12434
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942686053596782e-05,
+      "loss": 0.4421,
+      "step": 12435
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994266848810303e-05,
+      "loss": 0.4675,
+      "step": 12436
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942650919925707e-05,
+      "loss": 0.5049,
+      "step": 12437
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942633349064813e-05,
+      "loss": 0.4484,
+      "step": 12438
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942615775520356e-05,
+      "loss": 0.502,
+      "step": 12439
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942598199292343e-05,
+      "loss": 0.4733,
+      "step": 12440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942580620380774e-05,
+      "loss": 0.4934,
+      "step": 12441
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942563038785655e-05,
+      "loss": 0.514,
+      "step": 12442
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942545454506997e-05,
+      "loss": 0.404,
+      "step": 12443
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942527867544793e-05,
+      "loss": 0.4867,
+      "step": 12444
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994251027789906e-05,
+      "loss": 0.4656,
+      "step": 12445
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942492685569795e-05,
+      "loss": 0.5391,
+      "step": 12446
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942475090557003e-05,
+      "loss": 0.496,
+      "step": 12447
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942457492860693e-05,
+      "loss": 0.4948,
+      "step": 12448
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942439892480863e-05,
+      "loss": 0.4367,
+      "step": 12449
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942422289417525e-05,
+      "loss": 0.408,
+      "step": 12450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942404683670685e-05,
+      "loss": 0.447,
+      "step": 12451
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942387075240336e-05,
+      "loss": 0.4471,
+      "step": 12452
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942369464126495e-05,
+      "loss": 0.4528,
+      "step": 12453
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994235185032916e-05,
+      "loss": 0.4754,
+      "step": 12454
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942334233848338e-05,
+      "loss": 0.4886,
+      "step": 12455
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942316614684035e-05,
+      "loss": 0.4671,
+      "step": 12456
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942298992836254e-05,
+      "loss": 0.465,
+      "step": 12457
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942281368304998e-05,
+      "loss": 0.4486,
+      "step": 12458
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942263741090274e-05,
+      "loss": 0.4273,
+      "step": 12459
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942246111192088e-05,
+      "loss": 0.4777,
+      "step": 12460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942228478610444e-05,
+      "loss": 0.4245,
+      "step": 12461
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942210843345343e-05,
+      "loss": 0.4884,
+      "step": 12462
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942193205396797e-05,
+      "loss": 0.5142,
+      "step": 12463
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.99421755647648e-05,
+      "loss": 0.5131,
+      "step": 12464
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.994215792144937e-05,
+      "loss": 0.4835,
+      "step": 12465
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942140275450503e-05,
+      "loss": 0.4678,
+      "step": 12466
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942122626768206e-05,
+      "loss": 0.4337,
+      "step": 12467
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942104975402482e-05,
+      "loss": 0.508,
+      "step": 12468
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9942087321353337e-05,
+      "loss": 0.5243,
+      "step": 12469
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994206966462078e-05,
+      "loss": 0.4627,
+      "step": 12470
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994205200520481e-05,
+      "loss": 0.4281,
+      "step": 12471
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994203434310543e-05,
+      "loss": 0.4398,
+      "step": 12472
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994201667832265e-05,
+      "loss": 0.4989,
+      "step": 12473
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941999010856476e-05,
+      "loss": 0.5232,
+      "step": 12474
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941981340706912e-05,
+      "loss": 0.4476,
+      "step": 12475
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941963667873954e-05,
+      "loss": 0.4892,
+      "step": 12476
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994194599235762e-05,
+      "loss": 0.4663,
+      "step": 12477
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941928314157906e-05,
+      "loss": 0.5618,
+      "step": 12478
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994191063327482e-05,
+      "loss": 0.4572,
+      "step": 12479
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994189294970836e-05,
+      "loss": 0.5374,
+      "step": 12480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941875263458544e-05,
+      "loss": 0.4493,
+      "step": 12481
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941857574525364e-05,
+      "loss": 0.4015,
+      "step": 12482
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941839882908834e-05,
+      "loss": 0.5142,
+      "step": 12483
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994182218860895e-05,
+      "loss": 0.4781,
+      "step": 12484
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994180449162573e-05,
+      "loss": 0.548,
+      "step": 12485
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941786791959164e-05,
+      "loss": 0.5472,
+      "step": 12486
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941769089609267e-05,
+      "loss": 0.4477,
+      "step": 12487
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941751384576037e-05,
+      "loss": 0.5373,
+      "step": 12488
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941733676859484e-05,
+      "loss": 0.5187,
+      "step": 12489
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994171596645961e-05,
+      "loss": 0.494,
+      "step": 12490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994169825337642e-05,
+      "loss": 0.4214,
+      "step": 12491
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941680537609915e-05,
+      "loss": 0.4505,
+      "step": 12492
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994166281916011e-05,
+      "loss": 0.5202,
+      "step": 12493
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941645098027003e-05,
+      "loss": 0.5108,
+      "step": 12494
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941627374210597e-05,
+      "loss": 0.4586,
+      "step": 12495
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941609647710903e-05,
+      "loss": 0.4533,
+      "step": 12496
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994159191852792e-05,
+      "loss": 0.4489,
+      "step": 12497
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994157418666165e-05,
+      "loss": 0.4798,
+      "step": 12498
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941556452112108e-05,
+      "loss": 0.4913,
+      "step": 12499
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941538714879293e-05,
+      "loss": 0.4848,
+      "step": 12500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994152097496321e-05,
+      "loss": 0.4751,
+      "step": 12501
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941503232363865e-05,
+      "loss": 0.5146,
+      "step": 12502
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994148548708126e-05,
+      "loss": 0.516,
+      "step": 12503
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941467739115402e-05,
+      "loss": 0.5196,
+      "step": 12504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941449988466297e-05,
+      "loss": 0.4783,
+      "step": 12505
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941432235133947e-05,
+      "loss": 0.6056,
+      "step": 12506
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941414479118356e-05,
+      "loss": 0.4602,
+      "step": 12507
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941396720419534e-05,
+      "loss": 0.4129,
+      "step": 12508
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994137895903748e-05,
+      "loss": 0.4502,
+      "step": 12509
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941361194972204e-05,
+      "loss": 0.5409,
+      "step": 12510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941343428223706e-05,
+      "loss": 0.5501,
+      "step": 12511
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994132565879199e-05,
+      "loss": 0.4573,
+      "step": 12512
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994130788667707e-05,
+      "loss": 0.4744,
+      "step": 12513
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941290111878945e-05,
+      "loss": 0.3907,
+      "step": 12514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994127233439761e-05,
+      "loss": 0.4178,
+      "step": 12515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994125455423309e-05,
+      "loss": 0.4281,
+      "step": 12516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941236771385373e-05,
+      "loss": 0.4901,
+      "step": 12517
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994121898585447e-05,
+      "loss": 0.4303,
+      "step": 12518
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941201197640387e-05,
+      "loss": 0.4789,
+      "step": 12519
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994118340674313e-05,
+      "loss": 0.378,
+      "step": 12520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941165613162698e-05,
+      "loss": 0.4539,
+      "step": 12521
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941147816899096e-05,
+      "loss": 0.4992,
+      "step": 12522
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941130017952335e-05,
+      "loss": 0.5467,
+      "step": 12523
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941112216322417e-05,
+      "loss": 0.4635,
+      "step": 12524
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941094412009346e-05,
+      "loss": 0.4444,
+      "step": 12525
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941076605013126e-05,
+      "loss": 0.5694,
+      "step": 12526
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941058795333762e-05,
+      "loss": 0.4967,
+      "step": 12527
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941040982971263e-05,
+      "loss": 0.4884,
+      "step": 12528
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941023167925627e-05,
+      "loss": 0.4718,
+      "step": 12529
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9941005350196866e-05,
+      "loss": 0.4633,
+      "step": 12530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940987529784975e-05,
+      "loss": 0.4623,
+      "step": 12531
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940969706689972e-05,
+      "loss": 0.5491,
+      "step": 12532
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940951880911853e-05,
+      "loss": 0.4823,
+      "step": 12533
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940934052450625e-05,
+      "loss": 0.5332,
+      "step": 12534
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994091622130629e-05,
+      "loss": 0.5016,
+      "step": 12535
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994089838747886e-05,
+      "loss": 0.5259,
+      "step": 12536
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940880550968328e-05,
+      "loss": 0.4866,
+      "step": 12537
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940862711774708e-05,
+      "loss": 0.5093,
+      "step": 12538
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940844869898006e-05,
+      "loss": 0.5806,
+      "step": 12539
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994082702533822e-05,
+      "loss": 0.427,
+      "step": 12540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994080917809536e-05,
+      "loss": 0.4388,
+      "step": 12541
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994079132816943e-05,
+      "loss": 0.4584,
+      "step": 12542
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940773475560434e-05,
+      "loss": 0.4359,
+      "step": 12543
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940755620268374e-05,
+      "loss": 0.5375,
+      "step": 12544
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940737762293262e-05,
+      "loss": 0.5493,
+      "step": 12545
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940719901635096e-05,
+      "loss": 0.5782,
+      "step": 12546
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994070203829388e-05,
+      "loss": 0.5037,
+      "step": 12547
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940684172269625e-05,
+      "loss": 0.6592,
+      "step": 12548
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940666303562335e-05,
+      "loss": 0.5584,
+      "step": 12549
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994064843217201e-05,
+      "loss": 0.4511,
+      "step": 12550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940630558098658e-05,
+      "loss": 0.521,
+      "step": 12551
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940612681342284e-05,
+      "loss": 0.5213,
+      "step": 12552
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994059480190289e-05,
+      "loss": 0.538,
+      "step": 12553
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940576919780485e-05,
+      "loss": 0.4668,
+      "step": 12554
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940559034975073e-05,
+      "loss": 0.4945,
+      "step": 12555
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940541147486656e-05,
+      "loss": 0.5092,
+      "step": 12556
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994052325731524e-05,
+      "loss": 0.4572,
+      "step": 12557
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940505364460832e-05,
+      "loss": 0.4649,
+      "step": 12558
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940487468923434e-05,
+      "loss": 0.4267,
+      "step": 12559
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940469570703053e-05,
+      "loss": 0.4285,
+      "step": 12560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994045166979969e-05,
+      "loss": 0.4822,
+      "step": 12561
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940433766213356e-05,
+      "loss": 0.5203,
+      "step": 12562
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940415859944048e-05,
+      "loss": 0.4467,
+      "step": 12563
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940397950991783e-05,
+      "loss": 0.4929,
+      "step": 12564
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994038003935655e-05,
+      "loss": 0.4664,
+      "step": 12565
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9940362125038366e-05,
+      "loss": 0.4681,
+      "step": 12566
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.994034420803723e-05,
+      "loss": 0.5015,
+      "step": 12567
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.994032628835315e-05,
+      "loss": 0.4828,
+      "step": 12568
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940308365986133e-05,
+      "loss": 0.4131,
+      "step": 12569
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940290440936173e-05,
+      "loss": 0.5088,
+      "step": 12570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940272513203286e-05,
+      "loss": 0.4533,
+      "step": 12571
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940254582787474e-05,
+      "loss": 0.509,
+      "step": 12572
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.994023664968874e-05,
+      "loss": 0.4167,
+      "step": 12573
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940218713907088e-05,
+      "loss": 0.5485,
+      "step": 12574
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940200775442525e-05,
+      "loss": 0.4603,
+      "step": 12575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940182834295056e-05,
+      "loss": 0.5265,
+      "step": 12576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940164890464684e-05,
+      "loss": 0.6405,
+      "step": 12577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.994014694395142e-05,
+      "loss": 0.5122,
+      "step": 12578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.994012899475526e-05,
+      "loss": 0.4803,
+      "step": 12579
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.994011104287621e-05,
+      "loss": 0.5343,
+      "step": 12580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940093088314284e-05,
+      "loss": 0.4244,
+      "step": 12581
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940075131069472e-05,
+      "loss": 0.494,
+      "step": 12582
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940057171141792e-05,
+      "loss": 0.5391,
+      "step": 12583
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940039208531247e-05,
+      "loss": 0.4982,
+      "step": 12584
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940021243237836e-05,
+      "loss": 0.494,
+      "step": 12585
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9940003275261568e-05,
+      "loss": 0.4503,
+      "step": 12586
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939985304602444e-05,
+      "loss": 0.431,
+      "step": 12587
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939967331260473e-05,
+      "loss": 0.4916,
+      "step": 12588
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939949355235656e-05,
+      "loss": 0.4379,
+      "step": 12589
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939931376528005e-05,
+      "loss": 0.4485,
+      "step": 12590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939913395137517e-05,
+      "loss": 0.5123,
+      "step": 12591
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939895411064203e-05,
+      "loss": 0.4898,
+      "step": 12592
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939877424308062e-05,
+      "loss": 0.4963,
+      "step": 12593
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.99398594348691e-05,
+      "loss": 0.5207,
+      "step": 12594
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939841442747324e-05,
+      "loss": 0.4766,
+      "step": 12595
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993982344794274e-05,
+      "loss": 0.4218,
+      "step": 12596
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939805450455352e-05,
+      "loss": 0.5113,
+      "step": 12597
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939787450285164e-05,
+      "loss": 0.436,
+      "step": 12598
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993976944743218e-05,
+      "loss": 0.4652,
+      "step": 12599
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939751441896405e-05,
+      "loss": 0.4495,
+      "step": 12600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939733433677845e-05,
+      "loss": 0.5412,
+      "step": 12601
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939715422776508e-05,
+      "loss": 0.4249,
+      "step": 12602
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939697409192388e-05,
+      "loss": 0.4751,
+      "step": 12603
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.99396793929255e-05,
+      "loss": 0.4306,
+      "step": 12604
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993966137397585e-05,
+      "loss": 0.5216,
+      "step": 12605
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939643352343435e-05,
+      "loss": 0.4184,
+      "step": 12606
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939625328028267e-05,
+      "loss": 0.4639,
+      "step": 12607
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939607301030346e-05,
+      "loss": 0.5253,
+      "step": 12608
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993958927134968e-05,
+      "loss": 0.4243,
+      "step": 12609
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993957123898627e-05,
+      "loss": 0.4917,
+      "step": 12610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939553203940122e-05,
+      "loss": 0.4635,
+      "step": 12611
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939535166211246e-05,
+      "loss": 0.4743,
+      "step": 12612
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993951712579964e-05,
+      "loss": 0.4148,
+      "step": 12613
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939499082705312e-05,
+      "loss": 0.4608,
+      "step": 12614
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993948103692827e-05,
+      "loss": 0.4413,
+      "step": 12615
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939462988468514e-05,
+      "loss": 0.506,
+      "step": 12616
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939444937326047e-05,
+      "loss": 0.4967,
+      "step": 12617
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939426883500882e-05,
+      "loss": 0.4391,
+      "step": 12618
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939408826993018e-05,
+      "loss": 0.5232,
+      "step": 12619
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993939076780246e-05,
+      "loss": 0.4594,
+      "step": 12620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939372705929213e-05,
+      "loss": 0.4611,
+      "step": 12621
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993935464137328e-05,
+      "loss": 0.4465,
+      "step": 12622
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939336574134675e-05,
+      "loss": 0.4206,
+      "step": 12623
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939318504213397e-05,
+      "loss": 0.5195,
+      "step": 12624
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939300431609447e-05,
+      "loss": 0.525,
+      "step": 12625
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939282356322835e-05,
+      "loss": 0.4119,
+      "step": 12626
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939264278353562e-05,
+      "loss": 0.4718,
+      "step": 12627
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993924619770164e-05,
+      "loss": 0.4454,
+      "step": 12628
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939228114367063e-05,
+      "loss": 0.4591,
+      "step": 12629
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993921002834985e-05,
+      "loss": 0.4716,
+      "step": 12630
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993919193964999e-05,
+      "loss": 0.4097,
+      "step": 12631
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939173848267498e-05,
+      "loss": 0.5031,
+      "step": 12632
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939155754202375e-05,
+      "loss": 0.4435,
+      "step": 12633
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993913765745463e-05,
+      "loss": 0.5103,
+      "step": 12634
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993911955802426e-05,
+      "loss": 0.4237,
+      "step": 12635
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939101455911282e-05,
+      "loss": 0.4904,
+      "step": 12636
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939083351115693e-05,
+      "loss": 0.4321,
+      "step": 12637
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9939065243637495e-05,
+      "loss": 0.527,
+      "step": 12638
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.99390471334767e-05,
+      "loss": 0.4962,
+      "step": 12639
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993902902063331e-05,
+      "loss": 0.4572,
+      "step": 12640
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993901090510733e-05,
+      "loss": 0.5156,
+      "step": 12641
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993899278689876e-05,
+      "loss": 0.4413,
+      "step": 12642
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938974666007613e-05,
+      "loss": 0.4464,
+      "step": 12643
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993895654243389e-05,
+      "loss": 0.5056,
+      "step": 12644
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938938416177594e-05,
+      "loss": 0.4749,
+      "step": 12645
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938920287238733e-05,
+      "loss": 0.4862,
+      "step": 12646
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938902155617315e-05,
+      "loss": 0.6064,
+      "step": 12647
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938884021313337e-05,
+      "loss": 0.483,
+      "step": 12648
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993886588432681e-05,
+      "loss": 0.3875,
+      "step": 12649
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938847744657735e-05,
+      "loss": 0.4699,
+      "step": 12650
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938829602306117e-05,
+      "loss": 0.5047,
+      "step": 12651
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938811457271966e-05,
+      "loss": 0.4828,
+      "step": 12652
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993879330955528e-05,
+      "loss": 0.4927,
+      "step": 12653
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993877515915607e-05,
+      "loss": 0.3831,
+      "step": 12654
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938757006074336e-05,
+      "loss": 0.4953,
+      "step": 12655
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938738850310084e-05,
+      "loss": 0.4747,
+      "step": 12656
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938720691863322e-05,
+      "loss": 0.4838,
+      "step": 12657
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938702530734055e-05,
+      "loss": 0.4593,
+      "step": 12658
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993868436692228e-05,
+      "loss": 0.4081,
+      "step": 12659
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938666200428015e-05,
+      "loss": 0.4814,
+      "step": 12660
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993864803125125e-05,
+      "loss": 0.5138,
+      "step": 12661
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938629859392003e-05,
+      "loss": 0.4933,
+      "step": 12662
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938611684850267e-05,
+      "loss": 0.5279,
+      "step": 12663
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993859350762606e-05,
+      "loss": 0.4254,
+      "step": 12664
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.993857532771938e-05,
+      "loss": 0.4538,
+      "step": 12665
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9938557145130227e-05,
+      "loss": 0.4803,
+      "step": 12666
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938538959858613e-05,
+      "loss": 0.4654,
+      "step": 12667
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993852077190454e-05,
+      "loss": 0.4149,
+      "step": 12668
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993850258126802e-05,
+      "loss": 0.4108,
+      "step": 12669
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938484387949044e-05,
+      "loss": 0.4994,
+      "step": 12670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938466191947626e-05,
+      "loss": 0.4302,
+      "step": 12671
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938447993263774e-05,
+      "loss": 0.4223,
+      "step": 12672
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938429791897484e-05,
+      "loss": 0.5551,
+      "step": 12673
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993841158784877e-05,
+      "loss": 0.4897,
+      "step": 12674
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938393381117626e-05,
+      "loss": 0.5401,
+      "step": 12675
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993837517170407e-05,
+      "loss": 0.4688,
+      "step": 12676
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938356959608093e-05,
+      "loss": 0.4785,
+      "step": 12677
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993833874482971e-05,
+      "loss": 0.4841,
+      "step": 12678
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938320527368927e-05,
+      "loss": 0.5705,
+      "step": 12679
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938302307225742e-05,
+      "loss": 0.4122,
+      "step": 12680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993828408440016e-05,
+      "loss": 0.4445,
+      "step": 12681
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993826585889219e-05,
+      "loss": 0.5261,
+      "step": 12682
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938247630701837e-05,
+      "loss": 0.5314,
+      "step": 12683
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.99382293998291e-05,
+      "loss": 0.404,
+      "step": 12684
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938211166273996e-05,
+      "loss": 0.4907,
+      "step": 12685
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938192930036518e-05,
+      "loss": 0.4344,
+      "step": 12686
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938174691116673e-05,
+      "loss": 0.4451,
+      "step": 12687
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993815644951447e-05,
+      "loss": 0.4787,
+      "step": 12688
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938138205229915e-05,
+      "loss": 0.4706,
+      "step": 12689
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993811995826301e-05,
+      "loss": 0.3988,
+      "step": 12690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938101708613758e-05,
+      "loss": 0.4794,
+      "step": 12691
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938083456282166e-05,
+      "loss": 0.4735,
+      "step": 12692
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938065201268238e-05,
+      "loss": 0.4179,
+      "step": 12693
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993804694357198e-05,
+      "loss": 0.4346,
+      "step": 12694
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938028683193398e-05,
+      "loss": 0.5665,
+      "step": 12695
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9938010420132494e-05,
+      "loss": 0.4819,
+      "step": 12696
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937992154389274e-05,
+      "loss": 0.5319,
+      "step": 12697
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937973885963748e-05,
+      "loss": 0.4162,
+      "step": 12698
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937955614855912e-05,
+      "loss": 0.5905,
+      "step": 12699
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937937341065777e-05,
+      "loss": 0.3926,
+      "step": 12700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937919064593346e-05,
+      "loss": 0.5137,
+      "step": 12701
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937900785438623e-05,
+      "loss": 0.5894,
+      "step": 12702
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937882503601614e-05,
+      "loss": 0.4481,
+      "step": 12703
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937864219082327e-05,
+      "loss": 0.491,
+      "step": 12704
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993784593188076e-05,
+      "loss": 0.5115,
+      "step": 12705
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937827641996924e-05,
+      "loss": 0.4404,
+      "step": 12706
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937809349430824e-05,
+      "loss": 0.4645,
+      "step": 12707
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937791054182458e-05,
+      "loss": 0.5036,
+      "step": 12708
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993777275625184e-05,
+      "loss": 0.5167,
+      "step": 12709
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993775445563897e-05,
+      "loss": 0.4513,
+      "step": 12710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937736152343852e-05,
+      "loss": 0.4971,
+      "step": 12711
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937717846366492e-05,
+      "loss": 0.4556,
+      "step": 12712
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937699537706897e-05,
+      "loss": 0.5281,
+      "step": 12713
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993768122636507e-05,
+      "loss": 0.5606,
+      "step": 12714
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937662912341017e-05,
+      "loss": 0.4114,
+      "step": 12715
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993764459563474e-05,
+      "loss": 0.4613,
+      "step": 12716
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993762627624625e-05,
+      "loss": 0.5359,
+      "step": 12717
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937607954175547e-05,
+      "loss": 0.488,
+      "step": 12718
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937589629422635e-05,
+      "loss": 0.376,
+      "step": 12719
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937571301987525e-05,
+      "loss": 0.5045,
+      "step": 12720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937552971870214e-05,
+      "loss": 0.4173,
+      "step": 12721
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937534639070715e-05,
+      "loss": 0.5542,
+      "step": 12722
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937516303589026e-05,
+      "loss": 0.5574,
+      "step": 12723
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937497965425153e-05,
+      "loss": 0.5221,
+      "step": 12724
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993747962457911e-05,
+      "loss": 0.4886,
+      "step": 12725
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993746128105089e-05,
+      "loss": 0.5615,
+      "step": 12726
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.99374429348405e-05,
+      "loss": 0.4961,
+      "step": 12727
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937424585947952e-05,
+      "loss": 0.5096,
+      "step": 12728
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937406234373244e-05,
+      "loss": 0.4928,
+      "step": 12729
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937387880116388e-05,
+      "loss": 0.4357,
+      "step": 12730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993736952317738e-05,
+      "loss": 0.4802,
+      "step": 12731
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993735116355623e-05,
+      "loss": 0.5201,
+      "step": 12732
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937332801252945e-05,
+      "loss": 0.4641,
+      "step": 12733
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937314436267526e-05,
+      "loss": 0.4107,
+      "step": 12734
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937296068599977e-05,
+      "loss": 0.446,
+      "step": 12735
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993727769825031e-05,
+      "loss": 0.5096,
+      "step": 12736
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937259325218525e-05,
+      "loss": 0.4393,
+      "step": 12737
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993724094950462e-05,
+      "loss": 0.4619,
+      "step": 12738
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937222571108617e-05,
+      "loss": 0.5127,
+      "step": 12739
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937204190030504e-05,
+      "loss": 0.4156,
+      "step": 12740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937185806270295e-05,
+      "loss": 0.5212,
+      "step": 12741
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937167419827993e-05,
+      "loss": 0.5764,
+      "step": 12742
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937149030703606e-05,
+      "loss": 0.5189,
+      "step": 12743
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937130638897133e-05,
+      "loss": 0.4587,
+      "step": 12744
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937112244408584e-05,
+      "loss": 0.4682,
+      "step": 12745
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993709384723796e-05,
+      "loss": 0.4574,
+      "step": 12746
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993707544738527e-05,
+      "loss": 0.4375,
+      "step": 12747
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937057044850515e-05,
+      "loss": 0.4504,
+      "step": 12748
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937038639633704e-05,
+      "loss": 0.4292,
+      "step": 12749
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937020231734835e-05,
+      "loss": 0.4615,
+      "step": 12750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9937001821153924e-05,
+      "loss": 0.4994,
+      "step": 12751
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9936983407890965e-05,
+      "loss": 0.4808,
+      "step": 12752
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993696499194597e-05,
+      "loss": 0.417,
+      "step": 12753
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993694657331894e-05,
+      "loss": 0.6105,
+      "step": 12754
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9936928152009883e-05,
+      "loss": 0.4879,
+      "step": 12755
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9936909728018804e-05,
+      "loss": 0.48,
+      "step": 12756
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9936891301345707e-05,
+      "loss": 0.5,
+      "step": 12757
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9936872871990592e-05,
+      "loss": 0.5121,
+      "step": 12758
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9936854439953473e-05,
+      "loss": 0.4982,
+      "step": 12759
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993683600523435e-05,
+      "loss": 0.4032,
+      "step": 12760
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9936817567833227e-05,
+      "loss": 0.5317,
+      "step": 12761
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993679912775011e-05,
+      "loss": 0.5109,
+      "step": 12762
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9936780684985005e-05,
+      "loss": 0.4229,
+      "step": 12763
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.993676223953792e-05,
+      "loss": 0.4586,
+      "step": 12764
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993674379140885e-05,
+      "loss": 0.4696,
+      "step": 12765
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993672534059781e-05,
+      "loss": 0.4634,
+      "step": 12766
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936706887104804e-05,
+      "loss": 0.4274,
+      "step": 12767
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993668843092983e-05,
+      "loss": 0.4507,
+      "step": 12768
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936669972072896e-05,
+      "loss": 0.4714,
+      "step": 12769
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936651510534012e-05,
+      "loss": 0.479,
+      "step": 12770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936633046313178e-05,
+      "loss": 0.5058,
+      "step": 12771
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936614579410404e-05,
+      "loss": 0.5128,
+      "step": 12772
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936596109825686e-05,
+      "loss": 0.5952,
+      "step": 12773
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936577637559035e-05,
+      "loss": 0.4131,
+      "step": 12774
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936559162610457e-05,
+      "loss": 0.4926,
+      "step": 12775
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936540684979956e-05,
+      "loss": 0.4967,
+      "step": 12776
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936522204667536e-05,
+      "loss": 0.5078,
+      "step": 12777
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.99365037216732e-05,
+      "loss": 0.4185,
+      "step": 12778
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936485235996956e-05,
+      "loss": 0.5817,
+      "step": 12779
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936466747638806e-05,
+      "loss": 0.5056,
+      "step": 12780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993644825659876e-05,
+      "loss": 0.477,
+      "step": 12781
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936429762876823e-05,
+      "loss": 0.4084,
+      "step": 12782
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936411266472992e-05,
+      "loss": 0.5077,
+      "step": 12783
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936392767387282e-05,
+      "loss": 0.462,
+      "step": 12784
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936374265619686e-05,
+      "loss": 0.4834,
+      "step": 12785
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993635576117022e-05,
+      "loss": 0.4865,
+      "step": 12786
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936337254038888e-05,
+      "loss": 0.4523,
+      "step": 12787
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993631874422569e-05,
+      "loss": 0.5906,
+      "step": 12788
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993630023173063e-05,
+      "loss": 0.5065,
+      "step": 12789
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993628171655372e-05,
+      "loss": 0.4609,
+      "step": 12790
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936263198694962e-05,
+      "loss": 0.4991,
+      "step": 12791
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993624467815436e-05,
+      "loss": 0.6065,
+      "step": 12792
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936226154931916e-05,
+      "loss": 0.43,
+      "step": 12793
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993620762902764e-05,
+      "loss": 0.5589,
+      "step": 12794
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936189100441533e-05,
+      "loss": 0.4926,
+      "step": 12795
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936170569173603e-05,
+      "loss": 0.4575,
+      "step": 12796
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936152035223854e-05,
+      "loss": 0.4689,
+      "step": 12797
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993613349859229e-05,
+      "loss": 0.5148,
+      "step": 12798
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936114959278923e-05,
+      "loss": 0.5467,
+      "step": 12799
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936096417283747e-05,
+      "loss": 0.4035,
+      "step": 12800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9936077872606774e-05,
+      "loss": 0.4754,
+      "step": 12801
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993605932524801e-05,
+      "loss": 0.4551,
+      "step": 12802
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993604077520745e-05,
+      "loss": 0.529,
+      "step": 12803
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993602222248511e-05,
+      "loss": 0.4765,
+      "step": 12804
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993600366708099e-05,
+      "loss": 0.5199,
+      "step": 12805
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935985108995098e-05,
+      "loss": 0.4591,
+      "step": 12806
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935966548227436e-05,
+      "loss": 0.4443,
+      "step": 12807
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993594798477801e-05,
+      "loss": 0.5076,
+      "step": 12808
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935929418646823e-05,
+      "loss": 0.5862,
+      "step": 12809
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993591084983389e-05,
+      "loss": 0.4915,
+      "step": 12810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.99358922783392e-05,
+      "loss": 0.4155,
+      "step": 12811
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935873704162768e-05,
+      "loss": 0.5252,
+      "step": 12812
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.99358551273046e-05,
+      "loss": 0.5562,
+      "step": 12813
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.99358365477647e-05,
+      "loss": 0.5063,
+      "step": 12814
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935817965543065e-05,
+      "loss": 0.5283,
+      "step": 12815
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993579938063971e-05,
+      "loss": 0.5075,
+      "step": 12816
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935780793054636e-05,
+      "loss": 0.5195,
+      "step": 12817
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993576220278785e-05,
+      "loss": 0.4669,
+      "step": 12818
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993574360983935e-05,
+      "loss": 0.5528,
+      "step": 12819
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993572501420915e-05,
+      "loss": 0.4732,
+      "step": 12820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993570641589725e-05,
+      "loss": 0.5305,
+      "step": 12821
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935687814903657e-05,
+      "loss": 0.5136,
+      "step": 12822
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935669211228375e-05,
+      "loss": 0.375,
+      "step": 12823
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993565060487141e-05,
+      "loss": 0.4947,
+      "step": 12824
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935631995832768e-05,
+      "loss": 0.4499,
+      "step": 12825
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935613384112452e-05,
+      "loss": 0.4725,
+      "step": 12826
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935594769710467e-05,
+      "loss": 0.4488,
+      "step": 12827
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935576152626816e-05,
+      "loss": 0.4706,
+      "step": 12828
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993555753286151e-05,
+      "loss": 0.4633,
+      "step": 12829
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993553891041455e-05,
+      "loss": 0.4302,
+      "step": 12830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935520285285942e-05,
+      "loss": 0.5341,
+      "step": 12831
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935501657475686e-05,
+      "loss": 0.4531,
+      "step": 12832
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935483026983796e-05,
+      "loss": 0.4236,
+      "step": 12833
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993546439381027e-05,
+      "loss": 0.4603,
+      "step": 12834
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993544575795512e-05,
+      "loss": 0.4732,
+      "step": 12835
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935427119418343e-05,
+      "loss": 0.4921,
+      "step": 12836
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935408478199953e-05,
+      "loss": 0.4706,
+      "step": 12837
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935389834299944e-05,
+      "loss": 0.5121,
+      "step": 12838
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993537118771833e-05,
+      "loss": 0.5001,
+      "step": 12839
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935352538455112e-05,
+      "loss": 0.4493,
+      "step": 12840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935333886510292e-05,
+      "loss": 0.4597,
+      "step": 12841
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993531523188388e-05,
+      "loss": 0.5829,
+      "step": 12842
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935296574575887e-05,
+      "loss": 0.4217,
+      "step": 12843
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935277914586307e-05,
+      "loss": 0.5717,
+      "step": 12844
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935259251915147e-05,
+      "loss": 0.5565,
+      "step": 12845
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935240586562416e-05,
+      "loss": 0.4905,
+      "step": 12846
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935221918528114e-05,
+      "loss": 0.4683,
+      "step": 12847
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935203247812255e-05,
+      "loss": 0.4531,
+      "step": 12848
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935184574414836e-05,
+      "loss": 0.5109,
+      "step": 12849
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935165898335862e-05,
+      "loss": 0.4357,
+      "step": 12850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993514721957534e-05,
+      "loss": 0.4568,
+      "step": 12851
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935128538133277e-05,
+      "loss": 0.4413,
+      "step": 12852
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993510985400968e-05,
+      "loss": 0.5504,
+      "step": 12853
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935091167204545e-05,
+      "loss": 0.5344,
+      "step": 12854
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935072477717883e-05,
+      "loss": 0.5116,
+      "step": 12855
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935053785549702e-05,
+      "loss": 0.5141,
+      "step": 12856
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.99350350907e-05,
+      "loss": 0.4901,
+      "step": 12857
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9935016393168786e-05,
+      "loss": 0.5159,
+      "step": 12858
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9934997692956065e-05,
+      "loss": 0.471,
+      "step": 12859
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993497899006184e-05,
+      "loss": 0.4546,
+      "step": 12860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993496028448612e-05,
+      "loss": 0.4693,
+      "step": 12861
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993494157622891e-05,
+      "loss": 0.4746,
+      "step": 12862
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.993492286529021e-05,
+      "loss": 0.4974,
+      "step": 12863
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934904151670027e-05,
+      "loss": 0.4276,
+      "step": 12864
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993488543536837e-05,
+      "loss": 0.4444,
+      "step": 12865
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993486671638524e-05,
+      "loss": 0.4724,
+      "step": 12866
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934847994720642e-05,
+      "loss": 0.3914,
+      "step": 12867
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934829270374584e-05,
+      "loss": 0.5019,
+      "step": 12868
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934810543347067e-05,
+      "loss": 0.5528,
+      "step": 12869
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934791813638097e-05,
+      "loss": 0.5085,
+      "step": 12870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934773081247682e-05,
+      "loss": 0.5298,
+      "step": 12871
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934754346175825e-05,
+      "loss": 0.5522,
+      "step": 12872
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993473560842253e-05,
+      "loss": 0.4402,
+      "step": 12873
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993471686798781e-05,
+      "loss": 0.421,
+      "step": 12874
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934698124871656e-05,
+      "loss": 0.5449,
+      "step": 12875
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993467937907408e-05,
+      "loss": 0.4783,
+      "step": 12876
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934660630595092e-05,
+      "loss": 0.4958,
+      "step": 12877
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934641879434694e-05,
+      "loss": 0.4561,
+      "step": 12878
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934623125592885e-05,
+      "loss": 0.5943,
+      "step": 12879
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934604369069675e-05,
+      "loss": 0.5107,
+      "step": 12880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993458560986507e-05,
+      "loss": 0.4608,
+      "step": 12881
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934566847979077e-05,
+      "loss": 0.4755,
+      "step": 12882
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934548083411696e-05,
+      "loss": 0.5207,
+      "step": 12883
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993452931616293e-05,
+      "loss": 0.5085,
+      "step": 12884
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934510546232795e-05,
+      "loss": 0.5068,
+      "step": 12885
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934491773621285e-05,
+      "loss": 0.4312,
+      "step": 12886
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993447299832841e-05,
+      "loss": 0.4876,
+      "step": 12887
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934454220354172e-05,
+      "loss": 0.5075,
+      "step": 12888
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934435439698583e-05,
+      "loss": 0.4845,
+      "step": 12889
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993441665636164e-05,
+      "loss": 0.4503,
+      "step": 12890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934397870343353e-05,
+      "loss": 0.5182,
+      "step": 12891
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934379081643726e-05,
+      "loss": 0.5418,
+      "step": 12892
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934360290262762e-05,
+      "loss": 0.565,
+      "step": 12893
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934341496200468e-05,
+      "loss": 0.5031,
+      "step": 12894
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934322699456853e-05,
+      "loss": 0.4584,
+      "step": 12895
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934303900031912e-05,
+      "loss": 0.5703,
+      "step": 12896
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934285097925657e-05,
+      "loss": 0.4356,
+      "step": 12897
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934266293138096e-05,
+      "loss": 0.5369,
+      "step": 12898
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934247485669225e-05,
+      "loss": 0.5353,
+      "step": 12899
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934228675519058e-05,
+      "loss": 0.4813,
+      "step": 12900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934209862687594e-05,
+      "loss": 0.4512,
+      "step": 12901
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993419104717484e-05,
+      "loss": 0.5295,
+      "step": 12902
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934172228980805e-05,
+      "loss": 0.5049,
+      "step": 12903
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934153408105486e-05,
+      "loss": 0.5771,
+      "step": 12904
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934134584548898e-05,
+      "loss": 0.5049,
+      "step": 12905
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934115758311034e-05,
+      "loss": 0.4035,
+      "step": 12906
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934096929391907e-05,
+      "loss": 0.4912,
+      "step": 12907
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934078097791525e-05,
+      "loss": 0.4809,
+      "step": 12908
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934059263509887e-05,
+      "loss": 0.4696,
+      "step": 12909
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934040426547e-05,
+      "loss": 0.4251,
+      "step": 12910
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934021586902865e-05,
+      "loss": 0.444,
+      "step": 12911
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9934002744577497e-05,
+      "loss": 0.4646,
+      "step": 12912
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933983899570894e-05,
+      "loss": 0.5515,
+      "step": 12913
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993396505188306e-05,
+      "loss": 0.4011,
+      "step": 12914
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933946201514e-05,
+      "loss": 0.4721,
+      "step": 12915
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993392734846373e-05,
+      "loss": 0.4645,
+      "step": 12916
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933908492732243e-05,
+      "loss": 0.4697,
+      "step": 12917
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933889634319544e-05,
+      "loss": 0.5407,
+      "step": 12918
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933870773225644e-05,
+      "loss": 0.4969,
+      "step": 12919
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933851909450546e-05,
+      "loss": 0.5471,
+      "step": 12920
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933833042994256e-05,
+      "loss": 0.4742,
+      "step": 12921
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933814173856776e-05,
+      "loss": 0.4823,
+      "step": 12922
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933795302038114e-05,
+      "loss": 0.438,
+      "step": 12923
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933776427538274e-05,
+      "loss": 0.4858,
+      "step": 12924
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993375755035726e-05,
+      "loss": 0.5283,
+      "step": 12925
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933738670495078e-05,
+      "loss": 0.4682,
+      "step": 12926
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993371978795174e-05,
+      "loss": 0.5131,
+      "step": 12927
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993370090272724e-05,
+      "loss": 0.4487,
+      "step": 12928
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933682014821585e-05,
+      "loss": 0.4631,
+      "step": 12929
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933663124234784e-05,
+      "loss": 0.4682,
+      "step": 12930
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933644230966846e-05,
+      "loss": 0.479,
+      "step": 12931
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933625335017767e-05,
+      "loss": 0.4975,
+      "step": 12932
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933606436387554e-05,
+      "loss": 0.4924,
+      "step": 12933
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933587535076218e-05,
+      "loss": 0.5027,
+      "step": 12934
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933568631083758e-05,
+      "loss": 0.4435,
+      "step": 12935
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933549724410182e-05,
+      "loss": 0.5143,
+      "step": 12936
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933530815055495e-05,
+      "loss": 0.4582,
+      "step": 12937
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933511903019702e-05,
+      "loss": 0.4738,
+      "step": 12938
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933492988302806e-05,
+      "loss": 0.4291,
+      "step": 12939
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933474070904813e-05,
+      "loss": 0.5301,
+      "step": 12940
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933455150825734e-05,
+      "loss": 0.3758,
+      "step": 12941
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933436228065566e-05,
+      "loss": 0.4856,
+      "step": 12942
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933417302624317e-05,
+      "loss": 0.5862,
+      "step": 12943
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933398374501993e-05,
+      "loss": 0.5079,
+      "step": 12944
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933379443698596e-05,
+      "loss": 0.5142,
+      "step": 12945
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933360510214136e-05,
+      "loss": 0.5139,
+      "step": 12946
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993334157404861e-05,
+      "loss": 0.5713,
+      "step": 12947
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933322635202036e-05,
+      "loss": 0.4297,
+      "step": 12948
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993330369367441e-05,
+      "loss": 0.5565,
+      "step": 12949
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933284749465737e-05,
+      "loss": 0.4096,
+      "step": 12950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993326580257603e-05,
+      "loss": 0.4967,
+      "step": 12951
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993324685300528e-05,
+      "loss": 0.4492,
+      "step": 12952
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933227900753503e-05,
+      "loss": 0.4012,
+      "step": 12953
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933208945820702e-05,
+      "loss": 0.5004,
+      "step": 12954
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993318998820688e-05,
+      "loss": 0.4799,
+      "step": 12955
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933171027912044e-05,
+      "loss": 0.6072,
+      "step": 12956
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.99331520649362e-05,
+      "loss": 0.5092,
+      "step": 12957
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993313309927935e-05,
+      "loss": 0.587,
+      "step": 12958
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933114130941504e-05,
+      "loss": 0.502,
+      "step": 12959
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9933095159922663e-05,
+      "loss": 0.4298,
+      "step": 12960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.993307618622283e-05,
+      "loss": 0.4875,
+      "step": 12961
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9933057209842018e-05,
+      "loss": 0.505,
+      "step": 12962
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9933038230780223e-05,
+      "loss": 0.471,
+      "step": 12963
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993301924903746e-05,
+      "loss": 0.3919,
+      "step": 12964
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9933000264613723e-05,
+      "loss": 0.4529,
+      "step": 12965
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932981277509025e-05,
+      "loss": 0.4877,
+      "step": 12966
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993296228772337e-05,
+      "loss": 0.4828,
+      "step": 12967
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993294329525676e-05,
+      "loss": 0.4231,
+      "step": 12968
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932924300109204e-05,
+      "loss": 0.538,
+      "step": 12969
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932905302280706e-05,
+      "loss": 0.4537,
+      "step": 12970
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932886301771268e-05,
+      "loss": 0.4543,
+      "step": 12971
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.99328672985809e-05,
+      "loss": 0.4367,
+      "step": 12972
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.99328482927096e-05,
+      "loss": 0.5098,
+      "step": 12973
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932829284157385e-05,
+      "loss": 0.445,
+      "step": 12974
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993281027292425e-05,
+      "loss": 0.4999,
+      "step": 12975
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.99327912590102e-05,
+      "loss": 0.4138,
+      "step": 12976
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932772242415246e-05,
+      "loss": 0.4591,
+      "step": 12977
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993275322313939e-05,
+      "loss": 0.4274,
+      "step": 12978
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932734201182637e-05,
+      "loss": 0.446,
+      "step": 12979
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993271517654499e-05,
+      "loss": 0.4108,
+      "step": 12980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932696149226462e-05,
+      "loss": 0.4968,
+      "step": 12981
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993267711922705e-05,
+      "loss": 0.482,
+      "step": 12982
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932658086546762e-05,
+      "loss": 0.4379,
+      "step": 12983
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932639051185604e-05,
+      "loss": 0.4908,
+      "step": 12984
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932620013143577e-05,
+      "loss": 0.4177,
+      "step": 12985
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932600972420694e-05,
+      "loss": 0.4652,
+      "step": 12986
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932581929016955e-05,
+      "loss": 0.5421,
+      "step": 12987
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932562882932364e-05,
+      "loss": 0.3766,
+      "step": 12988
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932543834166927e-05,
+      "loss": 0.4344,
+      "step": 12989
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993252478272065e-05,
+      "loss": 0.484,
+      "step": 12990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993250572859354e-05,
+      "loss": 0.4835,
+      "step": 12991
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932486671785603e-05,
+      "loss": 0.4701,
+      "step": 12992
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932467612296837e-05,
+      "loss": 0.4911,
+      "step": 12993
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932448550127256e-05,
+      "loss": 0.5158,
+      "step": 12994
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932429485276853e-05,
+      "loss": 0.5208,
+      "step": 12995
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932410417745648e-05,
+      "loss": 0.4716,
+      "step": 12996
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932391347533635e-05,
+      "loss": 0.4561,
+      "step": 12997
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932372274640827e-05,
+      "loss": 0.4255,
+      "step": 12998
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993235319906722e-05,
+      "loss": 0.5348,
+      "step": 12999
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993233412081283e-05,
+      "loss": 0.4868,
+      "step": 13000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932315039877654e-05,
+      "loss": 0.5341,
+      "step": 13001
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932295956261697e-05,
+      "loss": 0.4991,
+      "step": 13002
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932276869964968e-05,
+      "loss": 0.4508,
+      "step": 13003
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932257780987476e-05,
+      "loss": 0.4922,
+      "step": 13004
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993223868932922e-05,
+      "loss": 0.5723,
+      "step": 13005
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.99322195949902e-05,
+      "loss": 0.5089,
+      "step": 13006
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993220049797043e-05,
+      "loss": 0.4811,
+      "step": 13007
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932181398269918e-05,
+      "loss": 0.6177,
+      "step": 13008
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932162295888657e-05,
+      "loss": 0.4536,
+      "step": 13009
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932143190826663e-05,
+      "loss": 0.4692,
+      "step": 13010
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932124083083938e-05,
+      "loss": 0.4131,
+      "step": 13011
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932104972660482e-05,
+      "loss": 0.5003,
+      "step": 13012
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932085859556306e-05,
+      "loss": 0.4677,
+      "step": 13013
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932066743771417e-05,
+      "loss": 0.5179,
+      "step": 13014
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932047625305814e-05,
+      "loss": 0.4786,
+      "step": 13015
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932028504159504e-05,
+      "loss": 0.5442,
+      "step": 13016
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9932009380332494e-05,
+      "loss": 0.4339,
+      "step": 13017
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931990253824787e-05,
+      "loss": 0.4208,
+      "step": 13018
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931971124636394e-05,
+      "loss": 0.4843,
+      "step": 13019
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993195199276731e-05,
+      "loss": 0.5191,
+      "step": 13020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993193285821755e-05,
+      "loss": 0.4923,
+      "step": 13021
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993191372098711e-05,
+      "loss": 0.4807,
+      "step": 13022
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931894581076005e-05,
+      "loss": 0.4986,
+      "step": 13023
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931875438484233e-05,
+      "loss": 0.4643,
+      "step": 13024
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931856293211803e-05,
+      "loss": 0.436,
+      "step": 13025
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931837145258716e-05,
+      "loss": 0.4096,
+      "step": 13026
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993181799462498e-05,
+      "loss": 0.4826,
+      "step": 13027
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931798841310602e-05,
+      "loss": 0.4312,
+      "step": 13028
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931779685315584e-05,
+      "loss": 0.4204,
+      "step": 13029
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931760526639935e-05,
+      "loss": 0.4853,
+      "step": 13030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931741365283653e-05,
+      "loss": 0.4772,
+      "step": 13031
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993172220124675e-05,
+      "loss": 0.4439,
+      "step": 13032
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931703034529226e-05,
+      "loss": 0.4315,
+      "step": 13033
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993168386513109e-05,
+      "loss": 0.5024,
+      "step": 13034
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931664693052348e-05,
+      "loss": 0.4741,
+      "step": 13035
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931645518293003e-05,
+      "loss": 0.4397,
+      "step": 13036
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993162634085306e-05,
+      "loss": 0.4934,
+      "step": 13037
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931607160732525e-05,
+      "loss": 0.4965,
+      "step": 13038
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931587977931398e-05,
+      "loss": 0.5168,
+      "step": 13039
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931568792449697e-05,
+      "loss": 0.468,
+      "step": 13040
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931549604287414e-05,
+      "loss": 0.4884,
+      "step": 13041
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993153041344456e-05,
+      "loss": 0.4209,
+      "step": 13042
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993151121992114e-05,
+      "loss": 0.4883,
+      "step": 13043
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931492023717158e-05,
+      "loss": 0.5367,
+      "step": 13044
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931472824832624e-05,
+      "loss": 0.5961,
+      "step": 13045
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931453623267532e-05,
+      "loss": 0.5378,
+      "step": 13046
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.99314344190219e-05,
+      "loss": 0.4572,
+      "step": 13047
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931415212095727e-05,
+      "loss": 0.403,
+      "step": 13048
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931396002489016e-05,
+      "loss": 0.5428,
+      "step": 13049
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931376790201774e-05,
+      "loss": 0.4198,
+      "step": 13050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993135757523401e-05,
+      "loss": 0.4077,
+      "step": 13051
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931338357585727e-05,
+      "loss": 0.5021,
+      "step": 13052
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931319137256927e-05,
+      "loss": 0.4745,
+      "step": 13053
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931299914247618e-05,
+      "loss": 0.4348,
+      "step": 13054
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931280688557805e-05,
+      "loss": 0.5263,
+      "step": 13055
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993126146018749e-05,
+      "loss": 0.4868,
+      "step": 13056
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931242229136684e-05,
+      "loss": 0.3897,
+      "step": 13057
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931222995405388e-05,
+      "loss": 0.5021,
+      "step": 13058
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9931203758993608e-05,
+      "loss": 0.4959,
+      "step": 13059
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993118451990135e-05,
+      "loss": 0.5279,
+      "step": 13060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9931165278128622e-05,
+      "loss": 0.4806,
+      "step": 13061
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9931146033675427e-05,
+      "loss": 0.5408,
+      "step": 13062
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9931126786541765e-05,
+      "loss": 0.4401,
+      "step": 13063
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9931107536727644e-05,
+      "loss": 0.5173,
+      "step": 13064
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9931088284233076e-05,
+      "loss": 0.4796,
+      "step": 13065
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993106902905806e-05,
+      "loss": 0.4914,
+      "step": 13066
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9931049771202597e-05,
+      "loss": 0.499,
+      "step": 13067
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9931030510666702e-05,
+      "loss": 0.4075,
+      "step": 13068
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9931011247450375e-05,
+      "loss": 0.4473,
+      "step": 13069
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993099198155362e-05,
+      "loss": 0.4941,
+      "step": 13070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930972712976444e-05,
+      "loss": 0.5124,
+      "step": 13071
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930953441718853e-05,
+      "loss": 0.4857,
+      "step": 13072
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993093416778085e-05,
+      "loss": 0.4493,
+      "step": 13073
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930914891162443e-05,
+      "loss": 0.4292,
+      "step": 13074
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930895611863633e-05,
+      "loss": 0.4679,
+      "step": 13075
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930876329884432e-05,
+      "loss": 0.5847,
+      "step": 13076
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993085704522484e-05,
+      "loss": 0.5488,
+      "step": 13077
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993083775788486e-05,
+      "loss": 0.4417,
+      "step": 13078
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.99308184678645e-05,
+      "loss": 0.5673,
+      "step": 13079
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993079917516377e-05,
+      "loss": 0.5235,
+      "step": 13080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930779879782668e-05,
+      "loss": 0.5166,
+      "step": 13081
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930760581721204e-05,
+      "loss": 0.4685,
+      "step": 13082
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930741280979383e-05,
+      "loss": 0.4433,
+      "step": 13083
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930721977557204e-05,
+      "loss": 0.5016,
+      "step": 13084
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930702671454677e-05,
+      "loss": 0.4258,
+      "step": 13085
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993068336267181e-05,
+      "loss": 0.5153,
+      "step": 13086
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930664051208604e-05,
+      "loss": 0.4436,
+      "step": 13087
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930644737065064e-05,
+      "loss": 0.4856,
+      "step": 13088
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930625420241197e-05,
+      "loss": 0.5369,
+      "step": 13089
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993060610073701e-05,
+      "loss": 0.4702,
+      "step": 13090
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930586778552502e-05,
+      "loss": 0.5729,
+      "step": 13091
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930567453687687e-05,
+      "loss": 0.4698,
+      "step": 13092
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930548126142565e-05,
+      "loss": 0.4974,
+      "step": 13093
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993052879591714e-05,
+      "loss": 0.4245,
+      "step": 13094
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930509463011417e-05,
+      "loss": 0.4976,
+      "step": 13095
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930490127425405e-05,
+      "loss": 0.4922,
+      "step": 13096
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930470789159106e-05,
+      "loss": 0.4788,
+      "step": 13097
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930451448212527e-05,
+      "loss": 0.5402,
+      "step": 13098
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930432104585674e-05,
+      "loss": 0.4676,
+      "step": 13099
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993041275827855e-05,
+      "loss": 0.427,
+      "step": 13100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993039340929116e-05,
+      "loss": 0.4518,
+      "step": 13101
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930374057623513e-05,
+      "loss": 0.4654,
+      "step": 13102
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993035470327561e-05,
+      "loss": 0.4329,
+      "step": 13103
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993033534624746e-05,
+      "loss": 0.5248,
+      "step": 13104
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930315986539063e-05,
+      "loss": 0.5256,
+      "step": 13105
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930296624150427e-05,
+      "loss": 0.4021,
+      "step": 13106
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993027725908156e-05,
+      "loss": 0.49,
+      "step": 13107
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930257891332462e-05,
+      "loss": 0.4854,
+      "step": 13108
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930238520903143e-05,
+      "loss": 0.4906,
+      "step": 13109
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930219147793605e-05,
+      "loss": 0.5437,
+      "step": 13110
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930199772003858e-05,
+      "loss": 0.5098,
+      "step": 13111
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.99301803935339e-05,
+      "loss": 0.4368,
+      "step": 13112
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993016101238374e-05,
+      "loss": 0.4352,
+      "step": 13113
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930141628553385e-05,
+      "loss": 0.4346,
+      "step": 13114
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930122242042837e-05,
+      "loss": 0.4665,
+      "step": 13115
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930102852852104e-05,
+      "loss": 0.4938,
+      "step": 13116
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993008346098119e-05,
+      "loss": 0.4881,
+      "step": 13117
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930064066430096e-05,
+      "loss": 0.5219,
+      "step": 13118
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930044669198835e-05,
+      "loss": 0.4591,
+      "step": 13119
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993002526928741e-05,
+      "loss": 0.5141,
+      "step": 13120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930005866695822e-05,
+      "loss": 0.4201,
+      "step": 13121
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929986461424077e-05,
+      "loss": 0.3996,
+      "step": 13122
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929967053472187e-05,
+      "loss": 0.4843,
+      "step": 13123
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929947642840147e-05,
+      "loss": 0.5199,
+      "step": 13124
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929928229527972e-05,
+      "loss": 0.5874,
+      "step": 13125
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929908813535663e-05,
+      "loss": 0.5773,
+      "step": 13126
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929889394863223e-05,
+      "loss": 0.4518,
+      "step": 13127
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.992986997351066e-05,
+      "loss": 0.4953,
+      "step": 13128
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.992985054947798e-05,
+      "loss": 0.4981,
+      "step": 13129
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929831122765186e-05,
+      "loss": 0.5128,
+      "step": 13130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929811693372284e-05,
+      "loss": 0.5043,
+      "step": 13131
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.992979226129928e-05,
+      "loss": 0.4547,
+      "step": 13132
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.992977282654618e-05,
+      "loss": 0.4673,
+      "step": 13133
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929753389112985e-05,
+      "loss": 0.4955,
+      "step": 13134
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929733948999702e-05,
+      "loss": 0.4211,
+      "step": 13135
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929714506206342e-05,
+      "loss": 0.5123,
+      "step": 13136
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929695060732906e-05,
+      "loss": 0.4418,
+      "step": 13137
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929675612579396e-05,
+      "loss": 0.4753,
+      "step": 13138
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.992965616174582e-05,
+      "loss": 0.482,
+      "step": 13139
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929636708232184e-05,
+      "loss": 0.5702,
+      "step": 13140
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929617252038492e-05,
+      "loss": 0.4907,
+      "step": 13141
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.992959779316475e-05,
+      "loss": 0.5696,
+      "step": 13142
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929578331610962e-05,
+      "loss": 0.5346,
+      "step": 13143
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929558867377135e-05,
+      "loss": 0.4615,
+      "step": 13144
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929539400463275e-05,
+      "loss": 0.4837,
+      "step": 13145
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929519930869387e-05,
+      "loss": 0.4698,
+      "step": 13146
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929500458595475e-05,
+      "loss": 0.3817,
+      "step": 13147
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.992948098364154e-05,
+      "loss": 0.4834,
+      "step": 13148
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929461506007596e-05,
+      "loss": 0.5904,
+      "step": 13149
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.992944202569364e-05,
+      "loss": 0.4542,
+      "step": 13150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929422542699687e-05,
+      "loss": 0.4706,
+      "step": 13151
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929403057025732e-05,
+      "loss": 0.5171,
+      "step": 13152
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929383568671788e-05,
+      "loss": 0.4549,
+      "step": 13153
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929364077637855e-05,
+      "loss": 0.4117,
+      "step": 13154
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929344583923937e-05,
+      "loss": 0.5463,
+      "step": 13155
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929325087530044e-05,
+      "loss": 0.4018,
+      "step": 13156
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929305588456183e-05,
+      "loss": 0.4709,
+      "step": 13157
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929286086702354e-05,
+      "loss": 0.4414,
+      "step": 13158
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929266582268564e-05,
+      "loss": 0.483,
+      "step": 13159
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9929247075154822e-05,
+      "loss": 0.5266,
+      "step": 13160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9929227565361125e-05,
+      "loss": 0.4246,
+      "step": 13161
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9929208052887484e-05,
+      "loss": 0.4477,
+      "step": 13162
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9929188537733906e-05,
+      "loss": 0.4677,
+      "step": 13163
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9929169019900393e-05,
+      "loss": 0.4933,
+      "step": 13164
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992914949938695e-05,
+      "loss": 0.537,
+      "step": 13165
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992912997619358e-05,
+      "loss": 0.4521,
+      "step": 13166
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.99291104503203e-05,
+      "loss": 0.4644,
+      "step": 13167
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9929090921767096e-05,
+      "loss": 0.4264,
+      "step": 13168
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992907139053399e-05,
+      "loss": 0.4963,
+      "step": 13169
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992905185662098e-05,
+      "loss": 0.4734,
+      "step": 13170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992903232002807e-05,
+      "loss": 0.4938,
+      "step": 13171
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9929012780755273e-05,
+      "loss": 0.4474,
+      "step": 13172
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992899323880259e-05,
+      "loss": 0.4134,
+      "step": 13173
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928973694170017e-05,
+      "loss": 0.5653,
+      "step": 13174
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928954146857572e-05,
+      "loss": 0.4245,
+      "step": 13175
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928934596865257e-05,
+      "loss": 0.4749,
+      "step": 13176
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928915044193076e-05,
+      "loss": 0.4941,
+      "step": 13177
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928895488841035e-05,
+      "loss": 0.5273,
+      "step": 13178
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928875930809134e-05,
+      "loss": 0.5676,
+      "step": 13179
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928856370097387e-05,
+      "loss": 0.5418,
+      "step": 13180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928836806705797e-05,
+      "loss": 0.5343,
+      "step": 13181
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928817240634364e-05,
+      "loss": 0.4205,
+      "step": 13182
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928797671883098e-05,
+      "loss": 0.5343,
+      "step": 13183
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928778100452003e-05,
+      "loss": 0.4888,
+      "step": 13184
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928758526341086e-05,
+      "loss": 0.4562,
+      "step": 13185
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992873894955035e-05,
+      "loss": 0.4665,
+      "step": 13186
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928719370079797e-05,
+      "loss": 0.5051,
+      "step": 13187
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992869978792944e-05,
+      "loss": 0.4096,
+      "step": 13188
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992868020309928e-05,
+      "loss": 0.4849,
+      "step": 13189
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992866061558932e-05,
+      "loss": 0.5653,
+      "step": 13190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928641025399574e-05,
+      "loss": 0.5604,
+      "step": 13191
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928621432530038e-05,
+      "loss": 0.4798,
+      "step": 13192
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992860183698072e-05,
+      "loss": 0.4612,
+      "step": 13193
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928582238751627e-05,
+      "loss": 0.4395,
+      "step": 13194
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928562637842762e-05,
+      "loss": 0.486,
+      "step": 13195
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928543034254133e-05,
+      "loss": 0.4967,
+      "step": 13196
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928523427985745e-05,
+      "loss": 0.4692,
+      "step": 13197
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928503819037603e-05,
+      "loss": 0.5095,
+      "step": 13198
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928484207409705e-05,
+      "loss": 0.508,
+      "step": 13199
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928464593102067e-05,
+      "loss": 0.4405,
+      "step": 13200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928444976114688e-05,
+      "loss": 0.4413,
+      "step": 13201
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928425356447577e-05,
+      "loss": 0.474,
+      "step": 13202
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992840573410074e-05,
+      "loss": 0.4805,
+      "step": 13203
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928386109074176e-05,
+      "loss": 0.459,
+      "step": 13204
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928366481367896e-05,
+      "loss": 0.5167,
+      "step": 13205
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928346850981902e-05,
+      "loss": 0.4228,
+      "step": 13206
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928327217916204e-05,
+      "loss": 0.4354,
+      "step": 13207
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.99283075821708e-05,
+      "loss": 0.4268,
+      "step": 13208
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.99282879437457e-05,
+      "loss": 0.4891,
+      "step": 13209
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992826830264091e-05,
+      "loss": 0.5862,
+      "step": 13210
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928248658856434e-05,
+      "loss": 0.4963,
+      "step": 13211
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928229012392278e-05,
+      "loss": 0.5408,
+      "step": 13212
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928209363248444e-05,
+      "loss": 0.4456,
+      "step": 13213
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928189711424944e-05,
+      "loss": 0.4439,
+      "step": 13214
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928170056921778e-05,
+      "loss": 0.4805,
+      "step": 13215
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928150399738948e-05,
+      "loss": 0.4503,
+      "step": 13216
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928130739876468e-05,
+      "loss": 0.4928,
+      "step": 13217
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992811107733434e-05,
+      "loss": 0.4482,
+      "step": 13218
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928091412112566e-05,
+      "loss": 0.4939,
+      "step": 13219
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928071744211154e-05,
+      "loss": 0.4907,
+      "step": 13220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992805207363011e-05,
+      "loss": 0.5731,
+      "step": 13221
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928032400369435e-05,
+      "loss": 0.4614,
+      "step": 13222
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928012724429142e-05,
+      "loss": 0.4677,
+      "step": 13223
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992799304580923e-05,
+      "loss": 0.4821,
+      "step": 13224
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927973364509708e-05,
+      "loss": 0.419,
+      "step": 13225
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927953680530577e-05,
+      "loss": 0.465,
+      "step": 13226
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927933993871848e-05,
+      "loss": 0.4833,
+      "step": 13227
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992791430453352e-05,
+      "loss": 0.4119,
+      "step": 13228
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927894612515602e-05,
+      "loss": 0.4172,
+      "step": 13229
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.99278749178181e-05,
+      "loss": 0.4397,
+      "step": 13230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927855220441015e-05,
+      "loss": 0.5058,
+      "step": 13231
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992783552038436e-05,
+      "loss": 0.4597,
+      "step": 13232
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927815817648134e-05,
+      "loss": 0.4504,
+      "step": 13233
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927796112232345e-05,
+      "loss": 0.3863,
+      "step": 13234
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927776404136997e-05,
+      "loss": 0.4089,
+      "step": 13235
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927756693362094e-05,
+      "loss": 0.4461,
+      "step": 13236
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927736979907647e-05,
+      "loss": 0.4448,
+      "step": 13237
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927717263773655e-05,
+      "loss": 0.4619,
+      "step": 13238
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992769754496012e-05,
+      "loss": 0.4659,
+      "step": 13239
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992767782346706e-05,
+      "loss": 0.5093,
+      "step": 13240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992765809929447e-05,
+      "loss": 0.5186,
+      "step": 13241
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927638372442365e-05,
+      "loss": 0.5501,
+      "step": 13242
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927618642910738e-05,
+      "loss": 0.5141,
+      "step": 13243
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927598910699603e-05,
+      "loss": 0.4686,
+      "step": 13244
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992757917580896e-05,
+      "loss": 0.5331,
+      "step": 13245
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927559438238818e-05,
+      "loss": 0.4599,
+      "step": 13246
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992753969798918e-05,
+      "loss": 0.4437,
+      "step": 13247
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927519955060054e-05,
+      "loss": 0.5094,
+      "step": 13248
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927500209451446e-05,
+      "loss": 0.4835,
+      "step": 13249
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992748046116336e-05,
+      "loss": 0.4746,
+      "step": 13250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927460710195797e-05,
+      "loss": 0.478,
+      "step": 13251
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927440956548765e-05,
+      "loss": 0.5036,
+      "step": 13252
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992742120022227e-05,
+      "loss": 0.4841,
+      "step": 13253
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927401441216326e-05,
+      "loss": 0.4569,
+      "step": 13254
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992738167953092e-05,
+      "loss": 0.4894,
+      "step": 13255
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927361915166074e-05,
+      "loss": 0.4848,
+      "step": 13256
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927342148121783e-05,
+      "loss": 0.4396,
+      "step": 13257
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927322378398056e-05,
+      "loss": 0.4516,
+      "step": 13258
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.99273026059949e-05,
+      "loss": 0.4797,
+      "step": 13259
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927282830912317e-05,
+      "loss": 0.5896,
+      "step": 13260
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927263053150315e-05,
+      "loss": 0.5584,
+      "step": 13261
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927243272708897e-05,
+      "loss": 0.5004,
+      "step": 13262
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992722348958807e-05,
+      "loss": 0.5148,
+      "step": 13263
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992720370378784e-05,
+      "loss": 0.4792,
+      "step": 13264
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927183915308212e-05,
+      "loss": 0.5038,
+      "step": 13265
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927164124149188e-05,
+      "loss": 0.5469,
+      "step": 13266
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992714433031078e-05,
+      "loss": 0.4748,
+      "step": 13267
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927124533792987e-05,
+      "loss": 0.4754,
+      "step": 13268
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927104734595817e-05,
+      "loss": 0.3678,
+      "step": 13269
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927084932719275e-05,
+      "loss": 0.4781,
+      "step": 13270
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927065128163365e-05,
+      "loss": 0.5142,
+      "step": 13271
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9927045320928092e-05,
+      "loss": 0.465,
+      "step": 13272
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992702551101347e-05,
+      "loss": 0.504,
+      "step": 13273
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992700569841949e-05,
+      "loss": 0.5289,
+      "step": 13274
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992698588314617e-05,
+      "loss": 0.4693,
+      "step": 13275
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926966065193508e-05,
+      "loss": 0.5064,
+      "step": 13276
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926946244561512e-05,
+      "loss": 0.4639,
+      "step": 13277
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926926421250184e-05,
+      "loss": 0.4862,
+      "step": 13278
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926906595259536e-05,
+      "loss": 0.5145,
+      "step": 13279
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992688676658957e-05,
+      "loss": 0.492,
+      "step": 13280
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926866935240286e-05,
+      "loss": 0.5287,
+      "step": 13281
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926847101211698e-05,
+      "loss": 0.4964,
+      "step": 13282
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926827264503806e-05,
+      "loss": 0.4367,
+      "step": 13283
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926807425116617e-05,
+      "loss": 0.4304,
+      "step": 13284
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926787583050137e-05,
+      "loss": 0.4251,
+      "step": 13285
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926767738304373e-05,
+      "loss": 0.445,
+      "step": 13286
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926747890879322e-05,
+      "loss": 0.5279,
+      "step": 13287
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926728040775e-05,
+      "loss": 0.4561,
+      "step": 13288
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926708187991406e-05,
+      "loss": 0.4918,
+      "step": 13289
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926688332528547e-05,
+      "loss": 0.459,
+      "step": 13290
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992666847438643e-05,
+      "loss": 0.5213,
+      "step": 13291
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926648613565057e-05,
+      "loss": 0.5261,
+      "step": 13292
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926628750064436e-05,
+      "loss": 0.4556,
+      "step": 13293
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992660888388457e-05,
+      "loss": 0.4771,
+      "step": 13294
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926589015025466e-05,
+      "loss": 0.4763,
+      "step": 13295
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926569143487132e-05,
+      "loss": 0.4684,
+      "step": 13296
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926549269269572e-05,
+      "loss": 0.4321,
+      "step": 13297
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926529392372786e-05,
+      "loss": 0.5141,
+      "step": 13298
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992650951279678e-05,
+      "loss": 0.4597,
+      "step": 13299
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992648963054157e-05,
+      "loss": 0.4631,
+      "step": 13300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992646974560715e-05,
+      "loss": 0.4032,
+      "step": 13301
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992644985799353e-05,
+      "loss": 0.4962,
+      "step": 13302
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926429967700714e-05,
+      "loss": 0.413,
+      "step": 13303
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926410074728713e-05,
+      "loss": 0.5869,
+      "step": 13304
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926390179077523e-05,
+      "loss": 0.4779,
+      "step": 13305
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926370280747153e-05,
+      "loss": 0.4795,
+      "step": 13306
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926350379737612e-05,
+      "loss": 0.4849,
+      "step": 13307
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.99263304760489e-05,
+      "loss": 0.4998,
+      "step": 13308
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926310569681028e-05,
+      "loss": 0.4793,
+      "step": 13309
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926290660633998e-05,
+      "loss": 0.5271,
+      "step": 13310
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926270748907813e-05,
+      "loss": 0.5202,
+      "step": 13311
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926250834502486e-05,
+      "loss": 0.4232,
+      "step": 13312
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926230917418012e-05,
+      "loss": 0.4323,
+      "step": 13313
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926210997654406e-05,
+      "loss": 0.4337,
+      "step": 13314
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926191075211668e-05,
+      "loss": 0.5358,
+      "step": 13315
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926171150089803e-05,
+      "loss": 0.5712,
+      "step": 13316
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992615122228882e-05,
+      "loss": 0.4052,
+      "step": 13317
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992613129180872e-05,
+      "loss": 0.3986,
+      "step": 13318
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992611135864951e-05,
+      "loss": 0.47,
+      "step": 13319
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.99260914228112e-05,
+      "loss": 0.6248,
+      "step": 13320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992607148429379e-05,
+      "loss": 0.4675,
+      "step": 13321
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992605154309729e-05,
+      "loss": 0.5046,
+      "step": 13322
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.99260315992217e-05,
+      "loss": 0.4504,
+      "step": 13323
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9926011652667023e-05,
+      "loss": 0.5338,
+      "step": 13324
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925991703433274e-05,
+      "loss": 0.4611,
+      "step": 13325
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992597175152045e-05,
+      "loss": 0.4679,
+      "step": 13326
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925951796928564e-05,
+      "loss": 0.5561,
+      "step": 13327
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925931839657614e-05,
+      "loss": 0.4146,
+      "step": 13328
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992591187970761e-05,
+      "loss": 0.4057,
+      "step": 13329
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925891917078558e-05,
+      "loss": 0.4714,
+      "step": 13330
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925871951770457e-05,
+      "loss": 0.4428,
+      "step": 13331
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992585198378332e-05,
+      "loss": 0.4473,
+      "step": 13332
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925832013117148e-05,
+      "loss": 0.503,
+      "step": 13333
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992581203977195e-05,
+      "loss": 0.4446,
+      "step": 13334
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925792063747726e-05,
+      "loss": 0.4177,
+      "step": 13335
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925772085044485e-05,
+      "loss": 0.4228,
+      "step": 13336
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925752103662234e-05,
+      "loss": 0.5546,
+      "step": 13337
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992573211960097e-05,
+      "loss": 0.4809,
+      "step": 13338
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992571213286071e-05,
+      "loss": 0.4045,
+      "step": 13339
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925692143441455e-05,
+      "loss": 0.4664,
+      "step": 13340
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925672151343203e-05,
+      "loss": 0.4904,
+      "step": 13341
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992565215656597e-05,
+      "loss": 0.4275,
+      "step": 13342
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925632159109754e-05,
+      "loss": 0.5357,
+      "step": 13343
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925612158974566e-05,
+      "loss": 0.519,
+      "step": 13344
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925592156160413e-05,
+      "loss": 0.4242,
+      "step": 13345
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992557215066729e-05,
+      "loss": 0.5181,
+      "step": 13346
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992555214249521e-05,
+      "loss": 0.476,
+      "step": 13347
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992553213164418e-05,
+      "loss": 0.533,
+      "step": 13348
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925512118114198e-05,
+      "loss": 0.4937,
+      "step": 13349
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925492101905276e-05,
+      "loss": 0.4774,
+      "step": 13350
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.992547208301742e-05,
+      "loss": 0.48,
+      "step": 13351
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925452061450627e-05,
+      "loss": 0.4542,
+      "step": 13352
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925432037204914e-05,
+      "loss": 0.4556,
+      "step": 13353
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925412010280273e-05,
+      "loss": 0.4472,
+      "step": 13354
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925391980676724e-05,
+      "loss": 0.5032,
+      "step": 13355
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9925371948394264e-05,
+      "loss": 0.4641,
+      "step": 13356
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9925351913432897e-05,
+      "loss": 0.5414,
+      "step": 13357
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992533187579263e-05,
+      "loss": 0.4782,
+      "step": 13358
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9925311835473476e-05,
+      "loss": 0.4373,
+      "step": 13359
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992529179247543e-05,
+      "loss": 0.4554,
+      "step": 13360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.99252717467985e-05,
+      "loss": 0.4419,
+      "step": 13361
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9925251698442697e-05,
+      "loss": 0.4909,
+      "step": 13362
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992523164740802e-05,
+      "loss": 0.4246,
+      "step": 13363
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9925211593694473e-05,
+      "loss": 0.5038,
+      "step": 13364
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9925191537302068e-05,
+      "loss": 0.4467,
+      "step": 13365
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992517147823081e-05,
+      "loss": 0.5698,
+      "step": 13366
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.99251514164807e-05,
+      "loss": 0.519,
+      "step": 13367
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992513135205174e-05,
+      "loss": 0.4666,
+      "step": 13368
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9925111284943947e-05,
+      "loss": 0.5044,
+      "step": 13369
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992509121515732e-05,
+      "loss": 0.4817,
+      "step": 13370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9925071142691862e-05,
+      "loss": 0.4105,
+      "step": 13371
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992505106754758e-05,
+      "loss": 0.4608,
+      "step": 13372
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9925030989724482e-05,
+      "loss": 0.4651,
+      "step": 13373
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9925010909222574e-05,
+      "loss": 0.5136,
+      "step": 13374
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924990826041854e-05,
+      "loss": 0.4475,
+      "step": 13375
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924970740182337e-05,
+      "loss": 0.4626,
+      "step": 13376
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924950651644025e-05,
+      "loss": 0.4704,
+      "step": 13377
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924930560426917e-05,
+      "loss": 0.4736,
+      "step": 13378
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924910466531026e-05,
+      "loss": 0.5937,
+      "step": 13379
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924890369956357e-05,
+      "loss": 0.4366,
+      "step": 13380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992487027070291e-05,
+      "loss": 0.4419,
+      "step": 13381
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.99248501687707e-05,
+      "loss": 0.4561,
+      "step": 13382
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924830064159725e-05,
+      "loss": 0.5819,
+      "step": 13383
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924809956869987e-05,
+      "loss": 0.5248,
+      "step": 13384
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.99247898469015e-05,
+      "loss": 0.4439,
+      "step": 13385
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924769734254266e-05,
+      "loss": 0.3786,
+      "step": 13386
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992474961892829e-05,
+      "loss": 0.5117,
+      "step": 13387
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924729500923577e-05,
+      "loss": 0.5008,
+      "step": 13388
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992470938024013e-05,
+      "loss": 0.4701,
+      "step": 13389
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924689256877963e-05,
+      "loss": 0.5096,
+      "step": 13390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924669130837077e-05,
+      "loss": 0.4966,
+      "step": 13391
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924649002117474e-05,
+      "loss": 0.4507,
+      "step": 13392
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924628870719162e-05,
+      "loss": 0.4529,
+      "step": 13393
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924608736642143e-05,
+      "loss": 0.4524,
+      "step": 13394
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992458859988643e-05,
+      "loss": 0.5227,
+      "step": 13395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924568460452024e-05,
+      "loss": 0.5044,
+      "step": 13396
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924548318338926e-05,
+      "loss": 0.4847,
+      "step": 13397
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992452817354715e-05,
+      "loss": 0.4351,
+      "step": 13398
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924508026076696e-05,
+      "loss": 0.4768,
+      "step": 13399
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924487875927572e-05,
+      "loss": 0.494,
+      "step": 13400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992446772309978e-05,
+      "loss": 0.4679,
+      "step": 13401
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992444756759333e-05,
+      "loss": 0.5147,
+      "step": 13402
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924427409408228e-05,
+      "loss": 0.476,
+      "step": 13403
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924407248544472e-05,
+      "loss": 0.4534,
+      "step": 13404
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924387085002073e-05,
+      "loss": 0.4356,
+      "step": 13405
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924366918781035e-05,
+      "loss": 0.4463,
+      "step": 13406
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924346749881366e-05,
+      "loss": 0.4311,
+      "step": 13407
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924326578303067e-05,
+      "loss": 0.4773,
+      "step": 13408
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992430640404615e-05,
+      "loss": 0.5309,
+      "step": 13409
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992428622711061e-05,
+      "loss": 0.4678,
+      "step": 13410
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924266047496466e-05,
+      "loss": 0.4853,
+      "step": 13411
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992424586520371e-05,
+      "loss": 0.4676,
+      "step": 13412
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924225680232356e-05,
+      "loss": 0.5254,
+      "step": 13413
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992420549258241e-05,
+      "loss": 0.4637,
+      "step": 13414
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924185302253872e-05,
+      "loss": 0.5172,
+      "step": 13415
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924165109246748e-05,
+      "loss": 0.4519,
+      "step": 13416
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924144913561048e-05,
+      "loss": 0.4153,
+      "step": 13417
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924124715196772e-05,
+      "loss": 0.4724,
+      "step": 13418
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992410451415393e-05,
+      "loss": 0.469,
+      "step": 13419
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992408431043253e-05,
+      "loss": 0.5709,
+      "step": 13420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992406410403257e-05,
+      "loss": 0.485,
+      "step": 13421
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924043894954055e-05,
+      "loss": 0.4456,
+      "step": 13422
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9924023683197e-05,
+      "loss": 0.4985,
+      "step": 13423
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.99240034687614e-05,
+      "loss": 0.4983,
+      "step": 13424
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992398325164727e-05,
+      "loss": 0.4051,
+      "step": 13425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992396303185461e-05,
+      "loss": 0.4728,
+      "step": 13426
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992394280938342e-05,
+      "loss": 0.4859,
+      "step": 13427
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923922584233716e-05,
+      "loss": 0.4578,
+      "step": 13428
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923902356405497e-05,
+      "loss": 0.4649,
+      "step": 13429
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992388212589877e-05,
+      "loss": 0.5095,
+      "step": 13430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923861892713543e-05,
+      "loss": 0.548,
+      "step": 13431
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923841656849818e-05,
+      "loss": 0.4402,
+      "step": 13432
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923821418307602e-05,
+      "loss": 0.4652,
+      "step": 13433
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.99238011770869e-05,
+      "loss": 0.4657,
+      "step": 13434
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992378093318772e-05,
+      "loss": 0.5162,
+      "step": 13435
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992376068661006e-05,
+      "loss": 0.4806,
+      "step": 13436
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923740437353936e-05,
+      "loss": 0.5512,
+      "step": 13437
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923720185419346e-05,
+      "loss": 0.4795,
+      "step": 13438
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.99236999308063e-05,
+      "loss": 0.4841,
+      "step": 13439
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.99236796735148e-05,
+      "loss": 0.5006,
+      "step": 13440
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923659413544848e-05,
+      "loss": 0.4749,
+      "step": 13441
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923639150896457e-05,
+      "loss": 0.4335,
+      "step": 13442
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992361888556963e-05,
+      "loss": 0.5088,
+      "step": 13443
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992359861756437e-05,
+      "loss": 0.4766,
+      "step": 13444
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923578346880687e-05,
+      "loss": 0.6134,
+      "step": 13445
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923558073518582e-05,
+      "loss": 0.4665,
+      "step": 13446
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923537797478062e-05,
+      "loss": 0.512,
+      "step": 13447
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992351751875913e-05,
+      "loss": 0.4901,
+      "step": 13448
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923497237361798e-05,
+      "loss": 0.4199,
+      "step": 13449
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992347695328607e-05,
+      "loss": 0.4084,
+      "step": 13450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923456666531947e-05,
+      "loss": 0.4756,
+      "step": 13451
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923436377099436e-05,
+      "loss": 0.4682,
+      "step": 13452
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.992341608498854e-05,
+      "loss": 0.499,
+      "step": 13453
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9923395790199272e-05,
+      "loss": 0.4926,
+      "step": 13454
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9923375492731633e-05,
+      "loss": 0.4564,
+      "step": 13455
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9923355192585626e-05,
+      "loss": 0.5792,
+      "step": 13456
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9923334889761263e-05,
+      "loss": 0.5102,
+      "step": 13457
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9923314584258544e-05,
+      "loss": 0.4659,
+      "step": 13458
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9923294276077473e-05,
+      "loss": 0.5303,
+      "step": 13459
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992327396521806e-05,
+      "loss": 0.5648,
+      "step": 13460
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9923253651680313e-05,
+      "loss": 0.4164,
+      "step": 13461
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992323333546423e-05,
+      "loss": 0.4873,
+      "step": 13462
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992321301656982e-05,
+      "loss": 0.4949,
+      "step": 13463
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9923192694997087e-05,
+      "loss": 0.4162,
+      "step": 13464
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992317237074604e-05,
+      "loss": 0.435,
+      "step": 13465
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992315204381668e-05,
+      "loss": 0.4701,
+      "step": 13466
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992313171420902e-05,
+      "loss": 0.5395,
+      "step": 13467
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9923111381923052e-05,
+      "loss": 0.4888,
+      "step": 13468
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.99230910469588e-05,
+      "loss": 0.4615,
+      "step": 13469
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992307070931625e-05,
+      "loss": 0.544,
+      "step": 13470
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992305036899542e-05,
+      "loss": 0.5586,
+      "step": 13471
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9923030025996316e-05,
+      "loss": 0.4875,
+      "step": 13472
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9923009680318937e-05,
+      "loss": 0.397,
+      "step": 13473
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922989331963293e-05,
+      "loss": 0.4714,
+      "step": 13474
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922968980929386e-05,
+      "loss": 0.4411,
+      "step": 13475
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992294862721722e-05,
+      "loss": 0.4981,
+      "step": 13476
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992292827082681e-05,
+      "loss": 0.4821,
+      "step": 13477
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992290791175815e-05,
+      "loss": 0.4364,
+      "step": 13478
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922887550011256e-05,
+      "loss": 0.4157,
+      "step": 13479
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922867185586123e-05,
+      "loss": 0.4556,
+      "step": 13480
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922846818482764e-05,
+      "loss": 0.6106,
+      "step": 13481
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922826448701185e-05,
+      "loss": 0.5009,
+      "step": 13482
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922806076241386e-05,
+      "loss": 0.4568,
+      "step": 13483
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922785701103374e-05,
+      "loss": 0.4823,
+      "step": 13484
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992276532328716e-05,
+      "loss": 0.4602,
+      "step": 13485
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992274494279274e-05,
+      "loss": 0.4445,
+      "step": 13486
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992272455962013e-05,
+      "loss": 0.4498,
+      "step": 13487
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922704173769327e-05,
+      "loss": 0.5535,
+      "step": 13488
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992268378524034e-05,
+      "loss": 0.5031,
+      "step": 13489
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922663394033175e-05,
+      "loss": 0.5067,
+      "step": 13490
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922643000147833e-05,
+      "loss": 0.4165,
+      "step": 13491
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992262260358433e-05,
+      "loss": 0.4072,
+      "step": 13492
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992260220434266e-05,
+      "loss": 0.492,
+      "step": 13493
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922581802422835e-05,
+      "loss": 0.4532,
+      "step": 13494
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992256139782486e-05,
+      "loss": 0.6152,
+      "step": 13495
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992254099054874e-05,
+      "loss": 0.458,
+      "step": 13496
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922520580594474e-05,
+      "loss": 0.5027,
+      "step": 13497
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922500167962078e-05,
+      "loss": 0.4391,
+      "step": 13498
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922479752651553e-05,
+      "loss": 0.4327,
+      "step": 13499
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922459334662903e-05,
+      "loss": 0.5192,
+      "step": 13500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922438913996135e-05,
+      "loss": 0.4579,
+      "step": 13501
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922418490651255e-05,
+      "loss": 0.4848,
+      "step": 13502
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922398064628267e-05,
+      "loss": 0.4575,
+      "step": 13503
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922377635927174e-05,
+      "loss": 0.4951,
+      "step": 13504
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992235720454799e-05,
+      "loss": 0.4394,
+      "step": 13505
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922336770490712e-05,
+      "loss": 0.5289,
+      "step": 13506
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922316333755353e-05,
+      "loss": 0.5507,
+      "step": 13507
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922295894341913e-05,
+      "loss": 0.4689,
+      "step": 13508
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.99222754522504e-05,
+      "loss": 0.4374,
+      "step": 13509
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922255007480813e-05,
+      "loss": 0.4891,
+      "step": 13510
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922234560033167e-05,
+      "loss": 0.4908,
+      "step": 13511
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922214109907464e-05,
+      "loss": 0.4517,
+      "step": 13512
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922193657103706e-05,
+      "loss": 0.478,
+      "step": 13513
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922173201621905e-05,
+      "loss": 0.5128,
+      "step": 13514
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992215274346206e-05,
+      "loss": 0.444,
+      "step": 13515
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922132282624186e-05,
+      "loss": 0.5175,
+      "step": 13516
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922111819108277e-05,
+      "loss": 0.4851,
+      "step": 13517
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922091352914342e-05,
+      "loss": 0.5044,
+      "step": 13518
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992207088404239e-05,
+      "loss": 0.4729,
+      "step": 13519
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922050412492425e-05,
+      "loss": 0.4437,
+      "step": 13520
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922029938264453e-05,
+      "loss": 0.4609,
+      "step": 13521
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9922009461358478e-05,
+      "loss": 0.519,
+      "step": 13522
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921988981774504e-05,
+      "loss": 0.5491,
+      "step": 13523
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921968499512543e-05,
+      "loss": 0.5177,
+      "step": 13524
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921948014572593e-05,
+      "loss": 0.4269,
+      "step": 13525
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921927526954664e-05,
+      "loss": 0.57,
+      "step": 13526
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992190703665876e-05,
+      "loss": 0.4651,
+      "step": 13527
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992188654368489e-05,
+      "loss": 0.3926,
+      "step": 13528
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921866048033053e-05,
+      "loss": 0.4793,
+      "step": 13529
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921845549703257e-05,
+      "loss": 0.5644,
+      "step": 13530
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921825048695512e-05,
+      "loss": 0.5006,
+      "step": 13531
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921804545009818e-05,
+      "loss": 0.4194,
+      "step": 13532
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921784038646182e-05,
+      "loss": 0.4214,
+      "step": 13533
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992176352960461e-05,
+      "loss": 0.5585,
+      "step": 13534
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992174301788511e-05,
+      "loss": 0.5316,
+      "step": 13535
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921722503487686e-05,
+      "loss": 0.5207,
+      "step": 13536
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992170198641234e-05,
+      "loss": 0.5481,
+      "step": 13537
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992168146665908e-05,
+      "loss": 0.5024,
+      "step": 13538
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921660944227915e-05,
+      "loss": 0.4721,
+      "step": 13539
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921640419118847e-05,
+      "loss": 0.5138,
+      "step": 13540
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921619891331877e-05,
+      "loss": 0.4993,
+      "step": 13541
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992159936086702e-05,
+      "loss": 0.4991,
+      "step": 13542
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921578827724275e-05,
+      "loss": 0.4885,
+      "step": 13543
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921558291903648e-05,
+      "loss": 0.4118,
+      "step": 13544
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921537753405148e-05,
+      "loss": 0.4393,
+      "step": 13545
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992151721222878e-05,
+      "loss": 0.4658,
+      "step": 13546
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921496668374547e-05,
+      "loss": 0.5374,
+      "step": 13547
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921476121842454e-05,
+      "loss": 0.4857,
+      "step": 13548
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.992145557263251e-05,
+      "loss": 0.4288,
+      "step": 13549
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921435020744716e-05,
+      "loss": 0.4968,
+      "step": 13550
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921414466179084e-05,
+      "loss": 0.4718,
+      "step": 13551
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921393908935616e-05,
+      "loss": 0.3681,
+      "step": 13552
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9921373349014313e-05,
+      "loss": 0.4759,
+      "step": 13553
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921352786415188e-05,
+      "loss": 0.5418,
+      "step": 13554
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921332221138242e-05,
+      "loss": 0.4896,
+      "step": 13555
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921311653183486e-05,
+      "loss": 0.5957,
+      "step": 13556
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992129108255092e-05,
+      "loss": 0.5249,
+      "step": 13557
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921270509240548e-05,
+      "loss": 0.5124,
+      "step": 13558
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921249933252382e-05,
+      "loss": 0.521,
+      "step": 13559
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992122935458642e-05,
+      "loss": 0.4744,
+      "step": 13560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921208773242676e-05,
+      "loss": 0.4491,
+      "step": 13561
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921188189221152e-05,
+      "loss": 0.6599,
+      "step": 13562
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921167602521852e-05,
+      "loss": 0.4874,
+      "step": 13563
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.443,
+      "step": 13564
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921126421089947e-05,
+      "loss": 0.4858,
+      "step": 13565
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921105826357353e-05,
+      "loss": 0.4334,
+      "step": 13566
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921085228947006e-05,
+      "loss": 0.4935,
+      "step": 13567
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921064628858917e-05,
+      "loss": 0.4976,
+      "step": 13568
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9921044026093082e-05,
+      "loss": 0.5254,
+      "step": 13569
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992102342064951e-05,
+      "loss": 0.4083,
+      "step": 13570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992100281252821e-05,
+      "loss": 0.5439,
+      "step": 13571
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920982201729182e-05,
+      "loss": 0.4251,
+      "step": 13572
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920961588252433e-05,
+      "loss": 0.5023,
+      "step": 13573
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920940972097976e-05,
+      "loss": 0.3792,
+      "step": 13574
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992092035326581e-05,
+      "loss": 0.5079,
+      "step": 13575
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920899731755936e-05,
+      "loss": 0.4692,
+      "step": 13576
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992087910756837e-05,
+      "loss": 0.4126,
+      "step": 13577
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992085848070311e-05,
+      "loss": 0.448,
+      "step": 13578
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920837851160165e-05,
+      "loss": 0.5145,
+      "step": 13579
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992081721893954e-05,
+      "loss": 0.5076,
+      "step": 13580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920796584041238e-05,
+      "loss": 0.4901,
+      "step": 13581
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920775946465266e-05,
+      "loss": 0.4973,
+      "step": 13582
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992075530621163e-05,
+      "loss": 0.4343,
+      "step": 13583
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992073466328034e-05,
+      "loss": 0.4662,
+      "step": 13584
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920714017671393e-05,
+      "loss": 0.5104,
+      "step": 13585
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.99206933693848e-05,
+      "loss": 0.4497,
+      "step": 13586
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920672718420566e-05,
+      "loss": 0.502,
+      "step": 13587
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920652064778697e-05,
+      "loss": 0.5388,
+      "step": 13588
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920631408459198e-05,
+      "loss": 0.398,
+      "step": 13589
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992061074946207e-05,
+      "loss": 0.5367,
+      "step": 13590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920590087787328e-05,
+      "loss": 0.4638,
+      "step": 13591
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992056942343497e-05,
+      "loss": 0.4921,
+      "step": 13592
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920548756405004e-05,
+      "loss": 0.5125,
+      "step": 13593
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920528086697434e-05,
+      "loss": 0.4732,
+      "step": 13594
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992050741431227e-05,
+      "loss": 0.4761,
+      "step": 13595
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920486739249514e-05,
+      "loss": 0.5014,
+      "step": 13596
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992046606150917e-05,
+      "loss": 0.5386,
+      "step": 13597
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992044538109125e-05,
+      "loss": 0.4984,
+      "step": 13598
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992042469799575e-05,
+      "loss": 0.4195,
+      "step": 13599
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920404012222686e-05,
+      "loss": 0.5527,
+      "step": 13600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920383323772056e-05,
+      "loss": 0.4236,
+      "step": 13601
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992036263264387e-05,
+      "loss": 0.5067,
+      "step": 13602
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992034193883813e-05,
+      "loss": 0.503,
+      "step": 13603
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920321242354843e-05,
+      "loss": 0.4989,
+      "step": 13604
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920300543194016e-05,
+      "loss": 0.4505,
+      "step": 13605
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920279841355654e-05,
+      "loss": 0.4692,
+      "step": 13606
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920259136839763e-05,
+      "loss": 0.4167,
+      "step": 13607
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920238429646343e-05,
+      "loss": 0.4136,
+      "step": 13608
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920217719775404e-05,
+      "loss": 0.4852,
+      "step": 13609
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920197007226958e-05,
+      "loss": 0.4907,
+      "step": 13610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920176292001002e-05,
+      "loss": 0.5471,
+      "step": 13611
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920155574097545e-05,
+      "loss": 0.4464,
+      "step": 13612
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920134853516587e-05,
+      "loss": 0.4698,
+      "step": 13613
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920114130258144e-05,
+      "loss": 0.4725,
+      "step": 13614
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920093404322212e-05,
+      "loss": 0.4483,
+      "step": 13615
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.99200726757088e-05,
+      "loss": 0.5098,
+      "step": 13616
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920051944417916e-05,
+      "loss": 0.4353,
+      "step": 13617
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9920031210449565e-05,
+      "loss": 0.5012,
+      "step": 13618
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.992001047380375e-05,
+      "loss": 0.434,
+      "step": 13619
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.991998973448048e-05,
+      "loss": 0.4391,
+      "step": 13620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.991996899247975e-05,
+      "loss": 0.4892,
+      "step": 13621
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919948247801583e-05,
+      "loss": 0.4601,
+      "step": 13622
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.991992750044597e-05,
+      "loss": 0.5369,
+      "step": 13623
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919906750412927e-05,
+      "loss": 0.4688,
+      "step": 13624
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919885997702454e-05,
+      "loss": 0.4757,
+      "step": 13625
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919865242314554e-05,
+      "loss": 0.5362,
+      "step": 13626
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919844484249236e-05,
+      "loss": 0.3966,
+      "step": 13627
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.991982372350651e-05,
+      "loss": 0.4065,
+      "step": 13628
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.991980296008637e-05,
+      "loss": 0.5024,
+      "step": 13629
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919782193988837e-05,
+      "loss": 0.4312,
+      "step": 13630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919761425213902e-05,
+      "loss": 0.4781,
+      "step": 13631
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919740653761582e-05,
+      "loss": 0.4909,
+      "step": 13632
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919719879631875e-05,
+      "loss": 0.4667,
+      "step": 13633
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.991969910282479e-05,
+      "loss": 0.5525,
+      "step": 13634
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.991967832334033e-05,
+      "loss": 0.5298,
+      "step": 13635
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919657541178506e-05,
+      "loss": 0.4837,
+      "step": 13636
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919636756339317e-05,
+      "loss": 0.4287,
+      "step": 13637
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919615968822776e-05,
+      "loss": 0.448,
+      "step": 13638
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.991959517862888e-05,
+      "loss": 0.4855,
+      "step": 13639
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919574385757636e-05,
+      "loss": 0.4875,
+      "step": 13640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919553590209057e-05,
+      "loss": 0.4506,
+      "step": 13641
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919532791983147e-05,
+      "loss": 0.4496,
+      "step": 13642
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919511991079904e-05,
+      "loss": 0.5057,
+      "step": 13643
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.991949118749934e-05,
+      "loss": 0.6034,
+      "step": 13644
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919470381241458e-05,
+      "loss": 0.5823,
+      "step": 13645
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919449572306263e-05,
+      "loss": 0.4932,
+      "step": 13646
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919428760693765e-05,
+      "loss": 0.4925,
+      "step": 13647
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919407946403966e-05,
+      "loss": 0.4676,
+      "step": 13648
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919387129436872e-05,
+      "loss": 0.4604,
+      "step": 13649
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.991936630979249e-05,
+      "loss": 0.4988,
+      "step": 13650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9919345487470826e-05,
+      "loss": 0.4096,
+      "step": 13651
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991932466247188e-05,
+      "loss": 0.4051,
+      "step": 13652
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9919303834795665e-05,
+      "loss": 0.4892,
+      "step": 13653
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9919283004442184e-05,
+      "loss": 0.4913,
+      "step": 13654
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991926217141144e-05,
+      "loss": 0.5373,
+      "step": 13655
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9919241335703443e-05,
+      "loss": 0.5226,
+      "step": 13656
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9919220497318195e-05,
+      "loss": 0.5548,
+      "step": 13657
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.99191996562557e-05,
+      "loss": 0.5427,
+      "step": 13658
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9919178812515972e-05,
+      "loss": 0.4906,
+      "step": 13659
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991915796609901e-05,
+      "loss": 0.5509,
+      "step": 13660
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991913711700482e-05,
+      "loss": 0.5006,
+      "step": 13661
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991911626523341e-05,
+      "loss": 0.5148,
+      "step": 13662
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9919095410784782e-05,
+      "loss": 0.4277,
+      "step": 13663
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9919074553658945e-05,
+      "loss": 0.5053,
+      "step": 13664
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9919053693855904e-05,
+      "loss": 0.4407,
+      "step": 13665
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9919032831375663e-05,
+      "loss": 0.4196,
+      "step": 13666
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991901196621823e-05,
+      "loss": 0.5537,
+      "step": 13667
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918991098383604e-05,
+      "loss": 0.4471,
+      "step": 13668
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.99189702278718e-05,
+      "loss": 0.4601,
+      "step": 13669
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991894935468282e-05,
+      "loss": 0.5118,
+      "step": 13670
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991892847881667e-05,
+      "loss": 0.432,
+      "step": 13671
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918907600273353e-05,
+      "loss": 0.4438,
+      "step": 13672
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918886719052874e-05,
+      "loss": 0.4671,
+      "step": 13673
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918865835155246e-05,
+      "loss": 0.5147,
+      "step": 13674
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918844948580468e-05,
+      "loss": 0.4958,
+      "step": 13675
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918824059328548e-05,
+      "loss": 0.4512,
+      "step": 13676
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991880316739949e-05,
+      "loss": 0.527,
+      "step": 13677
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.99187822727933e-05,
+      "loss": 0.4381,
+      "step": 13678
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918761375509987e-05,
+      "loss": 0.4666,
+      "step": 13679
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991874047554955e-05,
+      "loss": 0.4673,
+      "step": 13680
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918719572912003e-05,
+      "loss": 0.4731,
+      "step": 13681
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918698667597343e-05,
+      "loss": 0.4831,
+      "step": 13682
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991867775960558e-05,
+      "loss": 0.4524,
+      "step": 13683
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918656848936724e-05,
+      "loss": 0.4509,
+      "step": 13684
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918635935590772e-05,
+      "loss": 0.4518,
+      "step": 13685
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991861501956774e-05,
+      "loss": 0.4997,
+      "step": 13686
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918594100867617e-05,
+      "loss": 0.4818,
+      "step": 13687
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918573179490427e-05,
+      "loss": 0.4577,
+      "step": 13688
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918552255436166e-05,
+      "loss": 0.4907,
+      "step": 13689
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991853132870484e-05,
+      "loss": 0.4557,
+      "step": 13690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918510399296454e-05,
+      "loss": 0.4898,
+      "step": 13691
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918489467211023e-05,
+      "loss": 0.5392,
+      "step": 13692
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918468532448538e-05,
+      "loss": 0.5453,
+      "step": 13693
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918447595009018e-05,
+      "loss": 0.4228,
+      "step": 13694
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918426654892458e-05,
+      "loss": 0.4884,
+      "step": 13695
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991840571209887e-05,
+      "loss": 0.4573,
+      "step": 13696
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991838476662826e-05,
+      "loss": 0.4986,
+      "step": 13697
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991836381848063e-05,
+      "loss": 0.4287,
+      "step": 13698
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918342867655984e-05,
+      "loss": 0.4902,
+      "step": 13699
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918321914154335e-05,
+      "loss": 0.4581,
+      "step": 13700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991830095797568e-05,
+      "loss": 0.4558,
+      "step": 13701
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918279999120035e-05,
+      "loss": 0.4692,
+      "step": 13702
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918259037587395e-05,
+      "loss": 0.4829,
+      "step": 13703
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918238073377772e-05,
+      "loss": 0.4872,
+      "step": 13704
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918217106491173e-05,
+      "loss": 0.5054,
+      "step": 13705
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.99181961369276e-05,
+      "loss": 0.511,
+      "step": 13706
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918175164687057e-05,
+      "loss": 0.6502,
+      "step": 13707
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918154189769556e-05,
+      "loss": 0.4089,
+      "step": 13708
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918133212175094e-05,
+      "loss": 0.4716,
+      "step": 13709
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918112231903683e-05,
+      "loss": 0.4568,
+      "step": 13710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918091248955326e-05,
+      "loss": 0.5465,
+      "step": 13711
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991807026333003e-05,
+      "loss": 0.4783,
+      "step": 13712
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918049275027805e-05,
+      "loss": 0.5039,
+      "step": 13713
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9918028284048648e-05,
+      "loss": 0.5134,
+      "step": 13714
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991800729039257e-05,
+      "loss": 0.4163,
+      "step": 13715
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917986294059576e-05,
+      "loss": 0.4953,
+      "step": 13716
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991796529504967e-05,
+      "loss": 0.5126,
+      "step": 13717
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991794429336286e-05,
+      "loss": 0.4917,
+      "step": 13718
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917923288999152e-05,
+      "loss": 0.369,
+      "step": 13719
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991790228195855e-05,
+      "loss": 0.4581,
+      "step": 13720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917881272241054e-05,
+      "loss": 0.4658,
+      "step": 13721
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991786025984668e-05,
+      "loss": 0.4303,
+      "step": 13722
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991783924477543e-05,
+      "loss": 0.5042,
+      "step": 13723
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917818227027307e-05,
+      "loss": 0.5629,
+      "step": 13724
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991779720660232e-05,
+      "loss": 0.5337,
+      "step": 13725
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991777618350047e-05,
+      "loss": 0.4658,
+      "step": 13726
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917755157721768e-05,
+      "loss": 0.4998,
+      "step": 13727
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917734129266217e-05,
+      "loss": 0.4713,
+      "step": 13728
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917713098133825e-05,
+      "loss": 0.5826,
+      "step": 13729
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917692064324596e-05,
+      "loss": 0.4952,
+      "step": 13730
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991767102783853e-05,
+      "loss": 0.5444,
+      "step": 13731
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991764998867564e-05,
+      "loss": 0.4398,
+      "step": 13732
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917628946835936e-05,
+      "loss": 0.4921,
+      "step": 13733
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917607902319414e-05,
+      "loss": 0.565,
+      "step": 13734
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991758685512608e-05,
+      "loss": 0.4852,
+      "step": 13735
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991756580525595e-05,
+      "loss": 0.4084,
+      "step": 13736
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917544752709016e-05,
+      "loss": 0.4652,
+      "step": 13737
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917523697485292e-05,
+      "loss": 0.4562,
+      "step": 13738
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917502639584786e-05,
+      "loss": 0.5528,
+      "step": 13739
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917481579007493e-05,
+      "loss": 0.5804,
+      "step": 13740
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991746051575343e-05,
+      "loss": 0.5303,
+      "step": 13741
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917439449822596e-05,
+      "loss": 0.5237,
+      "step": 13742
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917418381215e-05,
+      "loss": 0.5393,
+      "step": 13743
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917397309930648e-05,
+      "loss": 0.4838,
+      "step": 13744
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917376235969543e-05,
+      "loss": 0.4622,
+      "step": 13745
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917355159331692e-05,
+      "loss": 0.5143,
+      "step": 13746
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917334080017098e-05,
+      "loss": 0.5568,
+      "step": 13747
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.991731299802577e-05,
+      "loss": 0.4797,
+      "step": 13748
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917291913357713e-05,
+      "loss": 0.4354,
+      "step": 13749
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9917270826012935e-05,
+      "loss": 0.4638,
+      "step": 13750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9917249735991438e-05,
+      "loss": 0.4952,
+      "step": 13751
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991722864329323e-05,
+      "loss": 0.4745,
+      "step": 13752
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991720754791831e-05,
+      "loss": 0.5696,
+      "step": 13753
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991718644986669e-05,
+      "loss": 0.5072,
+      "step": 13754
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9917165349138382e-05,
+      "loss": 0.4465,
+      "step": 13755
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991714424573338e-05,
+      "loss": 0.473,
+      "step": 13756
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9917123139651697e-05,
+      "loss": 0.4936,
+      "step": 13757
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9917102030893333e-05,
+      "loss": 0.4092,
+      "step": 13758
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9917080919458298e-05,
+      "loss": 0.4589,
+      "step": 13759
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9917059805346597e-05,
+      "loss": 0.4623,
+      "step": 13760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9917038688558236e-05,
+      "loss": 0.5855,
+      "step": 13761
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991701756909322e-05,
+      "loss": 0.4772,
+      "step": 13762
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916996446951553e-05,
+      "loss": 0.4784,
+      "step": 13763
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991697532213324e-05,
+      "loss": 0.5696,
+      "step": 13764
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916954194638295e-05,
+      "loss": 0.454,
+      "step": 13765
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916933064466715e-05,
+      "loss": 0.4633,
+      "step": 13766
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916911931618507e-05,
+      "loss": 0.4597,
+      "step": 13767
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991689079609368e-05,
+      "loss": 0.4768,
+      "step": 13768
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916869657892236e-05,
+      "loss": 0.4752,
+      "step": 13769
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916848517014183e-05,
+      "loss": 0.4856,
+      "step": 13770
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916827373459526e-05,
+      "loss": 0.4006,
+      "step": 13771
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916806227228272e-05,
+      "loss": 0.444,
+      "step": 13772
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916785078320425e-05,
+      "loss": 0.4864,
+      "step": 13773
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991676392673599e-05,
+      "loss": 0.4419,
+      "step": 13774
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916742772474976e-05,
+      "loss": 0.47,
+      "step": 13775
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991672161553739e-05,
+      "loss": 0.4784,
+      "step": 13776
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916700455923228e-05,
+      "loss": 0.4635,
+      "step": 13777
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916679293632507e-05,
+      "loss": 0.5313,
+      "step": 13778
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916658128665227e-05,
+      "loss": 0.4905,
+      "step": 13779
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916636961021393e-05,
+      "loss": 0.4488,
+      "step": 13780
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991661579070101e-05,
+      "loss": 0.4873,
+      "step": 13781
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991659461770409e-05,
+      "loss": 0.4267,
+      "step": 13782
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916573442030633e-05,
+      "loss": 0.501,
+      "step": 13783
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916552263680645e-05,
+      "loss": 0.5511,
+      "step": 13784
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991653108265414e-05,
+      "loss": 0.5081,
+      "step": 13785
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991650989895111e-05,
+      "loss": 0.3908,
+      "step": 13786
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916488712571568e-05,
+      "loss": 0.5189,
+      "step": 13787
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991646752351552e-05,
+      "loss": 0.4366,
+      "step": 13788
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916446331782975e-05,
+      "loss": 0.487,
+      "step": 13789
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916425137373934e-05,
+      "loss": 0.4995,
+      "step": 13790
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.99164039402884e-05,
+      "loss": 0.3923,
+      "step": 13791
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916382740526383e-05,
+      "loss": 0.5055,
+      "step": 13792
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991636153808789e-05,
+      "loss": 0.4715,
+      "step": 13793
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916340332972924e-05,
+      "loss": 0.4914,
+      "step": 13794
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916319125181488e-05,
+      "loss": 0.4765,
+      "step": 13795
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916297914713597e-05,
+      "loss": 0.3882,
+      "step": 13796
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916276701569247e-05,
+      "loss": 0.3966,
+      "step": 13797
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916255485748446e-05,
+      "loss": 0.5145,
+      "step": 13798
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916234267251204e-05,
+      "loss": 0.4422,
+      "step": 13799
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916213046077525e-05,
+      "loss": 0.4939,
+      "step": 13800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991619182222741e-05,
+      "loss": 0.4513,
+      "step": 13801
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916170595700876e-05,
+      "loss": 0.5211,
+      "step": 13802
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916149366497913e-05,
+      "loss": 0.5161,
+      "step": 13803
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916128134618537e-05,
+      "loss": 0.5747,
+      "step": 13804
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916106900062756e-05,
+      "loss": 0.468,
+      "step": 13805
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916085662830565e-05,
+      "loss": 0.5183,
+      "step": 13806
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991606442292198e-05,
+      "loss": 0.4606,
+      "step": 13807
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916043180337002e-05,
+      "loss": 0.4994,
+      "step": 13808
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9916021935075637e-05,
+      "loss": 0.5107,
+      "step": 13809
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991600068713789e-05,
+      "loss": 0.4436,
+      "step": 13810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915979436523772e-05,
+      "loss": 0.4028,
+      "step": 13811
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915958183233283e-05,
+      "loss": 0.4445,
+      "step": 13812
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991593692726643e-05,
+      "loss": 0.4733,
+      "step": 13813
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991591566862322e-05,
+      "loss": 0.5225,
+      "step": 13814
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915894407303657e-05,
+      "loss": 0.507,
+      "step": 13815
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991587314330775e-05,
+      "loss": 0.4799,
+      "step": 13816
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915851876635498e-05,
+      "loss": 0.5068,
+      "step": 13817
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915830607286914e-05,
+      "loss": 0.3918,
+      "step": 13818
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915809335262005e-05,
+      "loss": 0.4874,
+      "step": 13819
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991578806056077e-05,
+      "loss": 0.4642,
+      "step": 13820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991576678318321e-05,
+      "loss": 0.4736,
+      "step": 13821
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915745503129347e-05,
+      "loss": 0.4512,
+      "step": 13822
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915724220399174e-05,
+      "loss": 0.4775,
+      "step": 13823
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915702934992703e-05,
+      "loss": 0.4246,
+      "step": 13824
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915681646909936e-05,
+      "loss": 0.3902,
+      "step": 13825
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991566035615088e-05,
+      "loss": 0.4371,
+      "step": 13826
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915639062715542e-05,
+      "loss": 0.4471,
+      "step": 13827
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915617766603926e-05,
+      "loss": 0.3788,
+      "step": 13828
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991559646781604e-05,
+      "loss": 0.4551,
+      "step": 13829
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915575166351888e-05,
+      "loss": 0.6144,
+      "step": 13830
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991555386221147e-05,
+      "loss": 0.4786,
+      "step": 13831
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915532555394804e-05,
+      "loss": 0.4743,
+      "step": 13832
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915511245901887e-05,
+      "loss": 0.474,
+      "step": 13833
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915489933732732e-05,
+      "loss": 0.4451,
+      "step": 13834
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915468618887335e-05,
+      "loss": 0.4594,
+      "step": 13835
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915447301365708e-05,
+      "loss": 0.4119,
+      "step": 13836
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915425981167856e-05,
+      "loss": 0.4852,
+      "step": 13837
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991540465829378e-05,
+      "loss": 0.5041,
+      "step": 13838
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915383332743496e-05,
+      "loss": 0.4337,
+      "step": 13839
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915362004516998e-05,
+      "loss": 0.4718,
+      "step": 13840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.99153406736143e-05,
+      "loss": 0.5192,
+      "step": 13841
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915319340035408e-05,
+      "loss": 0.4016,
+      "step": 13842
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915298003780318e-05,
+      "loss": 0.5131,
+      "step": 13843
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991527666484905e-05,
+      "loss": 0.5242,
+      "step": 13844
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.99152553232416e-05,
+      "loss": 0.4672,
+      "step": 13845
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915233978957975e-05,
+      "loss": 0.4717,
+      "step": 13846
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9915212631998182e-05,
+      "loss": 0.3885,
+      "step": 13847
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991519128236223e-05,
+      "loss": 0.5567,
+      "step": 13848
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.991516993005012e-05,
+      "loss": 0.4864,
+      "step": 13849
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9915148575061855e-05,
+      "loss": 0.4487,
+      "step": 13850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991512721739745e-05,
+      "loss": 0.4685,
+      "step": 13851
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9915105857056903e-05,
+      "loss": 0.4729,
+      "step": 13852
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9915084494040225e-05,
+      "loss": 0.5319,
+      "step": 13853
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9915063128347417e-05,
+      "loss": 0.4316,
+      "step": 13854
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991504175997849e-05,
+      "loss": 0.4976,
+      "step": 13855
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9915020388933443e-05,
+      "loss": 0.4857,
+      "step": 13856
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914999015212287e-05,
+      "loss": 0.4754,
+      "step": 13857
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991497763881503e-05,
+      "loss": 0.3878,
+      "step": 13858
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991495625974167e-05,
+      "loss": 0.4853,
+      "step": 13859
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991493487799222e-05,
+      "loss": 0.5431,
+      "step": 13860
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991491349356668e-05,
+      "loss": 0.4408,
+      "step": 13861
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914892106465058e-05,
+      "loss": 0.5617,
+      "step": 13862
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914870716687363e-05,
+      "loss": 0.4837,
+      "step": 13863
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.99148493242336e-05,
+      "loss": 0.4666,
+      "step": 13864
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914827929103767e-05,
+      "loss": 0.5347,
+      "step": 13865
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991480653129788e-05,
+      "loss": 0.4783,
+      "step": 13866
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914785130815938e-05,
+      "loss": 0.5291,
+      "step": 13867
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914763727657948e-05,
+      "loss": 0.6006,
+      "step": 13868
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991474232182392e-05,
+      "loss": 0.4694,
+      "step": 13869
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914720913313857e-05,
+      "loss": 0.494,
+      "step": 13870
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991469950212776e-05,
+      "loss": 0.4736,
+      "step": 13871
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991467808826564e-05,
+      "loss": 0.5079,
+      "step": 13872
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991465667172751e-05,
+      "loss": 0.4218,
+      "step": 13873
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991463525251336e-05,
+      "loss": 0.4392,
+      "step": 13874
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914613830623205e-05,
+      "loss": 0.4816,
+      "step": 13875
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914592406057053e-05,
+      "loss": 0.4734,
+      "step": 13876
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.99145709788149e-05,
+      "loss": 0.4142,
+      "step": 13877
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914549548896758e-05,
+      "loss": 0.4703,
+      "step": 13878
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991452811630264e-05,
+      "loss": 0.497,
+      "step": 13879
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991450668103254e-05,
+      "loss": 0.4454,
+      "step": 13880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991448524308647e-05,
+      "loss": 0.4618,
+      "step": 13881
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991446380246443e-05,
+      "loss": 0.4701,
+      "step": 13882
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914442359166434e-05,
+      "loss": 0.6258,
+      "step": 13883
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914420913192483e-05,
+      "loss": 0.3957,
+      "step": 13884
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914399464542583e-05,
+      "loss": 0.5246,
+      "step": 13885
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991437801321674e-05,
+      "loss": 0.5145,
+      "step": 13886
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991435655921496e-05,
+      "loss": 0.4634,
+      "step": 13887
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914335102537247e-05,
+      "loss": 0.5246,
+      "step": 13888
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914313643183612e-05,
+      "loss": 0.4621,
+      "step": 13889
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914292181154057e-05,
+      "loss": 0.5043,
+      "step": 13890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914270716448586e-05,
+      "loss": 0.5573,
+      "step": 13891
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914249249067206e-05,
+      "loss": 0.536,
+      "step": 13892
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914227779009927e-05,
+      "loss": 0.4766,
+      "step": 13893
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914206306276752e-05,
+      "loss": 0.5968,
+      "step": 13894
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914184830867685e-05,
+      "loss": 0.4258,
+      "step": 13895
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914163352782733e-05,
+      "loss": 0.4839,
+      "step": 13896
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.99141418720219e-05,
+      "loss": 0.4333,
+      "step": 13897
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914120388585195e-05,
+      "loss": 0.5,
+      "step": 13898
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914098902472623e-05,
+      "loss": 0.5051,
+      "step": 13899
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991407741368419e-05,
+      "loss": 0.5423,
+      "step": 13900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.99140559222199e-05,
+      "loss": 0.4992,
+      "step": 13901
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991403442807976e-05,
+      "loss": 0.5697,
+      "step": 13902
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9914012931263776e-05,
+      "loss": 0.5374,
+      "step": 13903
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913991431771956e-05,
+      "loss": 0.4929,
+      "step": 13904
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913969929604305e-05,
+      "loss": 0.4774,
+      "step": 13905
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991394842476082e-05,
+      "loss": 0.5488,
+      "step": 13906
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991392691724152e-05,
+      "loss": 0.4886,
+      "step": 13907
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913905407046402e-05,
+      "loss": 0.5537,
+      "step": 13908
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913883894175473e-05,
+      "loss": 0.5425,
+      "step": 13909
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913862378628744e-05,
+      "loss": 0.4562,
+      "step": 13910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913840860406214e-05,
+      "loss": 0.4498,
+      "step": 13911
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913819339507893e-05,
+      "loss": 0.3818,
+      "step": 13912
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913797815933788e-05,
+      "loss": 0.48,
+      "step": 13913
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913776289683903e-05,
+      "loss": 0.4731,
+      "step": 13914
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991375476075824e-05,
+      "loss": 0.4288,
+      "step": 13915
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913733229156808e-05,
+      "loss": 0.4107,
+      "step": 13916
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913711694879615e-05,
+      "loss": 0.4307,
+      "step": 13917
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913690157926665e-05,
+      "loss": 0.5221,
+      "step": 13918
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913668618297962e-05,
+      "loss": 0.5813,
+      "step": 13919
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913647075993516e-05,
+      "loss": 0.4666,
+      "step": 13920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991362553101333e-05,
+      "loss": 0.4132,
+      "step": 13921
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913603983357408e-05,
+      "loss": 0.4797,
+      "step": 13922
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991358243302576e-05,
+      "loss": 0.4925,
+      "step": 13923
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991356088001839e-05,
+      "loss": 0.5219,
+      "step": 13924
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913539324335304e-05,
+      "loss": 0.4609,
+      "step": 13925
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913517765976504e-05,
+      "loss": 0.4971,
+      "step": 13926
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913496204942e-05,
+      "loss": 0.4528,
+      "step": 13927
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.99134746412318e-05,
+      "loss": 0.5471,
+      "step": 13928
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913453074845905e-05,
+      "loss": 0.4659,
+      "step": 13929
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991343150578432e-05,
+      "loss": 0.5009,
+      "step": 13930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991340993404706e-05,
+      "loss": 0.509,
+      "step": 13931
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913388359634118e-05,
+      "loss": 0.412,
+      "step": 13932
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991336678254551e-05,
+      "loss": 0.447,
+      "step": 13933
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913345202781237e-05,
+      "loss": 0.5181,
+      "step": 13934
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913323620341304e-05,
+      "loss": 0.4848,
+      "step": 13935
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913302035225722e-05,
+      "loss": 0.4393,
+      "step": 13936
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913280447434492e-05,
+      "loss": 0.475,
+      "step": 13937
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991325885696762e-05,
+      "loss": 0.5306,
+      "step": 13938
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913237263825114e-05,
+      "loss": 0.4442,
+      "step": 13939
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913215668006976e-05,
+      "loss": 0.4488,
+      "step": 13940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991319406951322e-05,
+      "loss": 0.4965,
+      "step": 13941
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913172468343844e-05,
+      "loss": 0.4327,
+      "step": 13942
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913150864498856e-05,
+      "loss": 0.5251,
+      "step": 13943
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913129257978263e-05,
+      "loss": 0.5455,
+      "step": 13944
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.991310764878207e-05,
+      "loss": 0.4618,
+      "step": 13945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913086036910283e-05,
+      "loss": 0.4319,
+      "step": 13946
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9913064422362905e-05,
+      "loss": 0.4322,
+      "step": 13947
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9913042805139947e-05,
+      "loss": 0.4117,
+      "step": 13948
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9913021185241414e-05,
+      "loss": 0.5061,
+      "step": 13949
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991299956266731e-05,
+      "loss": 0.4751,
+      "step": 13950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991297793741764e-05,
+      "loss": 0.4912,
+      "step": 13951
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912956309492412e-05,
+      "loss": 0.5555,
+      "step": 13952
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991293467889163e-05,
+      "loss": 0.4312,
+      "step": 13953
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.99129130456153e-05,
+      "loss": 0.4487,
+      "step": 13954
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912891409663425e-05,
+      "loss": 0.4726,
+      "step": 13955
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991286977103602e-05,
+      "loss": 0.4584,
+      "step": 13956
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912848129733082e-05,
+      "loss": 0.4478,
+      "step": 13957
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991282648575462e-05,
+      "loss": 0.4599,
+      "step": 13958
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991280483910064e-05,
+      "loss": 0.5595,
+      "step": 13959
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991278318977115e-05,
+      "loss": 0.4959,
+      "step": 13960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991276153776615e-05,
+      "loss": 0.4578,
+      "step": 13961
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991273988308565e-05,
+      "loss": 0.4709,
+      "step": 13962
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912718225729658e-05,
+      "loss": 0.4351,
+      "step": 13963
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912696565698176e-05,
+      "loss": 0.4605,
+      "step": 13964
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991267490299121e-05,
+      "loss": 0.5301,
+      "step": 13965
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991265323760877e-05,
+      "loss": 0.4498,
+      "step": 13966
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991263156955085e-05,
+      "loss": 0.4309,
+      "step": 13967
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912609898817473e-05,
+      "loss": 0.554,
+      "step": 13968
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991258822540863e-05,
+      "loss": 0.4765,
+      "step": 13969
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912566549324336e-05,
+      "loss": 0.4226,
+      "step": 13970
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912544870564596e-05,
+      "loss": 0.4681,
+      "step": 13971
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912523189129412e-05,
+      "loss": 0.5048,
+      "step": 13972
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912501505018792e-05,
+      "loss": 0.4281,
+      "step": 13973
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912479818232738e-05,
+      "loss": 0.4543,
+      "step": 13974
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912458128771263e-05,
+      "loss": 0.5246,
+      "step": 13975
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991243643663437e-05,
+      "loss": 0.5768,
+      "step": 13976
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991241474182206e-05,
+      "loss": 0.5469,
+      "step": 13977
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991239304433435e-05,
+      "loss": 0.4129,
+      "step": 13978
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991237134417123e-05,
+      "loss": 0.3949,
+      "step": 13979
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912349641332724e-05,
+      "loss": 0.4664,
+      "step": 13980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912327935818822e-05,
+      "loss": 0.5075,
+      "step": 13981
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912306227629536e-05,
+      "loss": 0.4635,
+      "step": 13982
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912284516764876e-05,
+      "loss": 0.4921,
+      "step": 13983
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912262803224842e-05,
+      "loss": 0.4296,
+      "step": 13984
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991224108700944e-05,
+      "loss": 0.4026,
+      "step": 13985
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991221936811868e-05,
+      "loss": 0.4496,
+      "step": 13986
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912197646552565e-05,
+      "loss": 0.4519,
+      "step": 13987
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912175922311104e-05,
+      "loss": 0.4423,
+      "step": 13988
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.99121541953943e-05,
+      "loss": 0.4928,
+      "step": 13989
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912132465802157e-05,
+      "loss": 0.4472,
+      "step": 13990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912110733534682e-05,
+      "loss": 0.4562,
+      "step": 13991
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912088998591884e-05,
+      "loss": 0.5411,
+      "step": 13992
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912067260973767e-05,
+      "loss": 0.4703,
+      "step": 13993
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991204552068034e-05,
+      "loss": 0.4915,
+      "step": 13994
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9912023777711596e-05,
+      "loss": 0.5213,
+      "step": 13995
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991200203206756e-05,
+      "loss": 0.4375,
+      "step": 13996
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911980283748225e-05,
+      "loss": 0.522,
+      "step": 13997
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.99119585327536e-05,
+      "loss": 0.5019,
+      "step": 13998
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991193677908369e-05,
+      "loss": 0.4252,
+      "step": 13999
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911915022738504e-05,
+      "loss": 0.4802,
+      "step": 14000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911893263718042e-05,
+      "loss": 0.4528,
+      "step": 14001
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991187150202232e-05,
+      "loss": 0.4961,
+      "step": 14002
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911849737651333e-05,
+      "loss": 0.4323,
+      "step": 14003
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991182797060509e-05,
+      "loss": 0.4829,
+      "step": 14004
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911806200883602e-05,
+      "loss": 0.5066,
+      "step": 14005
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911784428486867e-05,
+      "loss": 0.4932,
+      "step": 14006
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.99117626534149e-05,
+      "loss": 0.5001,
+      "step": 14007
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.99117408756677e-05,
+      "loss": 0.4626,
+      "step": 14008
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911719095245273e-05,
+      "loss": 0.4895,
+      "step": 14009
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991169731214763e-05,
+      "loss": 0.4733,
+      "step": 14010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991167552637477e-05,
+      "loss": 0.4937,
+      "step": 14011
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911653737926707e-05,
+      "loss": 0.433,
+      "step": 14012
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991163194680344e-05,
+      "loss": 0.4494,
+      "step": 14013
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911610153004974e-05,
+      "loss": 0.444,
+      "step": 14014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991158835653132e-05,
+      "loss": 0.5514,
+      "step": 14015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911566557382483e-05,
+      "loss": 0.4589,
+      "step": 14016
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911544755558472e-05,
+      "loss": 0.5054,
+      "step": 14017
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911522951059284e-05,
+      "loss": 0.4904,
+      "step": 14018
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911501143884928e-05,
+      "loss": 0.4488,
+      "step": 14019
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911479334035415e-05,
+      "loss": 0.4327,
+      "step": 14020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911457521510745e-05,
+      "loss": 0.4659,
+      "step": 14021
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911435706310928e-05,
+      "loss": 0.5055,
+      "step": 14022
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911413888435968e-05,
+      "loss": 0.4854,
+      "step": 14023
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911392067885867e-05,
+      "loss": 0.4299,
+      "step": 14024
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991137024466064e-05,
+      "loss": 0.4302,
+      "step": 14025
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911348418760287e-05,
+      "loss": 0.5165,
+      "step": 14026
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911326590184814e-05,
+      "loss": 0.4564,
+      "step": 14027
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911304758934227e-05,
+      "loss": 0.4254,
+      "step": 14028
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911282925008535e-05,
+      "loss": 0.4455,
+      "step": 14029
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991126108840774e-05,
+      "loss": 0.4183,
+      "step": 14030
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911239249131848e-05,
+      "loss": 0.462,
+      "step": 14031
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911217407180867e-05,
+      "loss": 0.4693,
+      "step": 14032
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911195562554804e-05,
+      "loss": 0.5431,
+      "step": 14033
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911173715253658e-05,
+      "loss": 0.4317,
+      "step": 14034
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911151865277443e-05,
+      "loss": 0.5061,
+      "step": 14035
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911130012626163e-05,
+      "loss": 0.4324,
+      "step": 14036
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911108157299824e-05,
+      "loss": 0.5215,
+      "step": 14037
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911086299298426e-05,
+      "loss": 0.4246,
+      "step": 14038
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991106443862198e-05,
+      "loss": 0.5012,
+      "step": 14039
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911042575270494e-05,
+      "loss": 0.4398,
+      "step": 14040
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9911020709243968e-05,
+      "loss": 0.5381,
+      "step": 14041
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9910998840542416e-05,
+      "loss": 0.4855,
+      "step": 14042
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9910976969165836e-05,
+      "loss": 0.5291,
+      "step": 14043
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.991095509511424e-05,
+      "loss": 0.513,
+      "step": 14044
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9910933218387628e-05,
+      "loss": 0.4338,
+      "step": 14045
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9910911338986007e-05,
+      "loss": 0.3968,
+      "step": 14046
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910889456909385e-05,
+      "loss": 0.4803,
+      "step": 14047
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991086757215777e-05,
+      "loss": 0.436,
+      "step": 14048
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910845684731167e-05,
+      "loss": 0.3865,
+      "step": 14049
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910823794629576e-05,
+      "loss": 0.5002,
+      "step": 14050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910801901853013e-05,
+      "loss": 0.4547,
+      "step": 14051
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910780006401473e-05,
+      "loss": 0.448,
+      "step": 14052
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991075810827497e-05,
+      "loss": 0.528,
+      "step": 14053
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910736207473508e-05,
+      "loss": 0.4642,
+      "step": 14054
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910714303997086e-05,
+      "loss": 0.4525,
+      "step": 14055
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910692397845724e-05,
+      "loss": 0.5296,
+      "step": 14056
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910670489019415e-05,
+      "loss": 0.4442,
+      "step": 14057
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991064857751817e-05,
+      "loss": 0.5834,
+      "step": 14058
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910626663341994e-05,
+      "loss": 0.4705,
+      "step": 14059
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910604746490895e-05,
+      "loss": 0.4761,
+      "step": 14060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991058282696488e-05,
+      "loss": 0.5122,
+      "step": 14061
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910560904763948e-05,
+      "loss": 0.5017,
+      "step": 14062
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991053897988811e-05,
+      "loss": 0.5566,
+      "step": 14063
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991051705233737e-05,
+      "loss": 0.4757,
+      "step": 14064
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991049512211174e-05,
+      "loss": 0.4905,
+      "step": 14065
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910473189211217e-05,
+      "loss": 0.5553,
+      "step": 14066
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991045125363581e-05,
+      "loss": 0.4592,
+      "step": 14067
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991042931538553e-05,
+      "loss": 0.4215,
+      "step": 14068
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910407374460375e-05,
+      "loss": 0.4792,
+      "step": 14069
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991038543086036e-05,
+      "loss": 0.5586,
+      "step": 14070
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910363484585483e-05,
+      "loss": 0.4715,
+      "step": 14071
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991034153563575e-05,
+      "loss": 0.4378,
+      "step": 14072
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910319584011172e-05,
+      "loss": 0.4309,
+      "step": 14073
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991029762971175e-05,
+      "loss": 0.5279,
+      "step": 14074
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910275672737493e-05,
+      "loss": 0.5085,
+      "step": 14075
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910253713088408e-05,
+      "loss": 0.446,
+      "step": 14076
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910231750764498e-05,
+      "loss": 0.5225,
+      "step": 14077
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991020978576577e-05,
+      "loss": 0.4726,
+      "step": 14078
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910187818092233e-05,
+      "loss": 0.4722,
+      "step": 14079
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910165847743888e-05,
+      "loss": 0.4779,
+      "step": 14080
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910143874720742e-05,
+      "loss": 0.519,
+      "step": 14081
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.99101218990228e-05,
+      "loss": 0.4808,
+      "step": 14082
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910099920650073e-05,
+      "loss": 0.4521,
+      "step": 14083
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910077939602567e-05,
+      "loss": 0.4613,
+      "step": 14084
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910055955880277e-05,
+      "loss": 0.4496,
+      "step": 14085
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.991003396948322e-05,
+      "loss": 0.4494,
+      "step": 14086
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9910011980411398e-05,
+      "loss": 0.4181,
+      "step": 14087
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990998998866482e-05,
+      "loss": 0.5011,
+      "step": 14088
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909967994243486e-05,
+      "loss": 0.4685,
+      "step": 14089
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990994599714741e-05,
+      "loss": 0.475,
+      "step": 14090
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990992399737659e-05,
+      "loss": 0.476,
+      "step": 14091
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909901994931032e-05,
+      "loss": 0.5844,
+      "step": 14092
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990987998981075e-05,
+      "loss": 0.5196,
+      "step": 14093
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990985798201574e-05,
+      "loss": 0.5832,
+      "step": 14094
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909835971546017e-05,
+      "loss": 0.5009,
+      "step": 14095
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909813958401584e-05,
+      "loss": 0.4611,
+      "step": 14096
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909791942582444e-05,
+      "loss": 0.5414,
+      "step": 14097
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909769924088604e-05,
+      "loss": 0.5425,
+      "step": 14098
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990974790292007e-05,
+      "loss": 0.5568,
+      "step": 14099
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909725879076853e-05,
+      "loss": 0.5336,
+      "step": 14100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990970385255895e-05,
+      "loss": 0.4191,
+      "step": 14101
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909681823366377e-05,
+      "loss": 0.5294,
+      "step": 14102
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990965979149913e-05,
+      "loss": 0.4794,
+      "step": 14103
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909637756957222e-05,
+      "loss": 0.5365,
+      "step": 14104
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909615719740657e-05,
+      "loss": 0.5253,
+      "step": 14105
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909593679849437e-05,
+      "loss": 0.4016,
+      "step": 14106
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990957163728357e-05,
+      "loss": 0.4961,
+      "step": 14107
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990954959204307e-05,
+      "loss": 0.5186,
+      "step": 14108
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990952754412793e-05,
+      "loss": 0.442,
+      "step": 14109
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909505493538167e-05,
+      "loss": 0.4541,
+      "step": 14110
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909483440273777e-05,
+      "loss": 0.4555,
+      "step": 14111
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909461384334776e-05,
+      "loss": 0.4124,
+      "step": 14112
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909439325721163e-05,
+      "loss": 0.3996,
+      "step": 14113
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909417264432945e-05,
+      "loss": 0.511,
+      "step": 14114
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909395200470133e-05,
+      "loss": 0.4264,
+      "step": 14115
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909373133832726e-05,
+      "loss": 0.5692,
+      "step": 14116
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990935106452073e-05,
+      "loss": 0.5067,
+      "step": 14117
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990932899253416e-05,
+      "loss": 0.5176,
+      "step": 14118
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990930691787301e-05,
+      "loss": 0.4605,
+      "step": 14119
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909284840537294e-05,
+      "loss": 0.5524,
+      "step": 14120
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909262760527016e-05,
+      "loss": 0.4414,
+      "step": 14121
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990924067784218e-05,
+      "loss": 0.4862,
+      "step": 14122
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909218592482794e-05,
+      "loss": 0.5234,
+      "step": 14123
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909196504448867e-05,
+      "loss": 0.4236,
+      "step": 14124
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.99091744137404e-05,
+      "loss": 0.5256,
+      "step": 14125
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.99091523203574e-05,
+      "loss": 0.4687,
+      "step": 14126
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909130224299872e-05,
+      "loss": 0.4521,
+      "step": 14127
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909108125567824e-05,
+      "loss": 0.508,
+      "step": 14128
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909086024161263e-05,
+      "loss": 0.4429,
+      "step": 14129
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9909063920080195e-05,
+      "loss": 0.4879,
+      "step": 14130
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990904181332462e-05,
+      "loss": 0.4616,
+      "step": 14131
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990901970389455e-05,
+      "loss": 0.5006,
+      "step": 14132
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9908997591789988e-05,
+      "loss": 0.4502,
+      "step": 14133
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9908975477010945e-05,
+      "loss": 0.552,
+      "step": 14134
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990895335955742e-05,
+      "loss": 0.5553,
+      "step": 14135
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990893123942942e-05,
+      "loss": 0.4605,
+      "step": 14136
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990890911662696e-05,
+      "loss": 0.4572,
+      "step": 14137
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9908886991150033e-05,
+      "loss": 0.4789,
+      "step": 14138
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9908864862998654e-05,
+      "loss": 0.4923,
+      "step": 14139
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9908842732172826e-05,
+      "loss": 0.5358,
+      "step": 14140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9908820598672553e-05,
+      "loss": 0.4737,
+      "step": 14141
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9908798462497844e-05,
+      "loss": 0.4182,
+      "step": 14142
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9908776323648707e-05,
+      "loss": 0.4622,
+      "step": 14143
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990875418212514e-05,
+      "loss": 0.4784,
+      "step": 14144
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908732037927156e-05,
+      "loss": 0.4515,
+      "step": 14145
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990870989105476e-05,
+      "loss": 0.4076,
+      "step": 14146
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908687741507955e-05,
+      "loss": 0.5589,
+      "step": 14147
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990866558928675e-05,
+      "loss": 0.4985,
+      "step": 14148
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990864343439115e-05,
+      "loss": 0.4275,
+      "step": 14149
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990862127682116e-05,
+      "loss": 0.5437,
+      "step": 14150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908599116576787e-05,
+      "loss": 0.4541,
+      "step": 14151
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990857695365804e-05,
+      "loss": 0.451,
+      "step": 14152
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908554788064918e-05,
+      "loss": 0.5128,
+      "step": 14153
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908532619797428e-05,
+      "loss": 0.4214,
+      "step": 14154
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908510448855584e-05,
+      "loss": 0.4133,
+      "step": 14155
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908488275239387e-05,
+      "loss": 0.5108,
+      "step": 14156
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990846609894884e-05,
+      "loss": 0.4882,
+      "step": 14157
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908443919983953e-05,
+      "loss": 0.531,
+      "step": 14158
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990842173834473e-05,
+      "loss": 0.5265,
+      "step": 14159
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908399554031177e-05,
+      "loss": 0.5266,
+      "step": 14160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.99083773670433e-05,
+      "loss": 0.431,
+      "step": 14161
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908355177381107e-05,
+      "loss": 0.4967,
+      "step": 14162
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908332985044604e-05,
+      "loss": 0.4659,
+      "step": 14163
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908310790033794e-05,
+      "loss": 0.4127,
+      "step": 14164
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908288592348684e-05,
+      "loss": 0.5658,
+      "step": 14165
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990826639198928e-05,
+      "loss": 0.4112,
+      "step": 14166
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908244188955592e-05,
+      "loss": 0.4424,
+      "step": 14167
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908221983247623e-05,
+      "loss": 0.5476,
+      "step": 14168
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908199774865372e-05,
+      "loss": 0.4911,
+      "step": 14169
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908177563808857e-05,
+      "loss": 0.4659,
+      "step": 14170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908155350078077e-05,
+      "loss": 0.4318,
+      "step": 14171
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990813313367304e-05,
+      "loss": 0.468,
+      "step": 14172
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908110914593753e-05,
+      "loss": 0.4162,
+      "step": 14173
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990808869284022e-05,
+      "loss": 0.3829,
+      "step": 14174
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908066468412442e-05,
+      "loss": 0.5264,
+      "step": 14175
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908044241310437e-05,
+      "loss": 0.4431,
+      "step": 14176
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9908022011534203e-05,
+      "loss": 0.4109,
+      "step": 14177
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907999779083748e-05,
+      "loss": 0.5695,
+      "step": 14178
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907977543959077e-05,
+      "loss": 0.4792,
+      "step": 14179
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907955306160194e-05,
+      "loss": 0.4453,
+      "step": 14180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990793306568711e-05,
+      "loss": 0.488,
+      "step": 14181
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907910822539828e-05,
+      "loss": 0.4221,
+      "step": 14182
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907888576718357e-05,
+      "loss": 0.506,
+      "step": 14183
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.99078663282227e-05,
+      "loss": 0.52,
+      "step": 14184
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990784407705286e-05,
+      "loss": 0.5186,
+      "step": 14185
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907821823208848e-05,
+      "loss": 0.4044,
+      "step": 14186
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990779956669067e-05,
+      "loss": 0.5114,
+      "step": 14187
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907777307498327e-05,
+      "loss": 0.4879,
+      "step": 14188
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907755045631833e-05,
+      "loss": 0.5507,
+      "step": 14189
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990773278109119e-05,
+      "loss": 0.5205,
+      "step": 14190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.99077105138764e-05,
+      "loss": 0.4494,
+      "step": 14191
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907688243987474e-05,
+      "loss": 0.487,
+      "step": 14192
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907665971424418e-05,
+      "loss": 0.5486,
+      "step": 14193
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907643696187235e-05,
+      "loss": 0.4523,
+      "step": 14194
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990762141827593e-05,
+      "loss": 0.5907,
+      "step": 14195
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907599137690518e-05,
+      "loss": 0.5096,
+      "step": 14196
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990757685443099e-05,
+      "loss": 0.4163,
+      "step": 14197
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990755456849737e-05,
+      "loss": 0.4608,
+      "step": 14198
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990753227988965e-05,
+      "loss": 0.5036,
+      "step": 14199
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907509988607843e-05,
+      "loss": 0.4724,
+      "step": 14200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990748769465195e-05,
+      "loss": 0.4988,
+      "step": 14201
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990746539802198e-05,
+      "loss": 0.5426,
+      "step": 14202
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990744309871794e-05,
+      "loss": 0.4633,
+      "step": 14203
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907420796739837e-05,
+      "loss": 0.5073,
+      "step": 14204
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990739849208767e-05,
+      "loss": 0.5403,
+      "step": 14205
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907376184761453e-05,
+      "loss": 0.5081,
+      "step": 14206
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907353874761187e-05,
+      "loss": 0.4341,
+      "step": 14207
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907331562086883e-05,
+      "loss": 0.5172,
+      "step": 14208
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907309246738542e-05,
+      "loss": 0.4629,
+      "step": 14209
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907286928716173e-05,
+      "loss": 0.5281,
+      "step": 14210
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990726460801978e-05,
+      "loss": 0.5578,
+      "step": 14211
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907242284649365e-05,
+      "loss": 0.4128,
+      "step": 14212
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907219958604947e-05,
+      "loss": 0.5889,
+      "step": 14213
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990719762988652e-05,
+      "loss": 0.4789,
+      "step": 14214
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907175298494095e-05,
+      "loss": 0.4534,
+      "step": 14215
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907152964427675e-05,
+      "loss": 0.4581,
+      "step": 14216
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907130627687274e-05,
+      "loss": 0.4435,
+      "step": 14217
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907108288272888e-05,
+      "loss": 0.4383,
+      "step": 14218
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907085946184527e-05,
+      "loss": 0.5247,
+      "step": 14219
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907063601422195e-05,
+      "loss": 0.412,
+      "step": 14220
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907041253985902e-05,
+      "loss": 0.4649,
+      "step": 14221
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9907018903875655e-05,
+      "loss": 0.5558,
+      "step": 14222
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906996551091457e-05,
+      "loss": 0.4741,
+      "step": 14223
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990697419563331e-05,
+      "loss": 0.4309,
+      "step": 14224
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906951837501228e-05,
+      "loss": 0.5077,
+      "step": 14225
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990692947669521e-05,
+      "loss": 0.5735,
+      "step": 14226
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990690711321527e-05,
+      "loss": 0.463,
+      "step": 14227
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906884747061406e-05,
+      "loss": 0.5245,
+      "step": 14228
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906862378233628e-05,
+      "loss": 0.4771,
+      "step": 14229
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906840006731943e-05,
+      "loss": 0.6046,
+      "step": 14230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906817632556355e-05,
+      "loss": 0.545,
+      "step": 14231
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990679525570687e-05,
+      "loss": 0.4492,
+      "step": 14232
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906772876183495e-05,
+      "loss": 0.4845,
+      "step": 14233
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906750493986234e-05,
+      "loss": 0.4671,
+      "step": 14234
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.99067281091151e-05,
+      "loss": 0.4881,
+      "step": 14235
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990670572157009e-05,
+      "loss": 0.4902,
+      "step": 14236
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906683331351212e-05,
+      "loss": 0.5338,
+      "step": 14237
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906660938458476e-05,
+      "loss": 0.5101,
+      "step": 14238
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906638542891884e-05,
+      "loss": 0.4804,
+      "step": 14239
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990661614465145e-05,
+      "loss": 0.441,
+      "step": 14240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.990659374373717e-05,
+      "loss": 0.4503,
+      "step": 14241
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906571340149052e-05,
+      "loss": 0.4628,
+      "step": 14242
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9906548933887104e-05,
+      "loss": 0.4586,
+      "step": 14243
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906526524951337e-05,
+      "loss": 0.464,
+      "step": 14244
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990650411334175e-05,
+      "loss": 0.4895,
+      "step": 14245
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990648169905835e-05,
+      "loss": 0.472,
+      "step": 14246
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906459282101143e-05,
+      "loss": 0.4381,
+      "step": 14247
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906436862470138e-05,
+      "loss": 0.4386,
+      "step": 14248
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990641444016534e-05,
+      "loss": 0.4443,
+      "step": 14249
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906392015186756e-05,
+      "loss": 0.4194,
+      "step": 14250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906369587534383e-05,
+      "loss": 0.4969,
+      "step": 14251
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906347157208242e-05,
+      "loss": 0.5073,
+      "step": 14252
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990632472420833e-05,
+      "loss": 0.4664,
+      "step": 14253
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906302288534657e-05,
+      "loss": 0.5261,
+      "step": 14254
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906279850187222e-05,
+      "loss": 0.5296,
+      "step": 14255
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906257409166037e-05,
+      "loss": 0.558,
+      "step": 14256
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906234965471107e-05,
+      "loss": 0.4821,
+      "step": 14257
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990621251910244e-05,
+      "loss": 0.465,
+      "step": 14258
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906190070060035e-05,
+      "loss": 0.4846,
+      "step": 14259
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906167618343907e-05,
+      "loss": 0.5928,
+      "step": 14260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906145163954058e-05,
+      "loss": 0.5168,
+      "step": 14261
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906122706890495e-05,
+      "loss": 0.4127,
+      "step": 14262
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906100247153222e-05,
+      "loss": 0.4767,
+      "step": 14263
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906077784742245e-05,
+      "loss": 0.4443,
+      "step": 14264
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9906055319657572e-05,
+      "loss": 0.537,
+      "step": 14265
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990603285189921e-05,
+      "loss": 0.4337,
+      "step": 14266
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990601038146716e-05,
+      "loss": 0.4549,
+      "step": 14267
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905987908361433e-05,
+      "loss": 0.5001,
+      "step": 14268
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905965432582037e-05,
+      "loss": 0.4504,
+      "step": 14269
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905942954128968e-05,
+      "loss": 0.4093,
+      "step": 14270
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905920473002242e-05,
+      "loss": 0.4893,
+      "step": 14271
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905897989201865e-05,
+      "loss": 0.5583,
+      "step": 14272
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905875502727837e-05,
+      "loss": 0.4723,
+      "step": 14273
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905853013580164e-05,
+      "loss": 0.4937,
+      "step": 14274
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990583052175886e-05,
+      "loss": 0.6033,
+      "step": 14275
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990580802726392e-05,
+      "loss": 0.3943,
+      "step": 14276
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905785530095364e-05,
+      "loss": 0.5263,
+      "step": 14277
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990576303025318e-05,
+      "loss": 0.4376,
+      "step": 14278
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905740527737394e-05,
+      "loss": 0.523,
+      "step": 14279
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905718022547997e-05,
+      "loss": 0.5458,
+      "step": 14280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905695514685003e-05,
+      "loss": 0.4692,
+      "step": 14281
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905673004148416e-05,
+      "loss": 0.4435,
+      "step": 14282
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990565049093824e-05,
+      "loss": 0.4643,
+      "step": 14283
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990562797505448e-05,
+      "loss": 0.4753,
+      "step": 14284
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905605456497146e-05,
+      "loss": 0.4791,
+      "step": 14285
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905582935266247e-05,
+      "loss": 0.4248,
+      "step": 14286
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905560411361783e-05,
+      "loss": 0.4985,
+      "step": 14287
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990553788478376e-05,
+      "loss": 0.455,
+      "step": 14288
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990551535553219e-05,
+      "loss": 0.4586,
+      "step": 14289
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990549282360707e-05,
+      "loss": 0.4395,
+      "step": 14290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905470289008413e-05,
+      "loss": 0.4334,
+      "step": 14291
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990544775173622e-05,
+      "loss": 0.4724,
+      "step": 14292
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905425211790508e-05,
+      "loss": 0.482,
+      "step": 14293
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990540266917127e-05,
+      "loss": 0.4457,
+      "step": 14294
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990538012387852e-05,
+      "loss": 0.566,
+      "step": 14295
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990535757591226e-05,
+      "loss": 0.5054,
+      "step": 14296
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905335025272498e-05,
+      "loss": 0.4833,
+      "step": 14297
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990531247195924e-05,
+      "loss": 0.4046,
+      "step": 14298
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905289915972492e-05,
+      "loss": 0.4744,
+      "step": 14299
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990526735731226e-05,
+      "loss": 0.5356,
+      "step": 14300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905244795978552e-05,
+      "loss": 0.4984,
+      "step": 14301
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905222231971368e-05,
+      "loss": 0.4354,
+      "step": 14302
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990519966529072e-05,
+      "loss": 0.5111,
+      "step": 14303
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905177095936613e-05,
+      "loss": 0.4591,
+      "step": 14304
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905154523909053e-05,
+      "loss": 0.4833,
+      "step": 14305
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990513194920804e-05,
+      "loss": 0.4683,
+      "step": 14306
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990510937183359e-05,
+      "loss": 0.4474,
+      "step": 14307
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905086791785706e-05,
+      "loss": 0.543,
+      "step": 14308
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990506420906439e-05,
+      "loss": 0.497,
+      "step": 14309
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9905041623669653e-05,
+      "loss": 0.5122,
+      "step": 14310
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.99050190356015e-05,
+      "loss": 0.4176,
+      "step": 14311
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904996444859934e-05,
+      "loss": 0.5234,
+      "step": 14312
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904973851444965e-05,
+      "loss": 0.5304,
+      "step": 14313
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904951255356594e-05,
+      "loss": 0.4842,
+      "step": 14314
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990492865659483e-05,
+      "loss": 0.4381,
+      "step": 14315
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904906055159683e-05,
+      "loss": 0.5788,
+      "step": 14316
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904883451051153e-05,
+      "loss": 0.556,
+      "step": 14317
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990486084426925e-05,
+      "loss": 0.4178,
+      "step": 14318
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990483823481398e-05,
+      "loss": 0.4494,
+      "step": 14319
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904815622685345e-05,
+      "loss": 0.4399,
+      "step": 14320
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904793007883356e-05,
+      "loss": 0.4532,
+      "step": 14321
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904770390408017e-05,
+      "loss": 0.4782,
+      "step": 14322
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904747770259332e-05,
+      "loss": 0.4808,
+      "step": 14323
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990472514743731e-05,
+      "loss": 0.406,
+      "step": 14324
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904702521941958e-05,
+      "loss": 0.6001,
+      "step": 14325
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990467989377328e-05,
+      "loss": 0.4902,
+      "step": 14326
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904657262931283e-05,
+      "loss": 0.5019,
+      "step": 14327
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904634629415972e-05,
+      "loss": 0.4156,
+      "step": 14328
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990461199322735e-05,
+      "loss": 0.4843,
+      "step": 14329
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904589354365435e-05,
+      "loss": 0.4901,
+      "step": 14330
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990456671283022e-05,
+      "loss": 0.4508,
+      "step": 14331
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904544068621716e-05,
+      "loss": 0.4864,
+      "step": 14332
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990452142173993e-05,
+      "loss": 0.5074,
+      "step": 14333
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904498772184867e-05,
+      "loss": 0.378,
+      "step": 14334
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904476119956535e-05,
+      "loss": 0.4049,
+      "step": 14335
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990445346505494e-05,
+      "loss": 0.4732,
+      "step": 14336
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990443080748008e-05,
+      "loss": 0.4489,
+      "step": 14337
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904408147231976e-05,
+      "loss": 0.4408,
+      "step": 14338
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.990438548431062e-05,
+      "loss": 0.4236,
+      "step": 14339
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904362818716024e-05,
+      "loss": 0.5644,
+      "step": 14340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9904340150448198e-05,
+      "loss": 0.5158,
+      "step": 14341
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9904317479507145e-05,
+      "loss": 0.4881,
+      "step": 14342
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9904294805892864e-05,
+      "loss": 0.5296,
+      "step": 14343
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9904272129605374e-05,
+      "loss": 0.4998,
+      "step": 14344
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990424945064467e-05,
+      "loss": 0.57,
+      "step": 14345
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9904226769010767e-05,
+      "loss": 0.4726,
+      "step": 14346
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9904204084703663e-05,
+      "loss": 0.4791,
+      "step": 14347
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990418139772337e-05,
+      "loss": 0.472,
+      "step": 14348
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9904158708069894e-05,
+      "loss": 0.4523,
+      "step": 14349
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9904136015743235e-05,
+      "loss": 0.4756,
+      "step": 14350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990411332074341e-05,
+      "loss": 0.4605,
+      "step": 14351
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990409062307041e-05,
+      "loss": 0.5189,
+      "step": 14352
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9904067922724256e-05,
+      "loss": 0.6005,
+      "step": 14353
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9904045219704945e-05,
+      "loss": 0.4739,
+      "step": 14354
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990402251401249e-05,
+      "loss": 0.4741,
+      "step": 14355
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990399980564689e-05,
+      "loss": 0.4777,
+      "step": 14356
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990397709460815e-05,
+      "loss": 0.4952,
+      "step": 14357
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903954380896285e-05,
+      "loss": 0.4164,
+      "step": 14358
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903931664511295e-05,
+      "loss": 0.4721,
+      "step": 14359
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903908945453188e-05,
+      "loss": 0.5355,
+      "step": 14360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990388622372197e-05,
+      "loss": 0.466,
+      "step": 14361
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903863499317645e-05,
+      "loss": 0.5573,
+      "step": 14362
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903840772240223e-05,
+      "loss": 0.4363,
+      "step": 14363
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990381804248971e-05,
+      "loss": 0.5262,
+      "step": 14364
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903795310066104e-05,
+      "loss": 0.5052,
+      "step": 14365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990377257496942e-05,
+      "loss": 0.4885,
+      "step": 14366
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903749837199665e-05,
+      "loss": 0.4554,
+      "step": 14367
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990372709675684e-05,
+      "loss": 0.4948,
+      "step": 14368
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903704353640953e-05,
+      "loss": 0.4901,
+      "step": 14369
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903681607852007e-05,
+      "loss": 0.4564,
+      "step": 14370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903658859390012e-05,
+      "loss": 0.5145,
+      "step": 14371
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903636108254973e-05,
+      "loss": 0.4356,
+      "step": 14372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.99036133544469e-05,
+      "loss": 0.5392,
+      "step": 14373
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903590597965792e-05,
+      "loss": 0.586,
+      "step": 14374
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990356783881166e-05,
+      "loss": 0.5219,
+      "step": 14375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903545076984507e-05,
+      "loss": 0.4408,
+      "step": 14376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903522312484342e-05,
+      "loss": 0.4874,
+      "step": 14377
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903499545311172e-05,
+      "loss": 0.448,
+      "step": 14378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903476775464997e-05,
+      "loss": 0.6574,
+      "step": 14379
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990345400294583e-05,
+      "loss": 0.5485,
+      "step": 14380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903431227753676e-05,
+      "loss": 0.5492,
+      "step": 14381
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903408449888535e-05,
+      "loss": 0.4593,
+      "step": 14382
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990338566935042e-05,
+      "loss": 0.4202,
+      "step": 14383
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903362886139336e-05,
+      "loss": 0.4252,
+      "step": 14384
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990334010025529e-05,
+      "loss": 0.4173,
+      "step": 14385
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990331731169828e-05,
+      "loss": 0.4971,
+      "step": 14386
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903294520468325e-05,
+      "loss": 0.4198,
+      "step": 14387
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990327172656542e-05,
+      "loss": 0.5159,
+      "step": 14388
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990324892998958e-05,
+      "loss": 0.4711,
+      "step": 14389
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903226130740803e-05,
+      "loss": 0.4636,
+      "step": 14390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.99032033288191e-05,
+      "loss": 0.4063,
+      "step": 14391
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903180524224478e-05,
+      "loss": 0.4742,
+      "step": 14392
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903157716956935e-05,
+      "loss": 0.4637,
+      "step": 14393
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990313490701649e-05,
+      "loss": 0.4917,
+      "step": 14394
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990311209440314e-05,
+      "loss": 0.4596,
+      "step": 14395
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903089279116895e-05,
+      "loss": 0.4258,
+      "step": 14396
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9903066461157757e-05,
+      "loss": 0.4987,
+      "step": 14397
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990304364052574e-05,
+      "loss": 0.4681,
+      "step": 14398
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990302081722084e-05,
+      "loss": 0.5273,
+      "step": 14399
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902997991243072e-05,
+      "loss": 0.49,
+      "step": 14400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902975162592437e-05,
+      "loss": 0.4778,
+      "step": 14401
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902952331268942e-05,
+      "loss": 0.4451,
+      "step": 14402
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902929497272597e-05,
+      "loss": 0.4729,
+      "step": 14403
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902906660603403e-05,
+      "loss": 0.4677,
+      "step": 14404
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902883821261366e-05,
+      "loss": 0.4835,
+      "step": 14405
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902860979246496e-05,
+      "loss": 0.4974,
+      "step": 14406
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.99028381345588e-05,
+      "loss": 0.5325,
+      "step": 14407
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990281528719828e-05,
+      "loss": 0.5123,
+      "step": 14408
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902792437164943e-05,
+      "loss": 0.5399,
+      "step": 14409
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902769584458795e-05,
+      "loss": 0.4916,
+      "step": 14410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902746729079846e-05,
+      "loss": 0.4886,
+      "step": 14411
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.99027238710281e-05,
+      "loss": 0.4169,
+      "step": 14412
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990270101030356e-05,
+      "loss": 0.5757,
+      "step": 14413
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902678146906235e-05,
+      "loss": 0.5976,
+      "step": 14414
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990265528083613e-05,
+      "loss": 0.5117,
+      "step": 14415
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902632412093254e-05,
+      "loss": 0.4704,
+      "step": 14416
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990260954067761e-05,
+      "loss": 0.4299,
+      "step": 14417
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902586666589207e-05,
+      "loss": 0.4283,
+      "step": 14418
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990256378982805e-05,
+      "loss": 0.5105,
+      "step": 14419
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902540910394143e-05,
+      "loss": 0.4712,
+      "step": 14420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902518028287493e-05,
+      "loss": 0.4905,
+      "step": 14421
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990249514350811e-05,
+      "loss": 0.4888,
+      "step": 14422
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902472256055995e-05,
+      "loss": 0.445,
+      "step": 14423
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902449365931157e-05,
+      "loss": 0.5809,
+      "step": 14424
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902426473133603e-05,
+      "loss": 0.4963,
+      "step": 14425
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902403577663337e-05,
+      "loss": 0.4482,
+      "step": 14426
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902380679520364e-05,
+      "loss": 0.4831,
+      "step": 14427
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902357778704693e-05,
+      "loss": 0.5208,
+      "step": 14428
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990233487521633e-05,
+      "loss": 0.4557,
+      "step": 14429
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990231196905528e-05,
+      "loss": 0.449,
+      "step": 14430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990228906022155e-05,
+      "loss": 0.4828,
+      "step": 14431
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902266148715144e-05,
+      "loss": 0.4663,
+      "step": 14432
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902243234536076e-05,
+      "loss": 0.4933,
+      "step": 14433
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990222031768434e-05,
+      "loss": 0.3985,
+      "step": 14434
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902197398159952e-05,
+      "loss": 0.525,
+      "step": 14435
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902174475962913e-05,
+      "loss": 0.4466,
+      "step": 14436
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902151551093235e-05,
+      "loss": 0.5217,
+      "step": 14437
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902128623550914e-05,
+      "loss": 0.5275,
+      "step": 14438
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9902105693335964e-05,
+      "loss": 0.5029,
+      "step": 14439
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.990208276044839e-05,
+      "loss": 0.5094,
+      "step": 14440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9902059824888197e-05,
+      "loss": 0.5317,
+      "step": 14441
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9902036886655392e-05,
+      "loss": 0.5211,
+      "step": 14442
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9902013945749984e-05,
+      "loss": 0.4591,
+      "step": 14443
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990199100217197e-05,
+      "loss": 0.4334,
+      "step": 14444
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990196805592137e-05,
+      "loss": 0.528,
+      "step": 14445
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901945106998174e-05,
+      "loss": 0.5584,
+      "step": 14446
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901922155402404e-05,
+      "loss": 0.4333,
+      "step": 14447
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901899201134056e-05,
+      "loss": 0.5064,
+      "step": 14448
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990187624419314e-05,
+      "loss": 0.5096,
+      "step": 14449
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901853284579656e-05,
+      "loss": 0.4461,
+      "step": 14450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901830322293623e-05,
+      "loss": 0.5041,
+      "step": 14451
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901807357335037e-05,
+      "loss": 0.4724,
+      "step": 14452
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901784389703907e-05,
+      "loss": 0.4583,
+      "step": 14453
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990176141940024e-05,
+      "loss": 0.4535,
+      "step": 14454
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901738446424036e-05,
+      "loss": 0.4881,
+      "step": 14455
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901715470775312e-05,
+      "loss": 0.4349,
+      "step": 14456
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901692492454067e-05,
+      "loss": 0.4196,
+      "step": 14457
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901669511460308e-05,
+      "loss": 0.4861,
+      "step": 14458
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901646527794044e-05,
+      "loss": 0.5631,
+      "step": 14459
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990162354145528e-05,
+      "loss": 0.5665,
+      "step": 14460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990160055244402e-05,
+      "loss": 0.5961,
+      "step": 14461
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990157756076027e-05,
+      "loss": 0.5439,
+      "step": 14462
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990155456640404e-05,
+      "loss": 0.4834,
+      "step": 14463
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990153156937533e-05,
+      "loss": 0.4847,
+      "step": 14464
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901508569674158e-05,
+      "loss": 0.4444,
+      "step": 14465
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901485567300518e-05,
+      "loss": 0.5199,
+      "step": 14466
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990146256225442e-05,
+      "loss": 0.5081,
+      "step": 14467
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901439554535874e-05,
+      "loss": 0.5509,
+      "step": 14468
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990141654414488e-05,
+      "loss": 0.4767,
+      "step": 14469
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990139353108145e-05,
+      "loss": 0.4252,
+      "step": 14470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990137051534559e-05,
+      "loss": 0.4774,
+      "step": 14471
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901347496937297e-05,
+      "loss": 0.4389,
+      "step": 14472
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990132447585659e-05,
+      "loss": 0.5412,
+      "step": 14473
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901301452103464e-05,
+      "loss": 0.5205,
+      "step": 14474
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990127842567793e-05,
+      "loss": 0.5106,
+      "step": 14475
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990125539658e-05,
+      "loss": 0.503,
+      "step": 14476
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901232364809676e-05,
+      "loss": 0.5033,
+      "step": 14477
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901209330366956e-05,
+      "loss": 0.6081,
+      "step": 14478
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990118629325186e-05,
+      "loss": 0.4526,
+      "step": 14479
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901163253464385e-05,
+      "loss": 0.4738,
+      "step": 14480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901140211004538e-05,
+      "loss": 0.5117,
+      "step": 14481
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990111716587233e-05,
+      "loss": 0.4398,
+      "step": 14482
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901094118067762e-05,
+      "loss": 0.4521,
+      "step": 14483
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901071067590843e-05,
+      "loss": 0.4806,
+      "step": 14484
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990104801444158e-05,
+      "loss": 0.4265,
+      "step": 14485
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901024958619974e-05,
+      "loss": 0.4571,
+      "step": 14486
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9901001900126037e-05,
+      "loss": 0.5002,
+      "step": 14487
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900978838959776e-05,
+      "loss": 0.4496,
+      "step": 14488
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990095577512119e-05,
+      "loss": 0.4806,
+      "step": 14489
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900932708610293e-05,
+      "loss": 0.4304,
+      "step": 14490
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990090963942709e-05,
+      "loss": 0.5592,
+      "step": 14491
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990088656757158e-05,
+      "loss": 0.4298,
+      "step": 14492
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900863493043777e-05,
+      "loss": 0.4898,
+      "step": 14493
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900840415843684e-05,
+      "loss": 0.4995,
+      "step": 14494
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900817335971306e-05,
+      "loss": 0.5097,
+      "step": 14495
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900794253426655e-05,
+      "loss": 0.5013,
+      "step": 14496
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900771168209734e-05,
+      "loss": 0.44,
+      "step": 14497
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900748080320545e-05,
+      "loss": 0.4916,
+      "step": 14498
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.99007249897591e-05,
+      "loss": 0.5179,
+      "step": 14499
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900701896525404e-05,
+      "loss": 0.5839,
+      "step": 14500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990067880061946e-05,
+      "loss": 0.5091,
+      "step": 14501
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900655702041277e-05,
+      "loss": 0.3842,
+      "step": 14502
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990063260079086e-05,
+      "loss": 0.4974,
+      "step": 14503
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900609496868214e-05,
+      "loss": 0.4677,
+      "step": 14504
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990058639027335e-05,
+      "loss": 0.5906,
+      "step": 14505
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900563281006273e-05,
+      "loss": 0.5046,
+      "step": 14506
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900540169066987e-05,
+      "loss": 0.4547,
+      "step": 14507
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900517054455498e-05,
+      "loss": 0.4676,
+      "step": 14508
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900493937171813e-05,
+      "loss": 0.4669,
+      "step": 14509
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990047081721594e-05,
+      "loss": 0.465,
+      "step": 14510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990044769458788e-05,
+      "loss": 0.4632,
+      "step": 14511
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900424569287648e-05,
+      "loss": 0.4677,
+      "step": 14512
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900401441315237e-05,
+      "loss": 0.5404,
+      "step": 14513
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990037831067067e-05,
+      "loss": 0.5172,
+      "step": 14514
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990035517735394e-05,
+      "loss": 0.5206,
+      "step": 14515
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990033204136506e-05,
+      "loss": 0.4616,
+      "step": 14516
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900308902704034e-05,
+      "loss": 0.4938,
+      "step": 14517
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900285761370868e-05,
+      "loss": 0.4621,
+      "step": 14518
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900262617365567e-05,
+      "loss": 0.5372,
+      "step": 14519
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990023947068814e-05,
+      "loss": 0.5304,
+      "step": 14520
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.990021632133859e-05,
+      "loss": 0.4698,
+      "step": 14521
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900193169316926e-05,
+      "loss": 0.5608,
+      "step": 14522
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900170014623157e-05,
+      "loss": 0.4856,
+      "step": 14523
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900146857257284e-05,
+      "loss": 0.5542,
+      "step": 14524
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900123697219316e-05,
+      "loss": 0.4798,
+      "step": 14525
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900100534509258e-05,
+      "loss": 0.4612,
+      "step": 14526
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900077369127112e-05,
+      "loss": 0.4342,
+      "step": 14527
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900054201072896e-05,
+      "loss": 0.4524,
+      "step": 14528
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900031030346606e-05,
+      "loss": 0.4709,
+      "step": 14529
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9900007856948252e-05,
+      "loss": 0.4732,
+      "step": 14530
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9899984680877838e-05,
+      "loss": 0.5154,
+      "step": 14531
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.989996150213537e-05,
+      "loss": 0.497,
+      "step": 14532
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9899938320720863e-05,
+      "loss": 0.5567,
+      "step": 14533
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.989991513663431e-05,
+      "loss": 0.4365,
+      "step": 14534
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9899891949875725e-05,
+      "loss": 0.5083,
+      "step": 14535
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9899868760445114e-05,
+      "loss": 0.4608,
+      "step": 14536
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.989984556834248e-05,
+      "loss": 0.4481,
+      "step": 14537
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9899822373567834e-05,
+      "loss": 0.4313,
+      "step": 14538
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9899799176121178e-05,
+      "loss": 0.4488,
+      "step": 14539
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899775976002523e-05,
+      "loss": 0.4391,
+      "step": 14540
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989975277321187e-05,
+      "loss": 0.4291,
+      "step": 14541
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899729567749225e-05,
+      "loss": 0.4621,
+      "step": 14542
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899706359614602e-05,
+      "loss": 0.4552,
+      "step": 14543
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899683148807997e-05,
+      "loss": 0.4435,
+      "step": 14544
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899659935329427e-05,
+      "loss": 0.415,
+      "step": 14545
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899636719178884e-05,
+      "loss": 0.4656,
+      "step": 14546
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989961350035639e-05,
+      "loss": 0.5904,
+      "step": 14547
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899590278861943e-05,
+      "loss": 0.444,
+      "step": 14548
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899567054695548e-05,
+      "loss": 0.5013,
+      "step": 14549
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989954382785721e-05,
+      "loss": 0.4522,
+      "step": 14550
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899520598346947e-05,
+      "loss": 0.5409,
+      "step": 14551
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899497366164754e-05,
+      "loss": 0.4153,
+      "step": 14552
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989947413131064e-05,
+      "loss": 0.5151,
+      "step": 14553
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989945089378461e-05,
+      "loss": 0.471,
+      "step": 14554
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899427653586673e-05,
+      "loss": 0.5217,
+      "step": 14555
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899404410716836e-05,
+      "loss": 0.4529,
+      "step": 14556
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.98993811651751e-05,
+      "loss": 0.4459,
+      "step": 14557
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989935791696148e-05,
+      "loss": 0.4809,
+      "step": 14558
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899334666075973e-05,
+      "loss": 0.4339,
+      "step": 14559
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989931141251859e-05,
+      "loss": 0.4501,
+      "step": 14560
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899288156289337e-05,
+      "loss": 0.4304,
+      "step": 14561
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989926489738822e-05,
+      "loss": 0.5581,
+      "step": 14562
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899241635815242e-05,
+      "loss": 0.5192,
+      "step": 14563
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899218371570416e-05,
+      "loss": 0.5425,
+      "step": 14564
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899195104653747e-05,
+      "loss": 0.4314,
+      "step": 14565
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899171835065234e-05,
+      "loss": 0.5059,
+      "step": 14566
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989914856280489e-05,
+      "loss": 0.5452,
+      "step": 14567
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899125287872718e-05,
+      "loss": 0.4503,
+      "step": 14568
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899102010268726e-05,
+      "loss": 0.4165,
+      "step": 14569
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899078729992923e-05,
+      "loss": 0.453,
+      "step": 14570
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989905544704531e-05,
+      "loss": 0.4782,
+      "step": 14571
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9899032161425897e-05,
+      "loss": 0.4552,
+      "step": 14572
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989900887313469e-05,
+      "loss": 0.4741,
+      "step": 14573
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898985582171694e-05,
+      "loss": 0.5345,
+      "step": 14574
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989896228853691e-05,
+      "loss": 0.5025,
+      "step": 14575
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898938992230357e-05,
+      "loss": 0.463,
+      "step": 14576
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898915693252033e-05,
+      "loss": 0.4271,
+      "step": 14577
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898892391601944e-05,
+      "loss": 0.4668,
+      "step": 14578
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898869087280096e-05,
+      "loss": 0.4222,
+      "step": 14579
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898845780286497e-05,
+      "loss": 0.4247,
+      "step": 14580
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898822470621155e-05,
+      "loss": 0.4057,
+      "step": 14581
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989879915828407e-05,
+      "loss": 0.4829,
+      "step": 14582
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989877584327526e-05,
+      "loss": 0.4953,
+      "step": 14583
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898752525594723e-05,
+      "loss": 0.5589,
+      "step": 14584
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898729205242462e-05,
+      "loss": 0.5513,
+      "step": 14585
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989870588221849e-05,
+      "loss": 0.4406,
+      "step": 14586
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898682556522812e-05,
+      "loss": 0.4687,
+      "step": 14587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898659228155434e-05,
+      "loss": 0.5452,
+      "step": 14588
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898635897116357e-05,
+      "loss": 0.5248,
+      "step": 14589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898612563405594e-05,
+      "loss": 0.4433,
+      "step": 14590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898589227023152e-05,
+      "loss": 0.4495,
+      "step": 14591
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898565887969034e-05,
+      "loss": 0.498,
+      "step": 14592
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989854254624324e-05,
+      "loss": 0.5366,
+      "step": 14593
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989851920184579e-05,
+      "loss": 0.4694,
+      "step": 14594
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898495854776684e-05,
+      "loss": 0.5269,
+      "step": 14595
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898472505035923e-05,
+      "loss": 0.4811,
+      "step": 14596
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898449152623522e-05,
+      "loss": 0.4067,
+      "step": 14597
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989842579753948e-05,
+      "loss": 0.47,
+      "step": 14598
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898402439783814e-05,
+      "loss": 0.4432,
+      "step": 14599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898379079356513e-05,
+      "loss": 0.3862,
+      "step": 14600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.98983557162576e-05,
+      "loss": 0.5171,
+      "step": 14601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989833235048707e-05,
+      "loss": 0.4921,
+      "step": 14602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898308982044935e-05,
+      "loss": 0.4895,
+      "step": 14603
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.98982856109312e-05,
+      "loss": 0.5103,
+      "step": 14604
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898262237145876e-05,
+      "loss": 0.4218,
+      "step": 14605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989823886068896e-05,
+      "loss": 0.4252,
+      "step": 14606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898215481560467e-05,
+      "loss": 0.4534,
+      "step": 14607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898192099760396e-05,
+      "loss": 0.4789,
+      "step": 14608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898168715288757e-05,
+      "loss": 0.4511,
+      "step": 14609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898145328145556e-05,
+      "loss": 0.511,
+      "step": 14610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898121938330803e-05,
+      "loss": 0.5523,
+      "step": 14611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898098545844494e-05,
+      "loss": 0.376,
+      "step": 14612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898075150686646e-05,
+      "loss": 0.454,
+      "step": 14613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989805175285726e-05,
+      "loss": 0.5106,
+      "step": 14614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898028352356343e-05,
+      "loss": 0.439,
+      "step": 14615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9898004949183904e-05,
+      "loss": 0.4096,
+      "step": 14616
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897981543339946e-05,
+      "loss": 0.4463,
+      "step": 14617
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897958134824477e-05,
+      "loss": 0.4507,
+      "step": 14618
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897934723637504e-05,
+      "loss": 0.458,
+      "step": 14619
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989791130977903e-05,
+      "loss": 0.5735,
+      "step": 14620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897887893249063e-05,
+      "loss": 0.4378,
+      "step": 14621
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989786447404761e-05,
+      "loss": 0.4831,
+      "step": 14622
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897841052174678e-05,
+      "loss": 0.4774,
+      "step": 14623
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897817627630273e-05,
+      "loss": 0.5334,
+      "step": 14624
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.98977942004144e-05,
+      "loss": 0.4973,
+      "step": 14625
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897770770527067e-05,
+      "loss": 0.4745,
+      "step": 14626
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989774733796828e-05,
+      "loss": 0.4638,
+      "step": 14627
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989772390273804e-05,
+      "loss": 0.4993,
+      "step": 14628
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989770046483636e-05,
+      "loss": 0.4785,
+      "step": 14629
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989767702426325e-05,
+      "loss": 0.4703,
+      "step": 14630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897653581018705e-05,
+      "loss": 0.4475,
+      "step": 14631
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989763013510274e-05,
+      "loss": 0.383,
+      "step": 14632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897606686515356e-05,
+      "loss": 0.4246,
+      "step": 14633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897583235256565e-05,
+      "loss": 0.485,
+      "step": 14634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897559781326364e-05,
+      "loss": 0.4733,
+      "step": 14635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.989753632472477e-05,
+      "loss": 0.4156,
+      "step": 14636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9897512865451784e-05,
+      "loss": 0.4523,
+      "step": 14637
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989748940350741e-05,
+      "loss": 0.4298,
+      "step": 14638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897465938891663e-05,
+      "loss": 0.4316,
+      "step": 14639
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989744247160454e-05,
+      "loss": 0.5727,
+      "step": 14640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989741900164605e-05,
+      "loss": 0.4323,
+      "step": 14641
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897395529016203e-05,
+      "loss": 0.4007,
+      "step": 14642
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897372053715e-05,
+      "loss": 0.5143,
+      "step": 14643
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897348575742455e-05,
+      "loss": 0.5046,
+      "step": 14644
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897325095098566e-05,
+      "loss": 0.4522,
+      "step": 14645
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897301611783342e-05,
+      "loss": 0.4639,
+      "step": 14646
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989727812579679e-05,
+      "loss": 0.4749,
+      "step": 14647
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989725463713892e-05,
+      "loss": 0.4876,
+      "step": 14648
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989723114580973e-05,
+      "loss": 0.4486,
+      "step": 14649
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897207651809233e-05,
+      "loss": 0.5057,
+      "step": 14650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897184155137435e-05,
+      "loss": 0.4791,
+      "step": 14651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897160655794335e-05,
+      "loss": 0.4207,
+      "step": 14652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897137153779954e-05,
+      "loss": 0.4371,
+      "step": 14653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897113649094282e-05,
+      "loss": 0.4704,
+      "step": 14654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897090141737336e-05,
+      "loss": 0.5199,
+      "step": 14655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897066631709115e-05,
+      "loss": 0.5915,
+      "step": 14656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897043119009637e-05,
+      "loss": 0.5182,
+      "step": 14657
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9897019603638895e-05,
+      "loss": 0.4548,
+      "step": 14658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896996085596903e-05,
+      "loss": 0.4692,
+      "step": 14659
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896972564883667e-05,
+      "loss": 0.5395,
+      "step": 14660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896949041499187e-05,
+      "loss": 0.4578,
+      "step": 14661
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989692551544348e-05,
+      "loss": 0.5062,
+      "step": 14662
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896901986716543e-05,
+      "loss": 0.4468,
+      "step": 14663
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896878455318386e-05,
+      "loss": 0.4192,
+      "step": 14664
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896854921249015e-05,
+      "loss": 0.4148,
+      "step": 14665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989683138450844e-05,
+      "loss": 0.4393,
+      "step": 14666
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896807845096662e-05,
+      "loss": 0.5543,
+      "step": 14667
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989678430301369e-05,
+      "loss": 0.431,
+      "step": 14668
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896760758259527e-05,
+      "loss": 0.4148,
+      "step": 14669
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989673721083418e-05,
+      "loss": 0.5357,
+      "step": 14670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896713660737664e-05,
+      "loss": 0.4081,
+      "step": 14671
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896690107969973e-05,
+      "loss": 0.48,
+      "step": 14672
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896666552531124e-05,
+      "loss": 0.4906,
+      "step": 14673
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896642994421115e-05,
+      "loss": 0.4786,
+      "step": 14674
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896619433639958e-05,
+      "loss": 0.4094,
+      "step": 14675
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896595870187655e-05,
+      "loss": 0.4953,
+      "step": 14676
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896572304064217e-05,
+      "loss": 0.5088,
+      "step": 14677
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896548735269646e-05,
+      "loss": 0.4223,
+      "step": 14678
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989652516380395e-05,
+      "loss": 0.4507,
+      "step": 14679
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896501589667137e-05,
+      "loss": 0.5145,
+      "step": 14680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896478012859212e-05,
+      "loss": 0.4662,
+      "step": 14681
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896454433380178e-05,
+      "loss": 0.4764,
+      "step": 14682
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989643085123005e-05,
+      "loss": 0.4292,
+      "step": 14683
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896407266408824e-05,
+      "loss": 0.4408,
+      "step": 14684
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896383678916516e-05,
+      "loss": 0.488,
+      "step": 14685
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896360088753127e-05,
+      "loss": 0.4592,
+      "step": 14686
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989633649591866e-05,
+      "loss": 0.4748,
+      "step": 14687
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896312900413127e-05,
+      "loss": 0.4539,
+      "step": 14688
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896289302236536e-05,
+      "loss": 0.3876,
+      "step": 14689
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896265701388887e-05,
+      "loss": 0.5028,
+      "step": 14690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989624209787019e-05,
+      "loss": 0.4624,
+      "step": 14691
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896218491680454e-05,
+      "loss": 0.4408,
+      "step": 14692
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989619488281968e-05,
+      "loss": 0.5237,
+      "step": 14693
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896171271287877e-05,
+      "loss": 0.5076,
+      "step": 14694
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896147657085053e-05,
+      "loss": 0.5383,
+      "step": 14695
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989612404021121e-05,
+      "loss": 0.5017,
+      "step": 14696
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896100420666354e-05,
+      "loss": 0.4996,
+      "step": 14697
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.98960767984505e-05,
+      "loss": 0.4627,
+      "step": 14698
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896053173563646e-05,
+      "loss": 0.4726,
+      "step": 14699
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896029546005803e-05,
+      "loss": 0.534,
+      "step": 14700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9896005915776973e-05,
+      "loss": 0.4938,
+      "step": 14701
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895982282877164e-05,
+      "loss": 0.5018,
+      "step": 14702
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895958647306386e-05,
+      "loss": 0.42,
+      "step": 14703
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989593500906464e-05,
+      "loss": 0.5458,
+      "step": 14704
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895911368151937e-05,
+      "loss": 0.4628,
+      "step": 14705
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895887724568278e-05,
+      "loss": 0.5004,
+      "step": 14706
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895864078313677e-05,
+      "loss": 0.527,
+      "step": 14707
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895840429388133e-05,
+      "loss": 0.4651,
+      "step": 14708
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895816777791657e-05,
+      "loss": 0.4934,
+      "step": 14709
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895793123524256e-05,
+      "loss": 0.3693,
+      "step": 14710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989576946658593e-05,
+      "loss": 0.4197,
+      "step": 14711
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989574580697669e-05,
+      "loss": 0.4837,
+      "step": 14712
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989572214469654e-05,
+      "loss": 0.5465,
+      "step": 14713
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895698479745493e-05,
+      "loss": 0.4285,
+      "step": 14714
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989567481212355e-05,
+      "loss": 0.5084,
+      "step": 14715
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895651141830716e-05,
+      "loss": 0.4955,
+      "step": 14716
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895627468867e-05,
+      "loss": 0.3871,
+      "step": 14717
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989560379323241e-05,
+      "loss": 0.4483,
+      "step": 14718
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895580114926946e-05,
+      "loss": 0.4249,
+      "step": 14719
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895556433950625e-05,
+      "loss": 0.4681,
+      "step": 14720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895532750303442e-05,
+      "loss": 0.4673,
+      "step": 14721
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989550906398541e-05,
+      "loss": 0.4771,
+      "step": 14722
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895485374996534e-05,
+      "loss": 0.4303,
+      "step": 14723
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989546168333682e-05,
+      "loss": 0.4949,
+      "step": 14724
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895437989006275e-05,
+      "loss": 0.4372,
+      "step": 14725
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895414292004906e-05,
+      "loss": 0.4883,
+      "step": 14726
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895390592332714e-05,
+      "loss": 0.4109,
+      "step": 14727
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895366889989715e-05,
+      "loss": 0.5101,
+      "step": 14728
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895343184975905e-05,
+      "loss": 0.4586,
+      "step": 14729
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.98953194772913e-05,
+      "loss": 0.5201,
+      "step": 14730
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.98952957669359e-05,
+      "loss": 0.5238,
+      "step": 14731
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895272053909716e-05,
+      "loss": 0.4465,
+      "step": 14732
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895248338212748e-05,
+      "loss": 0.4338,
+      "step": 14733
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.989522461984501e-05,
+      "loss": 0.4669,
+      "step": 14734
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.98952008988065e-05,
+      "loss": 0.4945,
+      "step": 14735
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9895177175097234e-05,
+      "loss": 0.4603,
+      "step": 14736
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9895153448717212e-05,
+      "loss": 0.4886,
+      "step": 14737
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989512971966644e-05,
+      "loss": 0.4259,
+      "step": 14738
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9895105987944927e-05,
+      "loss": 0.5246,
+      "step": 14739
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9895082253552678e-05,
+      "loss": 0.4513,
+      "step": 14740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9895058516489704e-05,
+      "loss": 0.491,
+      "step": 14741
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9895034776756e-05,
+      "loss": 0.4996,
+      "step": 14742
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9895011034351586e-05,
+      "loss": 0.3988,
+      "step": 14743
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894987289276462e-05,
+      "loss": 0.4883,
+      "step": 14744
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894963541530634e-05,
+      "loss": 0.4422,
+      "step": 14745
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894939791114107e-05,
+      "loss": 0.4458,
+      "step": 14746
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989491603802689e-05,
+      "loss": 0.4099,
+      "step": 14747
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894892282268993e-05,
+      "loss": 0.5171,
+      "step": 14748
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894868523840412e-05,
+      "loss": 0.4259,
+      "step": 14749
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894844762741163e-05,
+      "loss": 0.3627,
+      "step": 14750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989482099897125e-05,
+      "loss": 0.4801,
+      "step": 14751
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894797232530676e-05,
+      "loss": 0.4428,
+      "step": 14752
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989477346341945e-05,
+      "loss": 0.4964,
+      "step": 14753
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894749691637582e-05,
+      "loss": 0.4836,
+      "step": 14754
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894725917185073e-05,
+      "loss": 0.4918,
+      "step": 14755
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989470214006193e-05,
+      "loss": 0.4323,
+      "step": 14756
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894678360268162e-05,
+      "loss": 0.4686,
+      "step": 14757
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894654577803774e-05,
+      "loss": 0.3963,
+      "step": 14758
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894630792668773e-05,
+      "loss": 0.4532,
+      "step": 14759
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894607004863162e-05,
+      "loss": 0.4372,
+      "step": 14760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894583214386954e-05,
+      "loss": 0.4249,
+      "step": 14761
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989455942124015e-05,
+      "loss": 0.4314,
+      "step": 14762
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894535625422758e-05,
+      "loss": 0.4899,
+      "step": 14763
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894511826934784e-05,
+      "loss": 0.4929,
+      "step": 14764
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894488025776234e-05,
+      "loss": 0.4478,
+      "step": 14765
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894464221947117e-05,
+      "loss": 0.4634,
+      "step": 14766
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989444041544744e-05,
+      "loss": 0.4676,
+      "step": 14767
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894416606277206e-05,
+      "loss": 0.4646,
+      "step": 14768
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989439279443642e-05,
+      "loss": 0.4854,
+      "step": 14769
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894368979925095e-05,
+      "loss": 0.547,
+      "step": 14770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894345162743232e-05,
+      "loss": 0.529,
+      "step": 14771
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894321342890837e-05,
+      "loss": 0.4417,
+      "step": 14772
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989429752036792e-05,
+      "loss": 0.5046,
+      "step": 14773
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894273695174485e-05,
+      "loss": 0.5172,
+      "step": 14774
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989424986731054e-05,
+      "loss": 0.531,
+      "step": 14775
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989422603677609e-05,
+      "loss": 0.4888,
+      "step": 14776
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894202203571144e-05,
+      "loss": 0.5175,
+      "step": 14777
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894178367695708e-05,
+      "loss": 0.5105,
+      "step": 14778
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894154529149784e-05,
+      "loss": 0.492,
+      "step": 14779
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894130687933382e-05,
+      "loss": 0.4536,
+      "step": 14780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989410684404651e-05,
+      "loss": 0.604,
+      "step": 14781
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989408299748917e-05,
+      "loss": 0.4737,
+      "step": 14782
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989405914826137e-05,
+      "loss": 0.4695,
+      "step": 14783
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894035296363115e-05,
+      "loss": 0.4239,
+      "step": 14784
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9894011441794418e-05,
+      "loss": 0.4652,
+      "step": 14785
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893987584555283e-05,
+      "loss": 0.4314,
+      "step": 14786
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989396372464571e-05,
+      "loss": 0.4366,
+      "step": 14787
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989393986206571e-05,
+      "loss": 0.5048,
+      "step": 14788
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893915996815295e-05,
+      "loss": 0.4621,
+      "step": 14789
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893892128894458e-05,
+      "loss": 0.419,
+      "step": 14790
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893868258303218e-05,
+      "loss": 0.4724,
+      "step": 14791
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893844385041574e-05,
+      "loss": 0.3823,
+      "step": 14792
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989382050910954e-05,
+      "loss": 0.4574,
+      "step": 14793
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893796630507113e-05,
+      "loss": 0.5115,
+      "step": 14794
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893772749234306e-05,
+      "loss": 0.5014,
+      "step": 14795
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893748865291122e-05,
+      "loss": 0.6146,
+      "step": 14796
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989372497867757e-05,
+      "loss": 0.5617,
+      "step": 14797
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893701089393653e-05,
+      "loss": 0.4825,
+      "step": 14798
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893677197439385e-05,
+      "loss": 0.4561,
+      "step": 14799
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893653302814764e-05,
+      "loss": 0.5016,
+      "step": 14800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.98936294055198e-05,
+      "loss": 0.5448,
+      "step": 14801
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893605505554497e-05,
+      "loss": 0.5451,
+      "step": 14802
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989358160291887e-05,
+      "loss": 0.4976,
+      "step": 14803
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989355769761291e-05,
+      "loss": 0.5052,
+      "step": 14804
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893533789636638e-05,
+      "loss": 0.5031,
+      "step": 14805
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893509878990053e-05,
+      "loss": 0.5413,
+      "step": 14806
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893485965673167e-05,
+      "loss": 0.502,
+      "step": 14807
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989346204968598e-05,
+      "loss": 0.4117,
+      "step": 14808
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893438131028505e-05,
+      "loss": 0.4667,
+      "step": 14809
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989341420970074e-05,
+      "loss": 0.475,
+      "step": 14810
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893390285702696e-05,
+      "loss": 0.564,
+      "step": 14811
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989336635903438e-05,
+      "loss": 0.5006,
+      "step": 14812
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893342429695802e-05,
+      "loss": 0.4996,
+      "step": 14813
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893318497686965e-05,
+      "loss": 0.5172,
+      "step": 14814
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893294563007874e-05,
+      "loss": 0.4513,
+      "step": 14815
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893270625658534e-05,
+      "loss": 0.389,
+      "step": 14816
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893246685638954e-05,
+      "loss": 0.4657,
+      "step": 14817
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989322274294914e-05,
+      "loss": 0.463,
+      "step": 14818
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893198797589102e-05,
+      "loss": 0.4458,
+      "step": 14819
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893174849558844e-05,
+      "loss": 0.4308,
+      "step": 14820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893150898858368e-05,
+      "loss": 0.5113,
+      "step": 14821
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989312694548769e-05,
+      "loss": 0.4667,
+      "step": 14822
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893102989446806e-05,
+      "loss": 0.4304,
+      "step": 14823
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893079030735726e-05,
+      "loss": 0.4608,
+      "step": 14824
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9893055069354463e-05,
+      "loss": 0.504,
+      "step": 14825
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989303110530301e-05,
+      "loss": 0.5134,
+      "step": 14826
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989300713858139e-05,
+      "loss": 0.4971,
+      "step": 14827
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9892983169189598e-05,
+      "loss": 0.5034,
+      "step": 14828
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9892959197127642e-05,
+      "loss": 0.4886,
+      "step": 14829
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9892935222395534e-05,
+      "loss": 0.4563,
+      "step": 14830
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9892911244993273e-05,
+      "loss": 0.4144,
+      "step": 14831
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989288726492087e-05,
+      "loss": 0.4457,
+      "step": 14832
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989286328217833e-05,
+      "loss": 0.4033,
+      "step": 14833
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.989283929676566e-05,
+      "loss": 0.4495,
+      "step": 14834
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989281530868287e-05,
+      "loss": 0.4658,
+      "step": 14835
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989279131792996e-05,
+      "loss": 0.521,
+      "step": 14836
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892767324506938e-05,
+      "loss": 0.5195,
+      "step": 14837
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892743328413814e-05,
+      "loss": 0.3984,
+      "step": 14838
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892719329650592e-05,
+      "loss": 0.5297,
+      "step": 14839
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892695328217278e-05,
+      "loss": 0.4205,
+      "step": 14840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989267132411388e-05,
+      "loss": 0.5399,
+      "step": 14841
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892647317340404e-05,
+      "loss": 0.4847,
+      "step": 14842
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892623307896855e-05,
+      "loss": 0.5489,
+      "step": 14843
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989259929578324e-05,
+      "loss": 0.4334,
+      "step": 14844
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989257528099957e-05,
+      "loss": 0.4657,
+      "step": 14845
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892551263545844e-05,
+      "loss": 0.5033,
+      "step": 14846
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892527243422074e-05,
+      "loss": 0.565,
+      "step": 14847
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892503220628262e-05,
+      "loss": 0.5036,
+      "step": 14848
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989247919516442e-05,
+      "loss": 0.5472,
+      "step": 14849
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989245516703055e-05,
+      "loss": 0.4439,
+      "step": 14850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892431136226663e-05,
+      "loss": 0.4494,
+      "step": 14851
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989240710275276e-05,
+      "loss": 0.4559,
+      "step": 14852
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989238306660885e-05,
+      "loss": 0.4176,
+      "step": 14853
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989235902779494e-05,
+      "loss": 0.4107,
+      "step": 14854
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892334986311038e-05,
+      "loss": 0.495,
+      "step": 14855
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892310942157146e-05,
+      "loss": 0.4866,
+      "step": 14856
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892286895333274e-05,
+      "loss": 0.4326,
+      "step": 14857
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989226284583943e-05,
+      "loss": 0.4495,
+      "step": 14858
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892238793675614e-05,
+      "loss": 0.4471,
+      "step": 14859
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892214738841836e-05,
+      "loss": 0.5048,
+      "step": 14860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892190681338105e-05,
+      "loss": 0.4553,
+      "step": 14861
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892166621164425e-05,
+      "loss": 0.4822,
+      "step": 14862
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892142558320805e-05,
+      "loss": 0.4348,
+      "step": 14863
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892118492807246e-05,
+      "loss": 0.4416,
+      "step": 14864
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989209442462376e-05,
+      "loss": 0.5381,
+      "step": 14865
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989207035377035e-05,
+      "loss": 0.4527,
+      "step": 14866
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989204628024703e-05,
+      "loss": 0.4757,
+      "step": 14867
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9892022204053797e-05,
+      "loss": 0.5006,
+      "step": 14868
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891998125190656e-05,
+      "loss": 0.4417,
+      "step": 14869
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891974043657624e-05,
+      "loss": 0.4212,
+      "step": 14870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.98919499594547e-05,
+      "loss": 0.4742,
+      "step": 14871
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989192587258189e-05,
+      "loss": 0.5703,
+      "step": 14872
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989190178303921e-05,
+      "loss": 0.4769,
+      "step": 14873
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891877690826654e-05,
+      "loss": 0.4314,
+      "step": 14874
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891853595944233e-05,
+      "loss": 0.4638,
+      "step": 14875
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891829498391958e-05,
+      "loss": 0.5154,
+      "step": 14876
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989180539816983e-05,
+      "loss": 0.4543,
+      "step": 14877
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989178129527786e-05,
+      "loss": 0.481,
+      "step": 14878
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891757189716048e-05,
+      "loss": 0.4151,
+      "step": 14879
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989173308148441e-05,
+      "loss": 0.4291,
+      "step": 14880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989170897058294e-05,
+      "loss": 0.5196,
+      "step": 14881
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891684857011654e-05,
+      "loss": 0.4651,
+      "step": 14882
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989166074077056e-05,
+      "loss": 0.4257,
+      "step": 14883
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891636621859656e-05,
+      "loss": 0.5698,
+      "step": 14884
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891612500278954e-05,
+      "loss": 0.5234,
+      "step": 14885
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989158837602846e-05,
+      "loss": 0.4966,
+      "step": 14886
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989156424910818e-05,
+      "loss": 0.4512,
+      "step": 14887
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989154011951812e-05,
+      "loss": 0.5181,
+      "step": 14888
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989151598725829e-05,
+      "loss": 0.4307,
+      "step": 14889
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891491852328692e-05,
+      "loss": 0.4172,
+      "step": 14890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891467714729332e-05,
+      "loss": 0.5118,
+      "step": 14891
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989144357446022e-05,
+      "loss": 0.4626,
+      "step": 14892
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989141943152136e-05,
+      "loss": 0.4364,
+      "step": 14893
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891395285912763e-05,
+      "loss": 0.4499,
+      "step": 14894
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989137113763443e-05,
+      "loss": 0.5617,
+      "step": 14895
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989134698668637e-05,
+      "loss": 0.4724,
+      "step": 14896
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989132283306859e-05,
+      "loss": 0.5173,
+      "step": 14897
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891298676781092e-05,
+      "loss": 0.4691,
+      "step": 14898
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989127451782389e-05,
+      "loss": 0.4832,
+      "step": 14899
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891250356196986e-05,
+      "loss": 0.4175,
+      "step": 14900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891226191900388e-05,
+      "loss": 0.3928,
+      "step": 14901
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.98912020249341e-05,
+      "loss": 0.5225,
+      "step": 14902
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891177855298133e-05,
+      "loss": 0.4799,
+      "step": 14903
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989115368299249e-05,
+      "loss": 0.4057,
+      "step": 14904
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891129508017174e-05,
+      "loss": 0.4561,
+      "step": 14905
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.98911053303722e-05,
+      "loss": 0.4436,
+      "step": 14906
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989108115005757e-05,
+      "loss": 0.4041,
+      "step": 14907
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989105696707329e-05,
+      "loss": 0.532,
+      "step": 14908
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891032781419367e-05,
+      "loss": 0.4019,
+      "step": 14909
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9891008593095812e-05,
+      "loss": 0.4723,
+      "step": 14910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9890984402102623e-05,
+      "loss": 0.4448,
+      "step": 14911
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9890960208439813e-05,
+      "loss": 0.5398,
+      "step": 14912
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9890936012107385e-05,
+      "loss": 0.5062,
+      "step": 14913
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9890911813105354e-05,
+      "loss": 0.423,
+      "step": 14914
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9890887611433712e-05,
+      "loss": 0.4812,
+      "step": 14915
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9890863407092476e-05,
+      "loss": 0.5285,
+      "step": 14916
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989083920008165e-05,
+      "loss": 0.4876,
+      "step": 14917
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989081499040124e-05,
+      "loss": 0.5155,
+      "step": 14918
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989079077805125e-05,
+      "loss": 0.4421,
+      "step": 14919
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989076656303169e-05,
+      "loss": 0.4953,
+      "step": 14920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989074234534257e-05,
+      "loss": 0.5744,
+      "step": 14921
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989071812498389e-05,
+      "loss": 0.5117,
+      "step": 14922
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989069390195566e-05,
+      "loss": 0.4168,
+      "step": 14923
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9890669676257882e-05,
+      "loss": 0.5108,
+      "step": 14924
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989064544789057e-05,
+      "loss": 0.5986,
+      "step": 14925
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9890621216853726e-05,
+      "loss": 0.5116,
+      "step": 14926
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9890596983147355e-05,
+      "loss": 0.4817,
+      "step": 14927
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989057274677147e-05,
+      "loss": 0.4667,
+      "step": 14928
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9890548507726067e-05,
+      "loss": 0.5003,
+      "step": 14929
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989052426601116e-05,
+      "loss": 0.4971,
+      "step": 14930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989050002162676e-05,
+      "loss": 0.5501,
+      "step": 14931
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989047577457286e-05,
+      "loss": 0.5212,
+      "step": 14932
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.989045152484948e-05,
+      "loss": 0.4445,
+      "step": 14933
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890427272456617e-05,
+      "loss": 0.4322,
+      "step": 14934
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890403017394286e-05,
+      "loss": 0.4808,
+      "step": 14935
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890378759662483e-05,
+      "loss": 0.4588,
+      "step": 14936
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890354499261225e-05,
+      "loss": 0.5541,
+      "step": 14937
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890330236190513e-05,
+      "loss": 0.4688,
+      "step": 14938
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890305970450356e-05,
+      "loss": 0.3937,
+      "step": 14939
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890281702040757e-05,
+      "loss": 0.444,
+      "step": 14940
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890257430961725e-05,
+      "loss": 0.4556,
+      "step": 14941
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890233157213267e-05,
+      "loss": 0.5213,
+      "step": 14942
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.989020888079539e-05,
+      "loss": 0.4782,
+      "step": 14943
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890184601708097e-05,
+      "loss": 0.4785,
+      "step": 14944
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.98901603199514e-05,
+      "loss": 0.495,
+      "step": 14945
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890136035525297e-05,
+      "loss": 0.4598,
+      "step": 14946
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890111748429806e-05,
+      "loss": 0.5312,
+      "step": 14947
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890087458664925e-05,
+      "loss": 0.3906,
+      "step": 14948
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890063166230663e-05,
+      "loss": 0.51,
+      "step": 14949
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890038871127028e-05,
+      "loss": 0.5644,
+      "step": 14950
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9890014573354023e-05,
+      "loss": 0.5053,
+      "step": 14951
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889990272911658e-05,
+      "loss": 0.4042,
+      "step": 14952
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988996596979994e-05,
+      "loss": 0.5094,
+      "step": 14953
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988994166401887e-05,
+      "loss": 0.5047,
+      "step": 14954
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889917355568464e-05,
+      "loss": 0.5703,
+      "step": 14955
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988989304444872e-05,
+      "loss": 0.4073,
+      "step": 14956
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889868730659646e-05,
+      "loss": 0.4303,
+      "step": 14957
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889844414201253e-05,
+      "loss": 0.433,
+      "step": 14958
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988982009507354e-05,
+      "loss": 0.4828,
+      "step": 14959
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889795773276527e-05,
+      "loss": 0.5118,
+      "step": 14960
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889771448810203e-05,
+      "loss": 0.4772,
+      "step": 14961
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988974712167459e-05,
+      "loss": 0.4415,
+      "step": 14962
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889722791869687e-05,
+      "loss": 0.4775,
+      "step": 14963
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889698459395497e-05,
+      "loss": 0.4835,
+      "step": 14964
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889674124252035e-05,
+      "loss": 0.5404,
+      "step": 14965
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.98896497864393e-05,
+      "loss": 0.5331,
+      "step": 14966
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988962544595731e-05,
+      "loss": 0.4016,
+      "step": 14967
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988960110280606e-05,
+      "loss": 0.458,
+      "step": 14968
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889576756985557e-05,
+      "loss": 0.4675,
+      "step": 14969
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889552408495814e-05,
+      "loss": 0.4891,
+      "step": 14970
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889528057336832e-05,
+      "loss": 0.4618,
+      "step": 14971
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889503703508623e-05,
+      "loss": 0.5975,
+      "step": 14972
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988947934701119e-05,
+      "loss": 0.4274,
+      "step": 14973
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889454987844542e-05,
+      "loss": 0.5529,
+      "step": 14974
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988943062600868e-05,
+      "loss": 0.4412,
+      "step": 14975
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889406261503617e-05,
+      "loss": 0.4753,
+      "step": 14976
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889381894329357e-05,
+      "loss": 0.5379,
+      "step": 14977
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889357524485907e-05,
+      "loss": 0.5096,
+      "step": 14978
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889333151973272e-05,
+      "loss": 0.5058,
+      "step": 14979
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889308776791457e-05,
+      "loss": 0.4487,
+      "step": 14980
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889284398940476e-05,
+      "loss": 0.4531,
+      "step": 14981
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988926001842033e-05,
+      "loss": 0.6011,
+      "step": 14982
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988923563523103e-05,
+      "loss": 0.4916,
+      "step": 14983
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988921124937257e-05,
+      "loss": 0.4635,
+      "step": 14984
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889186860844972e-05,
+      "loss": 0.442,
+      "step": 14985
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889162469648235e-05,
+      "loss": 0.4818,
+      "step": 14986
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889138075782367e-05,
+      "loss": 0.433,
+      "step": 14987
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889113679247373e-05,
+      "loss": 0.6095,
+      "step": 14988
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889089280043264e-05,
+      "loss": 0.4566,
+      "step": 14989
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889064878170038e-05,
+      "loss": 0.5148,
+      "step": 14990
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889040473627713e-05,
+      "loss": 0.4709,
+      "step": 14991
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9889016066416286e-05,
+      "loss": 0.4863,
+      "step": 14992
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988899165653577e-05,
+      "loss": 0.5046,
+      "step": 14993
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988896724398617e-05,
+      "loss": 0.4579,
+      "step": 14994
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888942828767485e-05,
+      "loss": 0.4812,
+      "step": 14995
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888918410879733e-05,
+      "loss": 0.4764,
+      "step": 14996
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988889399032292e-05,
+      "loss": 0.4472,
+      "step": 14997
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988886956709704e-05,
+      "loss": 0.495,
+      "step": 14998
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988884514120211e-05,
+      "loss": 0.467,
+      "step": 14999
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888820712638136e-05,
+      "loss": 0.4949,
+      "step": 15000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888796281405125e-05,
+      "loss": 0.5863,
+      "step": 15001
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988877184750308e-05,
+      "loss": 0.5293,
+      "step": 15002
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888747410932007e-05,
+      "loss": 0.4732,
+      "step": 15003
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888722971691916e-05,
+      "loss": 0.465,
+      "step": 15004
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888698529782814e-05,
+      "loss": 0.4414,
+      "step": 15005
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888674085204705e-05,
+      "loss": 0.4802,
+      "step": 15006
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888649637957598e-05,
+      "loss": 0.4579,
+      "step": 15007
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888625188041494e-05,
+      "loss": 0.4696,
+      "step": 15008
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888600735456405e-05,
+      "loss": 0.4899,
+      "step": 15009
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988857628020234e-05,
+      "loss": 0.4612,
+      "step": 15010
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888551822279304e-05,
+      "loss": 0.4703,
+      "step": 15011
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888527361687295e-05,
+      "loss": 0.5303,
+      "step": 15012
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988850289842633e-05,
+      "loss": 0.5285,
+      "step": 15013
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888478432496413e-05,
+      "loss": 0.4958,
+      "step": 15014
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888453963897543e-05,
+      "loss": 0.5023,
+      "step": 15015
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988842949262974e-05,
+      "loss": 0.4556,
+      "step": 15016
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888405018693e-05,
+      "loss": 0.5519,
+      "step": 15017
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888380542087337e-05,
+      "loss": 0.4995,
+      "step": 15018
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888356062812752e-05,
+      "loss": 0.4421,
+      "step": 15019
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988833158086925e-05,
+      "loss": 0.4744,
+      "step": 15020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988830709625684e-05,
+      "loss": 0.4308,
+      "step": 15021
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888282608975538e-05,
+      "loss": 0.457,
+      "step": 15022
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888258119025336e-05,
+      "loss": 0.4187,
+      "step": 15023
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988823362640625e-05,
+      "loss": 0.4616,
+      "step": 15024
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888209131118283e-05,
+      "loss": 0.4847,
+      "step": 15025
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988818463316144e-05,
+      "loss": 0.4771,
+      "step": 15026
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988816013253573e-05,
+      "loss": 0.4383,
+      "step": 15027
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888135629241162e-05,
+      "loss": 0.4761,
+      "step": 15028
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888111123277738e-05,
+      "loss": 0.473,
+      "step": 15029
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.988808661464547e-05,
+      "loss": 0.4814,
+      "step": 15030
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9888062103344355e-05,
+      "loss": 0.5704,
+      "step": 15031
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988803758937441e-05,
+      "loss": 0.5222,
+      "step": 15032
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9888013072735636e-05,
+      "loss": 0.4607,
+      "step": 15033
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988798855342804e-05,
+      "loss": 0.454,
+      "step": 15034
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887964031451636e-05,
+      "loss": 0.4935,
+      "step": 15035
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887939506806416e-05,
+      "loss": 0.4756,
+      "step": 15036
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887914979492398e-05,
+      "loss": 0.5176,
+      "step": 15037
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988789044950959e-05,
+      "loss": 0.4558,
+      "step": 15038
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887865916857987e-05,
+      "loss": 0.4301,
+      "step": 15039
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887841381537604e-05,
+      "loss": 0.5462,
+      "step": 15040
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988781684354845e-05,
+      "loss": 0.5046,
+      "step": 15041
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887792302890526e-05,
+      "loss": 0.4446,
+      "step": 15042
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887767759563842e-05,
+      "loss": 0.4206,
+      "step": 15043
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.98877432135684e-05,
+      "loss": 0.4504,
+      "step": 15044
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887718664904213e-05,
+      "loss": 0.5994,
+      "step": 15045
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887694113571282e-05,
+      "loss": 0.5245,
+      "step": 15046
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887669559569617e-05,
+      "loss": 0.3957,
+      "step": 15047
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887645002899227e-05,
+      "loss": 0.5451,
+      "step": 15048
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887620443560113e-05,
+      "loss": 0.5263,
+      "step": 15049
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887595881552283e-05,
+      "loss": 0.4156,
+      "step": 15050
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887571316875744e-05,
+      "loss": 0.5037,
+      "step": 15051
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887546749530505e-05,
+      "loss": 0.4475,
+      "step": 15052
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988752217951657e-05,
+      "loss": 0.4335,
+      "step": 15053
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988749760683395e-05,
+      "loss": 0.504,
+      "step": 15054
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887473031482644e-05,
+      "loss": 0.5337,
+      "step": 15055
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887448453462666e-05,
+      "loss": 0.4519,
+      "step": 15056
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887423872774016e-05,
+      "loss": 0.4633,
+      "step": 15057
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887399289416707e-05,
+      "loss": 0.5219,
+      "step": 15058
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988737470339074e-05,
+      "loss": 0.4468,
+      "step": 15059
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887350114696127e-05,
+      "loss": 0.5607,
+      "step": 15060
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988732552333287e-05,
+      "loss": 0.5052,
+      "step": 15061
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988730092930098e-05,
+      "loss": 0.5112,
+      "step": 15062
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988727633260046e-05,
+      "loss": 0.4451,
+      "step": 15063
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887251733231317e-05,
+      "loss": 0.5155,
+      "step": 15064
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887227131193557e-05,
+      "loss": 0.4449,
+      "step": 15065
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887202526487193e-05,
+      "loss": 0.5135,
+      "step": 15066
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887177919112224e-05,
+      "loss": 0.4485,
+      "step": 15067
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887153309068658e-05,
+      "loss": 0.5225,
+      "step": 15068
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887128696356507e-05,
+      "loss": 0.4656,
+      "step": 15069
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988710408097577e-05,
+      "loss": 0.4328,
+      "step": 15070
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988707946292646e-05,
+      "loss": 0.49,
+      "step": 15071
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988705484220858e-05,
+      "loss": 0.5772,
+      "step": 15072
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887030218822136e-05,
+      "loss": 0.4328,
+      "step": 15073
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9887005592767136e-05,
+      "loss": 0.5179,
+      "step": 15074
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886980964043593e-05,
+      "loss": 0.5903,
+      "step": 15075
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886956332651503e-05,
+      "loss": 0.4718,
+      "step": 15076
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886931698590873e-05,
+      "loss": 0.4878,
+      "step": 15077
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886907061861723e-05,
+      "loss": 0.3968,
+      "step": 15078
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886882422464044e-05,
+      "loss": 0.435,
+      "step": 15079
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988685778039785e-05,
+      "loss": 0.4549,
+      "step": 15080
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988683313566315e-05,
+      "loss": 0.453,
+      "step": 15081
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886808488259944e-05,
+      "loss": 0.4649,
+      "step": 15082
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886783838188247e-05,
+      "loss": 0.4123,
+      "step": 15083
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886759185448057e-05,
+      "loss": 0.3729,
+      "step": 15084
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886734530039385e-05,
+      "loss": 0.4778,
+      "step": 15085
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886709871962237e-05,
+      "loss": 0.5044,
+      "step": 15086
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988668521121662e-05,
+      "loss": 0.4884,
+      "step": 15087
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988666054780254e-05,
+      "loss": 0.5289,
+      "step": 15088
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886635881720006e-05,
+      "loss": 0.477,
+      "step": 15089
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886611212969023e-05,
+      "loss": 0.4508,
+      "step": 15090
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886586541549595e-05,
+      "loss": 0.4742,
+      "step": 15091
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886561867461732e-05,
+      "loss": 0.4561,
+      "step": 15092
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988653719070544e-05,
+      "loss": 0.5625,
+      "step": 15093
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886512511280728e-05,
+      "loss": 0.4313,
+      "step": 15094
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886487829187594e-05,
+      "loss": 0.4696,
+      "step": 15095
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886463144426055e-05,
+      "loss": 0.5663,
+      "step": 15096
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886438456996112e-05,
+      "loss": 0.4959,
+      "step": 15097
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988641376689777e-05,
+      "loss": 0.4781,
+      "step": 15098
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886389074131046e-05,
+      "loss": 0.5757,
+      "step": 15099
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886364378695933e-05,
+      "loss": 0.4437,
+      "step": 15100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886339680592447e-05,
+      "loss": 0.4479,
+      "step": 15101
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886314979820593e-05,
+      "loss": 0.4776,
+      "step": 15102
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886290276380375e-05,
+      "loss": 0.5497,
+      "step": 15103
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.98862655702718e-05,
+      "loss": 0.5495,
+      "step": 15104
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886240861494875e-05,
+      "loss": 0.5207,
+      "step": 15105
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988621615004961e-05,
+      "loss": 0.441,
+      "step": 15106
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886191435936008e-05,
+      "loss": 0.5202,
+      "step": 15107
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886166719154076e-05,
+      "loss": 0.4224,
+      "step": 15108
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886141999703824e-05,
+      "loss": 0.4183,
+      "step": 15109
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886117277585252e-05,
+      "loss": 0.5103,
+      "step": 15110
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886092552798374e-05,
+      "loss": 0.5053,
+      "step": 15111
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988606782534319e-05,
+      "loss": 0.4708,
+      "step": 15112
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886043095219714e-05,
+      "loss": 0.4896,
+      "step": 15113
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9886018362427944e-05,
+      "loss": 0.4744,
+      "step": 15114
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9885993626967895e-05,
+      "loss": 0.5353,
+      "step": 15115
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988596888883957e-05,
+      "loss": 0.5403,
+      "step": 15116
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9885944148042978e-05,
+      "loss": 0.4694,
+      "step": 15117
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988591940457812e-05,
+      "loss": 0.5058,
+      "step": 15118
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9885894658445008e-05,
+      "loss": 0.5373,
+      "step": 15119
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988586990964364e-05,
+      "loss": 0.4908,
+      "step": 15120
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988584515817404e-05,
+      "loss": 0.5773,
+      "step": 15121
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9885820404036196e-05,
+      "loss": 0.5241,
+      "step": 15122
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9885795647230125e-05,
+      "loss": 0.431,
+      "step": 15123
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9885770887755835e-05,
+      "loss": 0.4613,
+      "step": 15124
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9885746125613327e-05,
+      "loss": 0.4353,
+      "step": 15125
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988572136080261e-05,
+      "loss": 0.5708,
+      "step": 15126
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.988569659332369e-05,
+      "loss": 0.5961,
+      "step": 15127
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9885671823176575e-05,
+      "loss": 0.5031,
+      "step": 15128
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9885647050361272e-05,
+      "loss": 0.4974,
+      "step": 15129
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9885622274877785e-05,
+      "loss": 0.4672,
+      "step": 15130
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988559749672612e-05,
+      "loss": 0.4378,
+      "step": 15131
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885572715906292e-05,
+      "loss": 0.4632,
+      "step": 15132
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.98855479324183e-05,
+      "loss": 0.4619,
+      "step": 15133
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988552314626215e-05,
+      "loss": 0.4921,
+      "step": 15134
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988549835743785e-05,
+      "loss": 0.4703,
+      "step": 15135
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988547356594541e-05,
+      "loss": 0.442,
+      "step": 15136
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885448771784833e-05,
+      "loss": 0.4852,
+      "step": 15137
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988542397495613e-05,
+      "loss": 0.4705,
+      "step": 15138
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885399175459302e-05,
+      "loss": 0.429,
+      "step": 15139
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988537437329436e-05,
+      "loss": 0.4518,
+      "step": 15140
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988534956846131e-05,
+      "loss": 0.4621,
+      "step": 15141
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988532476096016e-05,
+      "loss": 0.4988,
+      "step": 15142
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885299950790907e-05,
+      "loss": 0.5059,
+      "step": 15143
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885275137953572e-05,
+      "loss": 0.5135,
+      "step": 15144
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885250322448154e-05,
+      "loss": 0.5315,
+      "step": 15145
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885225504274657e-05,
+      "loss": 0.4816,
+      "step": 15146
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885200683433098e-05,
+      "loss": 0.4561,
+      "step": 15147
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988517585992347e-05,
+      "loss": 0.4548,
+      "step": 15148
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885151033745793e-05,
+      "loss": 0.5299,
+      "step": 15149
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885126204900063e-05,
+      "loss": 0.4183,
+      "step": 15150
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885101373386292e-05,
+      "loss": 0.451,
+      "step": 15151
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885076539204486e-05,
+      "loss": 0.4022,
+      "step": 15152
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885051702354654e-05,
+      "loss": 0.6394,
+      "step": 15153
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885026862836798e-05,
+      "loss": 0.4219,
+      "step": 15154
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9885002020650927e-05,
+      "loss": 0.5303,
+      "step": 15155
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988497717579705e-05,
+      "loss": 0.4501,
+      "step": 15156
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988495232827517e-05,
+      "loss": 0.4389,
+      "step": 15157
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884927478085294e-05,
+      "loss": 0.432,
+      "step": 15158
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988490262522743e-05,
+      "loss": 0.4612,
+      "step": 15159
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884877769701587e-05,
+      "loss": 0.5324,
+      "step": 15160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988485291150777e-05,
+      "loss": 0.4776,
+      "step": 15161
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988482805064598e-05,
+      "loss": 0.4287,
+      "step": 15162
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884803187116234e-05,
+      "loss": 0.4471,
+      "step": 15163
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988477832091853e-05,
+      "loss": 0.4683,
+      "step": 15164
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988475345205288e-05,
+      "loss": 0.5131,
+      "step": 15165
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884728580519285e-05,
+      "loss": 0.4683,
+      "step": 15166
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988470370631776e-05,
+      "loss": 0.4942,
+      "step": 15167
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988467882944831e-05,
+      "loss": 0.5161,
+      "step": 15168
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988465394991093e-05,
+      "loss": 0.5168,
+      "step": 15169
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988462906770564e-05,
+      "loss": 0.4186,
+      "step": 15170
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884604182832445e-05,
+      "loss": 0.427,
+      "step": 15171
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884579295291347e-05,
+      "loss": 0.507,
+      "step": 15172
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884554405082353e-05,
+      "loss": 0.6279,
+      "step": 15173
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884529512205474e-05,
+      "loss": 0.494,
+      "step": 15174
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884504616660714e-05,
+      "loss": 0.4879,
+      "step": 15175
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988447971844808e-05,
+      "loss": 0.4383,
+      "step": 15176
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884454817567578e-05,
+      "loss": 0.4258,
+      "step": 15177
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884429914019218e-05,
+      "loss": 0.4443,
+      "step": 15178
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884405007803e-05,
+      "loss": 0.4207,
+      "step": 15179
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884380098918938e-05,
+      "loss": 0.4213,
+      "step": 15180
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988435518736703e-05,
+      "loss": 0.512,
+      "step": 15181
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884330273147293e-05,
+      "loss": 0.5568,
+      "step": 15182
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884305356259732e-05,
+      "loss": 0.4506,
+      "step": 15183
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884280436704347e-05,
+      "loss": 0.5523,
+      "step": 15184
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884255514481147e-05,
+      "loss": 0.5344,
+      "step": 15185
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884230589590145e-05,
+      "loss": 0.4343,
+      "step": 15186
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884205662031338e-05,
+      "loss": 0.4548,
+      "step": 15187
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988418073180474e-05,
+      "loss": 0.4782,
+      "step": 15188
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884155798910355e-05,
+      "loss": 0.4023,
+      "step": 15189
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884130863348195e-05,
+      "loss": 0.5046,
+      "step": 15190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884105925118257e-05,
+      "loss": 0.4484,
+      "step": 15191
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988408098422055e-05,
+      "loss": 0.4989,
+      "step": 15192
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988405604065509e-05,
+      "loss": 0.4949,
+      "step": 15193
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9884031094421872e-05,
+      "loss": 0.5379,
+      "step": 15194
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988400614552091e-05,
+      "loss": 0.447,
+      "step": 15195
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883981193952208e-05,
+      "loss": 0.5306,
+      "step": 15196
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988395623971577e-05,
+      "loss": 0.4684,
+      "step": 15197
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883931282811612e-05,
+      "loss": 0.4467,
+      "step": 15198
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883906323239734e-05,
+      "loss": 0.4924,
+      "step": 15199
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883881361000135e-05,
+      "loss": 0.4544,
+      "step": 15200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988385639609284e-05,
+      "loss": 0.4926,
+      "step": 15201
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988383142851784e-05,
+      "loss": 0.5547,
+      "step": 15202
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883806458275152e-05,
+      "loss": 0.478,
+      "step": 15203
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988378148536478e-05,
+      "loss": 0.4366,
+      "step": 15204
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883756509786722e-05,
+      "loss": 0.4338,
+      "step": 15205
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883731531540997e-05,
+      "loss": 0.4368,
+      "step": 15206
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883706550627605e-05,
+      "loss": 0.439,
+      "step": 15207
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883681567046554e-05,
+      "loss": 0.523,
+      "step": 15208
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988365658079785e-05,
+      "loss": 0.4515,
+      "step": 15209
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883631591881506e-05,
+      "loss": 0.4912,
+      "step": 15210
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883606600297522e-05,
+      "loss": 0.408,
+      "step": 15211
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883581606045902e-05,
+      "loss": 0.4474,
+      "step": 15212
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988355660912666e-05,
+      "loss": 0.4486,
+      "step": 15213
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.98835316095398e-05,
+      "loss": 0.5362,
+      "step": 15214
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988350660728533e-05,
+      "loss": 0.4647,
+      "step": 15215
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883481602363254e-05,
+      "loss": 0.489,
+      "step": 15216
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988345659477358e-05,
+      "loss": 0.585,
+      "step": 15217
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883431584516313e-05,
+      "loss": 0.474,
+      "step": 15218
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883406571591463e-05,
+      "loss": 0.4259,
+      "step": 15219
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883381555999035e-05,
+      "loss": 0.5381,
+      "step": 15220
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988335653773904e-05,
+      "loss": 0.4023,
+      "step": 15221
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883331516811476e-05,
+      "loss": 0.5203,
+      "step": 15222
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883306493216358e-05,
+      "loss": 0.5276,
+      "step": 15223
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883281466953686e-05,
+      "loss": 0.4866,
+      "step": 15224
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883256438023474e-05,
+      "loss": 0.485,
+      "step": 15225
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988323140642572e-05,
+      "loss": 0.4274,
+      "step": 15226
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.988320637216044e-05,
+      "loss": 0.4619,
+      "step": 15227
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883181335227635e-05,
+      "loss": 0.4362,
+      "step": 15228
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9883156295627312e-05,
+      "loss": 0.4667,
+      "step": 15229
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988313125335948e-05,
+      "loss": 0.3939,
+      "step": 15230
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9883106208424145e-05,
+      "loss": 0.4572,
+      "step": 15231
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988308116082131e-05,
+      "loss": 0.5895,
+      "step": 15232
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9883056110550992e-05,
+      "loss": 0.4763,
+      "step": 15233
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9883031057613186e-05,
+      "loss": 0.5077,
+      "step": 15234
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9883006002007904e-05,
+      "loss": 0.4937,
+      "step": 15235
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882980943735153e-05,
+      "loss": 0.5536,
+      "step": 15236
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882955882794942e-05,
+      "loss": 0.5706,
+      "step": 15237
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882930819187272e-05,
+      "loss": 0.4456,
+      "step": 15238
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882905752912153e-05,
+      "loss": 0.4201,
+      "step": 15239
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882880683969592e-05,
+      "loss": 0.459,
+      "step": 15240
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882855612359595e-05,
+      "loss": 0.4699,
+      "step": 15241
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988283053808217e-05,
+      "loss": 0.4698,
+      "step": 15242
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882805461137323e-05,
+      "loss": 0.427,
+      "step": 15243
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988278038152506e-05,
+      "loss": 0.5383,
+      "step": 15244
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882755299245386e-05,
+      "loss": 0.5075,
+      "step": 15245
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882730214298313e-05,
+      "loss": 0.5727,
+      "step": 15246
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882705126683842e-05,
+      "loss": 0.464,
+      "step": 15247
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882680036401987e-05,
+      "loss": 0.4757,
+      "step": 15248
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882654943452747e-05,
+      "loss": 0.504,
+      "step": 15249
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882629847836135e-05,
+      "loss": 0.4813,
+      "step": 15250
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882604749552153e-05,
+      "loss": 0.4494,
+      "step": 15251
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882579648600813e-05,
+      "loss": 0.4597,
+      "step": 15252
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882554544982112e-05,
+      "loss": 0.4703,
+      "step": 15253
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882529438696067e-05,
+      "loss": 0.449,
+      "step": 15254
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882504329742685e-05,
+      "loss": 0.4553,
+      "step": 15255
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882479218121966e-05,
+      "loss": 0.555,
+      "step": 15256
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882454103833916e-05,
+      "loss": 0.496,
+      "step": 15257
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988242898687855e-05,
+      "loss": 0.468,
+      "step": 15258
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988240386725587e-05,
+      "loss": 0.418,
+      "step": 15259
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988237874496588e-05,
+      "loss": 0.4994,
+      "step": 15260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988235362000859e-05,
+      "loss": 0.4806,
+      "step": 15261
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882328492384007e-05,
+      "loss": 0.4835,
+      "step": 15262
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988230336209214e-05,
+      "loss": 0.5874,
+      "step": 15263
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988227822913299e-05,
+      "loss": 0.5265,
+      "step": 15264
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882253093506568e-05,
+      "loss": 0.4801,
+      "step": 15265
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988222795521288e-05,
+      "loss": 0.4001,
+      "step": 15266
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882202814251935e-05,
+      "loss": 0.5313,
+      "step": 15267
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882177670623734e-05,
+      "loss": 0.4728,
+      "step": 15268
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882152524328288e-05,
+      "loss": 0.5251,
+      "step": 15269
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882127375365603e-05,
+      "loss": 0.4927,
+      "step": 15270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882102223735685e-05,
+      "loss": 0.4179,
+      "step": 15271
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882077069438542e-05,
+      "loss": 0.5097,
+      "step": 15272
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882051912474177e-05,
+      "loss": 0.4982,
+      "step": 15273
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882026752842604e-05,
+      "loss": 0.5171,
+      "step": 15274
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9882001590543825e-05,
+      "loss": 0.5388,
+      "step": 15275
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988197642557785e-05,
+      "loss": 0.5175,
+      "step": 15276
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988195125794468e-05,
+      "loss": 0.5233,
+      "step": 15277
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881926087644327e-05,
+      "loss": 0.4897,
+      "step": 15278
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881900914676792e-05,
+      "loss": 0.5454,
+      "step": 15279
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988187573904209e-05,
+      "loss": 0.4541,
+      "step": 15280
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881850560740222e-05,
+      "loss": 0.4706,
+      "step": 15281
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881825379771194e-05,
+      "loss": 0.5495,
+      "step": 15282
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988180019613502e-05,
+      "loss": 0.502,
+      "step": 15283
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.98817750098317e-05,
+      "loss": 0.5068,
+      "step": 15284
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988174982086124e-05,
+      "loss": 0.5222,
+      "step": 15285
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988172462922365e-05,
+      "loss": 0.4513,
+      "step": 15286
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881699434918944e-05,
+      "loss": 0.4198,
+      "step": 15287
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881674237947112e-05,
+      "loss": 0.5109,
+      "step": 15288
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881649038308173e-05,
+      "loss": 0.5277,
+      "step": 15289
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988162383600213e-05,
+      "loss": 0.441,
+      "step": 15290
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881598631028992e-05,
+      "loss": 0.4562,
+      "step": 15291
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881573423388764e-05,
+      "loss": 0.4412,
+      "step": 15292
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881548213081455e-05,
+      "loss": 0.4361,
+      "step": 15293
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881523000107067e-05,
+      "loss": 0.4778,
+      "step": 15294
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988149778446561e-05,
+      "loss": 0.5098,
+      "step": 15295
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988147256615709e-05,
+      "loss": 0.5235,
+      "step": 15296
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881447345181516e-05,
+      "loss": 0.4031,
+      "step": 15297
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988142212153889e-05,
+      "loss": 0.4751,
+      "step": 15298
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988139689522923e-05,
+      "loss": 0.5257,
+      "step": 15299
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881371666252528e-05,
+      "loss": 0.5033,
+      "step": 15300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.98813464346088e-05,
+      "loss": 0.4516,
+      "step": 15301
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881321200298047e-05,
+      "loss": 0.4681,
+      "step": 15302
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881295963320283e-05,
+      "loss": 0.5326,
+      "step": 15303
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881270723675507e-05,
+      "loss": 0.527,
+      "step": 15304
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881245481363733e-05,
+      "loss": 0.5132,
+      "step": 15305
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881220236384964e-05,
+      "loss": 0.4484,
+      "step": 15306
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881194988739207e-05,
+      "loss": 0.4032,
+      "step": 15307
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988116973842647e-05,
+      "loss": 0.4112,
+      "step": 15308
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988114448544676e-05,
+      "loss": 0.4807,
+      "step": 15309
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988111922980008e-05,
+      "loss": 0.5019,
+      "step": 15310
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881093971486444e-05,
+      "loss": 0.4921,
+      "step": 15311
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881068710505853e-05,
+      "loss": 0.5082,
+      "step": 15312
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881043446858316e-05,
+      "loss": 0.5546,
+      "step": 15313
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9881018180543833e-05,
+      "loss": 0.4672,
+      "step": 15314
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880992911562424e-05,
+      "loss": 0.4596,
+      "step": 15315
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880967639914087e-05,
+      "loss": 0.4247,
+      "step": 15316
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.988094236559883e-05,
+      "loss": 0.4455,
+      "step": 15317
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880917088616663e-05,
+      "loss": 0.3857,
+      "step": 15318
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880891808967585e-05,
+      "loss": 0.4796,
+      "step": 15319
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880866526651614e-05,
+      "loss": 0.4154,
+      "step": 15320
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880841241668746e-05,
+      "loss": 0.4789,
+      "step": 15321
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880815954018994e-05,
+      "loss": 0.5228,
+      "step": 15322
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880790663702366e-05,
+      "loss": 0.4668,
+      "step": 15323
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880765370718866e-05,
+      "loss": 0.5319,
+      "step": 15324
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880740075068495e-05,
+      "loss": 0.5044,
+      "step": 15325
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9880714776751272e-05,
+      "loss": 0.4872,
+      "step": 15326
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.98806894757672e-05,
+      "loss": 0.5132,
+      "step": 15327
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.988066417211628e-05,
+      "loss": 0.4075,
+      "step": 15328
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880638865798523e-05,
+      "loss": 0.4091,
+      "step": 15329
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880613556813936e-05,
+      "loss": 0.4774,
+      "step": 15330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880588245162524e-05,
+      "loss": 0.4406,
+      "step": 15331
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880562930844296e-05,
+      "loss": 0.4955,
+      "step": 15332
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.988053761385926e-05,
+      "loss": 0.4478,
+      "step": 15333
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880512294207418e-05,
+      "loss": 0.508,
+      "step": 15334
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.988048697188878e-05,
+      "loss": 0.4561,
+      "step": 15335
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880461646903352e-05,
+      "loss": 0.5063,
+      "step": 15336
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880436319251138e-05,
+      "loss": 0.5517,
+      "step": 15337
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880410988932156e-05,
+      "loss": 0.5047,
+      "step": 15338
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.98803856559464e-05,
+      "loss": 0.484,
+      "step": 15339
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880360320293882e-05,
+      "loss": 0.4506,
+      "step": 15340
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.988033498197461e-05,
+      "loss": 0.4753,
+      "step": 15341
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880309640988586e-05,
+      "loss": 0.4861,
+      "step": 15342
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880284297335825e-05,
+      "loss": 0.456,
+      "step": 15343
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880258951016323e-05,
+      "loss": 0.4325,
+      "step": 15344
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.98802336020301e-05,
+      "loss": 0.4991,
+      "step": 15345
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.988020825037715e-05,
+      "loss": 0.4457,
+      "step": 15346
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880182896057487e-05,
+      "loss": 0.5109,
+      "step": 15347
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.988015753907112e-05,
+      "loss": 0.5266,
+      "step": 15348
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.988013217941805e-05,
+      "loss": 0.4064,
+      "step": 15349
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880106817098284e-05,
+      "loss": 0.4482,
+      "step": 15350
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880081452111836e-05,
+      "loss": 0.4749,
+      "step": 15351
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.98800560844587e-05,
+      "loss": 0.4965,
+      "step": 15352
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880030714138897e-05,
+      "loss": 0.496,
+      "step": 15353
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9880005341152426e-05,
+      "loss": 0.4123,
+      "step": 15354
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879979965499296e-05,
+      "loss": 0.4922,
+      "step": 15355
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987995458717951e-05,
+      "loss": 0.526,
+      "step": 15356
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879929206193083e-05,
+      "loss": 0.4901,
+      "step": 15357
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879903822540015e-05,
+      "loss": 0.6314,
+      "step": 15358
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879878436220313e-05,
+      "loss": 0.4062,
+      "step": 15359
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879853047233987e-05,
+      "loss": 0.417,
+      "step": 15360
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879827655581044e-05,
+      "loss": 0.4429,
+      "step": 15361
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879802261261486e-05,
+      "loss": 0.4492,
+      "step": 15362
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879776864275323e-05,
+      "loss": 0.5276,
+      "step": 15363
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879751464622564e-05,
+      "loss": 0.5026,
+      "step": 15364
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879726062303214e-05,
+      "loss": 0.5033,
+      "step": 15365
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987970065731728e-05,
+      "loss": 0.4675,
+      "step": 15366
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987967524966477e-05,
+      "loss": 0.4576,
+      "step": 15367
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879649839345685e-05,
+      "loss": 0.5137,
+      "step": 15368
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987962442636004e-05,
+      "loss": 0.4839,
+      "step": 15369
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879599010707837e-05,
+      "loss": 0.4667,
+      "step": 15370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879573592389085e-05,
+      "loss": 0.4435,
+      "step": 15371
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879548171403788e-05,
+      "loss": 0.5317,
+      "step": 15372
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879522747751956e-05,
+      "loss": 0.5001,
+      "step": 15373
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879497321433594e-05,
+      "loss": 0.4629,
+      "step": 15374
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987947189244871e-05,
+      "loss": 0.5311,
+      "step": 15375
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879446460797308e-05,
+      "loss": 0.4987,
+      "step": 15376
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.98794210264794e-05,
+      "loss": 0.4174,
+      "step": 15377
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879395589494993e-05,
+      "loss": 0.4602,
+      "step": 15378
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879370149844085e-05,
+      "loss": 0.4821,
+      "step": 15379
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879344707526694e-05,
+      "loss": 0.4569,
+      "step": 15380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987931926254282e-05,
+      "loss": 0.4536,
+      "step": 15381
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879293814892467e-05,
+      "loss": 0.5593,
+      "step": 15382
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987926836457565e-05,
+      "loss": 0.4557,
+      "step": 15383
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879242911592375e-05,
+      "loss": 0.581,
+      "step": 15384
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879217455942644e-05,
+      "loss": 0.4374,
+      "step": 15385
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879191997626464e-05,
+      "loss": 0.4528,
+      "step": 15386
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879166536643844e-05,
+      "loss": 0.4496,
+      "step": 15387
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879141072994792e-05,
+      "loss": 0.5276,
+      "step": 15388
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879115606679314e-05,
+      "loss": 0.4865,
+      "step": 15389
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879090137697417e-05,
+      "loss": 0.4133,
+      "step": 15390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879064666049107e-05,
+      "loss": 0.5155,
+      "step": 15391
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879039191734392e-05,
+      "loss": 0.474,
+      "step": 15392
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9879013714753275e-05,
+      "loss": 0.4367,
+      "step": 15393
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987898823510577e-05,
+      "loss": 0.4651,
+      "step": 15394
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987896275279188e-05,
+      "loss": 0.5253,
+      "step": 15395
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878937267811606e-05,
+      "loss": 0.4851,
+      "step": 15396
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878911780164965e-05,
+      "loss": 0.458,
+      "step": 15397
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987888628985196e-05,
+      "loss": 0.4788,
+      "step": 15398
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878860796872596e-05,
+      "loss": 0.4031,
+      "step": 15399
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987883530122688e-05,
+      "loss": 0.3724,
+      "step": 15400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878809802914824e-05,
+      "loss": 0.5249,
+      "step": 15401
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987878430193643e-05,
+      "loss": 0.574,
+      "step": 15402
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878758798291705e-05,
+      "loss": 0.5592,
+      "step": 15403
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878733291980653e-05,
+      "loss": 0.4205,
+      "step": 15404
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987870778300329e-05,
+      "loss": 0.4694,
+      "step": 15405
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878682271359616e-05,
+      "loss": 0.4417,
+      "step": 15406
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987865675704964e-05,
+      "loss": 0.4243,
+      "step": 15407
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878631240073367e-05,
+      "loss": 0.4311,
+      "step": 15408
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878605720430805e-05,
+      "loss": 0.4216,
+      "step": 15409
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878580198121964e-05,
+      "loss": 0.435,
+      "step": 15410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878554673146845e-05,
+      "loss": 0.4928,
+      "step": 15411
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987852914550546e-05,
+      "loss": 0.4984,
+      "step": 15412
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987850361519781e-05,
+      "loss": 0.4815,
+      "step": 15413
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878478082223913e-05,
+      "loss": 0.5085,
+      "step": 15414
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878452546583766e-05,
+      "loss": 0.4174,
+      "step": 15415
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878427008277375e-05,
+      "loss": 0.4821,
+      "step": 15416
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878401467304752e-05,
+      "loss": 0.4728,
+      "step": 15417
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.98783759236659e-05,
+      "loss": 0.4914,
+      "step": 15418
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878350377360834e-05,
+      "loss": 0.4608,
+      "step": 15419
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878324828389547e-05,
+      "loss": 0.4873,
+      "step": 15420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878299276752064e-05,
+      "loss": 0.3999,
+      "step": 15421
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878273722448372e-05,
+      "loss": 0.5167,
+      "step": 15422
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878248165478495e-05,
+      "loss": 0.4244,
+      "step": 15423
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987822260584243e-05,
+      "loss": 0.4459,
+      "step": 15424
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9878197043540184e-05,
+      "loss": 0.4933,
+      "step": 15425
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.987817147857177e-05,
+      "loss": 0.4454,
+      "step": 15426
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9878145910937187e-05,
+      "loss": 0.4818,
+      "step": 15427
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9878120340636455e-05,
+      "loss": 0.4664,
+      "step": 15428
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9878094767669564e-05,
+      "loss": 0.5087,
+      "step": 15429
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9878069192036532e-05,
+      "loss": 0.4449,
+      "step": 15430
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9878043613737362e-05,
+      "loss": 0.4431,
+      "step": 15431
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987801803277206e-05,
+      "loss": 0.4523,
+      "step": 15432
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877992449140636e-05,
+      "loss": 0.5071,
+      "step": 15433
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877966862843097e-05,
+      "loss": 0.4633,
+      "step": 15434
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987794127387945e-05,
+      "loss": 0.5024,
+      "step": 15435
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.98779156822497e-05,
+      "loss": 0.4644,
+      "step": 15436
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877890087953852e-05,
+      "loss": 0.4327,
+      "step": 15437
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877864490991918e-05,
+      "loss": 0.4966,
+      "step": 15438
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.98778388913639e-05,
+      "loss": 0.5027,
+      "step": 15439
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877813289069807e-05,
+      "loss": 0.5018,
+      "step": 15440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987778768410965e-05,
+      "loss": 0.5627,
+      "step": 15441
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877762076483427e-05,
+      "loss": 0.4863,
+      "step": 15442
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877736466191152e-05,
+      "loss": 0.5097,
+      "step": 15443
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877710853232827e-05,
+      "loss": 0.5401,
+      "step": 15444
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877685237608463e-05,
+      "loss": 0.4864,
+      "step": 15445
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987765961931807e-05,
+      "loss": 0.4517,
+      "step": 15446
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877633998361646e-05,
+      "loss": 0.4382,
+      "step": 15447
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877608374739203e-05,
+      "loss": 0.5788,
+      "step": 15448
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987758274845075e-05,
+      "loss": 0.4743,
+      "step": 15449
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877557119496288e-05,
+      "loss": 0.5098,
+      "step": 15450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877531487875828e-05,
+      "loss": 0.4231,
+      "step": 15451
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877505853589376e-05,
+      "loss": 0.4378,
+      "step": 15452
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987748021663694e-05,
+      "loss": 0.4589,
+      "step": 15453
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877454577018523e-05,
+      "loss": 0.5201,
+      "step": 15454
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987742893473414e-05,
+      "loss": 0.5154,
+      "step": 15455
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877403289783787e-05,
+      "loss": 0.4457,
+      "step": 15456
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987737764216748e-05,
+      "loss": 0.4675,
+      "step": 15457
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877351991885224e-05,
+      "loss": 0.4997,
+      "step": 15458
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877326338937023e-05,
+      "loss": 0.4827,
+      "step": 15459
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877300683322885e-05,
+      "loss": 0.4082,
+      "step": 15460
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877275025042816e-05,
+      "loss": 0.487,
+      "step": 15461
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877249364096825e-05,
+      "loss": 0.4176,
+      "step": 15462
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987722370048492e-05,
+      "loss": 0.5035,
+      "step": 15463
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877198034207105e-05,
+      "loss": 0.4325,
+      "step": 15464
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877172365263385e-05,
+      "loss": 0.5029,
+      "step": 15465
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877146693653775e-05,
+      "loss": 0.4918,
+      "step": 15466
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877121019378275e-05,
+      "loss": 0.5032,
+      "step": 15467
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877095342436897e-05,
+      "loss": 0.4016,
+      "step": 15468
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987706966282964e-05,
+      "loss": 0.4597,
+      "step": 15469
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877043980556518e-05,
+      "loss": 0.4854,
+      "step": 15470
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9877018295617537e-05,
+      "loss": 0.5283,
+      "step": 15471
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.98769926080127e-05,
+      "loss": 0.4619,
+      "step": 15472
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876966917742014e-05,
+      "loss": 0.4546,
+      "step": 15473
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876941224805493e-05,
+      "loss": 0.5388,
+      "step": 15474
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987691552920314e-05,
+      "loss": 0.4227,
+      "step": 15475
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987688983093496e-05,
+      "loss": 0.4748,
+      "step": 15476
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876864130000957e-05,
+      "loss": 0.4615,
+      "step": 15477
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987683842640115e-05,
+      "loss": 0.4892,
+      "step": 15478
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987681272013553e-05,
+      "loss": 0.5388,
+      "step": 15479
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987678701120412e-05,
+      "loss": 0.5001,
+      "step": 15480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876761299606913e-05,
+      "loss": 0.5137,
+      "step": 15481
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876735585343925e-05,
+      "loss": 0.4452,
+      "step": 15482
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987670986841516e-05,
+      "loss": 0.4111,
+      "step": 15483
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876684148820624e-05,
+      "loss": 0.4367,
+      "step": 15484
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876658426560325e-05,
+      "loss": 0.5205,
+      "step": 15485
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987663270163427e-05,
+      "loss": 0.5066,
+      "step": 15486
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876606974042463e-05,
+      "loss": 0.4092,
+      "step": 15487
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876581243784915e-05,
+      "loss": 0.5064,
+      "step": 15488
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876555510861633e-05,
+      "loss": 0.4948,
+      "step": 15489
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876529775272622e-05,
+      "loss": 0.5023,
+      "step": 15490
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987650403701789e-05,
+      "loss": 0.5014,
+      "step": 15491
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987647829609744e-05,
+      "loss": 0.4735,
+      "step": 15492
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876452552511284e-05,
+      "loss": 0.5505,
+      "step": 15493
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987642680625943e-05,
+      "loss": 0.4454,
+      "step": 15494
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876401057341882e-05,
+      "loss": 0.5284,
+      "step": 15495
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876375305758643e-05,
+      "loss": 0.4332,
+      "step": 15496
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876349551509727e-05,
+      "loss": 0.4121,
+      "step": 15497
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876323794595138e-05,
+      "loss": 0.6331,
+      "step": 15498
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987629803501488e-05,
+      "loss": 0.387,
+      "step": 15499
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876272272768968e-05,
+      "loss": 0.6179,
+      "step": 15500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876246507857405e-05,
+      "loss": 0.4146,
+      "step": 15501
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987622074028019e-05,
+      "loss": 0.4359,
+      "step": 15502
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876194970037342e-05,
+      "loss": 0.4397,
+      "step": 15503
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876169197128863e-05,
+      "loss": 0.5453,
+      "step": 15504
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876143421554757e-05,
+      "loss": 0.5296,
+      "step": 15505
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876117643315033e-05,
+      "loss": 0.5532,
+      "step": 15506
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876091862409702e-05,
+      "loss": 0.5017,
+      "step": 15507
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987606607883877e-05,
+      "loss": 0.5284,
+      "step": 15508
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876040292602233e-05,
+      "loss": 0.4639,
+      "step": 15509
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9876014503700115e-05,
+      "loss": 0.4417,
+      "step": 15510
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987598871213241e-05,
+      "loss": 0.4647,
+      "step": 15511
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987596291789913e-05,
+      "loss": 0.4753,
+      "step": 15512
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987593712100028e-05,
+      "loss": 0.5683,
+      "step": 15513
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9875911321435872e-05,
+      "loss": 0.4698,
+      "step": 15514
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9875885519205907e-05,
+      "loss": 0.3869,
+      "step": 15515
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9875859714310394e-05,
+      "loss": 0.4229,
+      "step": 15516
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9875833906749343e-05,
+      "loss": 0.5397,
+      "step": 15517
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9875808096522754e-05,
+      "loss": 0.5143,
+      "step": 15518
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9875782283630644e-05,
+      "loss": 0.5067,
+      "step": 15519
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987575646807301e-05,
+      "loss": 0.5447,
+      "step": 15520
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9875730649849863e-05,
+      "loss": 0.428,
+      "step": 15521
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9875704828961212e-05,
+      "loss": 0.4884,
+      "step": 15522
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.987567900540706e-05,
+      "loss": 0.4368,
+      "step": 15523
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9875653179187416e-05,
+      "loss": 0.4605,
+      "step": 15524
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987562735030229e-05,
+      "loss": 0.4143,
+      "step": 15525
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875601518751684e-05,
+      "loss": 0.4927,
+      "step": 15526
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875575684535607e-05,
+      "loss": 0.4774,
+      "step": 15527
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875549847654067e-05,
+      "loss": 0.4971,
+      "step": 15528
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987552400810707e-05,
+      "loss": 0.4704,
+      "step": 15529
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875498165894622e-05,
+      "loss": 0.3945,
+      "step": 15530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875472321016732e-05,
+      "loss": 0.4641,
+      "step": 15531
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875446473473405e-05,
+      "loss": 0.5201,
+      "step": 15532
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875420623264648e-05,
+      "loss": 0.4513,
+      "step": 15533
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875394770390472e-05,
+      "loss": 0.4746,
+      "step": 15534
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875368914850876e-05,
+      "loss": 0.5239,
+      "step": 15535
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875343056645874e-05,
+      "loss": 0.4202,
+      "step": 15536
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875317195775473e-05,
+      "loss": 0.4779,
+      "step": 15537
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875291332239676e-05,
+      "loss": 0.5121,
+      "step": 15538
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987526546603849e-05,
+      "loss": 0.4593,
+      "step": 15539
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875239597171925e-05,
+      "loss": 0.4624,
+      "step": 15540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875213725639985e-05,
+      "loss": 0.4885,
+      "step": 15541
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875187851442682e-05,
+      "loss": 0.4301,
+      "step": 15542
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875161974580018e-05,
+      "loss": 0.4463,
+      "step": 15543
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875136095052002e-05,
+      "loss": 0.5639,
+      "step": 15544
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987511021285864e-05,
+      "loss": 0.5508,
+      "step": 15545
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875084327999938e-05,
+      "loss": 0.4713,
+      "step": 15546
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875058440475907e-05,
+      "loss": 0.5173,
+      "step": 15547
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987503255028655e-05,
+      "loss": 0.4502,
+      "step": 15548
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9875006657431878e-05,
+      "loss": 0.5011,
+      "step": 15549
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874980761911896e-05,
+      "loss": 0.4257,
+      "step": 15550
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874954863726604e-05,
+      "loss": 0.4781,
+      "step": 15551
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874928962876024e-05,
+      "loss": 0.4978,
+      "step": 15552
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987490305936015e-05,
+      "loss": 0.4314,
+      "step": 15553
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874877153178996e-05,
+      "loss": 0.4691,
+      "step": 15554
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874851244332564e-05,
+      "loss": 0.5435,
+      "step": 15555
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987482533282086e-05,
+      "loss": 0.5547,
+      "step": 15556
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874799418643902e-05,
+      "loss": 0.5538,
+      "step": 15557
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874773501801686e-05,
+      "loss": 0.5378,
+      "step": 15558
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874747582294222e-05,
+      "loss": 0.461,
+      "step": 15559
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874721660121516e-05,
+      "loss": 0.4154,
+      "step": 15560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987469573528358e-05,
+      "loss": 0.4519,
+      "step": 15561
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874669807780417e-05,
+      "loss": 0.4854,
+      "step": 15562
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874643877612035e-05,
+      "loss": 0.4615,
+      "step": 15563
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987461794477844e-05,
+      "loss": 0.5159,
+      "step": 15564
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874592009279635e-05,
+      "loss": 0.4852,
+      "step": 15565
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874566071115634e-05,
+      "loss": 0.5473,
+      "step": 15566
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987454013028644e-05,
+      "loss": 0.539,
+      "step": 15567
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874514186792065e-05,
+      "loss": 0.4551,
+      "step": 15568
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874488240632514e-05,
+      "loss": 0.4399,
+      "step": 15569
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874462291807787e-05,
+      "loss": 0.4618,
+      "step": 15570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.98744363403179e-05,
+      "loss": 0.4939,
+      "step": 15571
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874410386162854e-05,
+      "loss": 0.5036,
+      "step": 15572
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874384429342665e-05,
+      "loss": 0.4652,
+      "step": 15573
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874358469857323e-05,
+      "loss": 0.4133,
+      "step": 15574
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874332507706854e-05,
+      "loss": 0.4982,
+      "step": 15575
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874306542891253e-05,
+      "loss": 0.5447,
+      "step": 15576
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987428057541053e-05,
+      "loss": 0.5108,
+      "step": 15577
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874254605264697e-05,
+      "loss": 0.5527,
+      "step": 15578
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874228632453752e-05,
+      "loss": 0.4932,
+      "step": 15579
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987420265697771e-05,
+      "loss": 0.451,
+      "step": 15580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874176678836566e-05,
+      "loss": 0.447,
+      "step": 15581
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874150698030345e-05,
+      "loss": 0.4359,
+      "step": 15582
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874124714559043e-05,
+      "loss": 0.4317,
+      "step": 15583
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874098728422667e-05,
+      "loss": 0.5173,
+      "step": 15584
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987407273962122e-05,
+      "loss": 0.4408,
+      "step": 15585
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987404674815472e-05,
+      "loss": 0.4188,
+      "step": 15586
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9874020754023174e-05,
+      "loss": 0.4628,
+      "step": 15587
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873994757226577e-05,
+      "loss": 0.482,
+      "step": 15588
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873968757764943e-05,
+      "loss": 0.4169,
+      "step": 15589
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987394275563828e-05,
+      "loss": 0.4456,
+      "step": 15590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873916750846593e-05,
+      "loss": 0.5385,
+      "step": 15591
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987389074338989e-05,
+      "loss": 0.4847,
+      "step": 15592
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873864733268177e-05,
+      "loss": 0.5569,
+      "step": 15593
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873838720481465e-05,
+      "loss": 0.4561,
+      "step": 15594
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873812705029753e-05,
+      "loss": 0.4867,
+      "step": 15595
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873786686913055e-05,
+      "loss": 0.4927,
+      "step": 15596
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873760666131375e-05,
+      "loss": 0.4354,
+      "step": 15597
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873734642684726e-05,
+      "loss": 0.5024,
+      "step": 15598
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873708616573105e-05,
+      "loss": 0.4562,
+      "step": 15599
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873682587796525e-05,
+      "loss": 0.5027,
+      "step": 15600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987365655635499e-05,
+      "loss": 0.4102,
+      "step": 15601
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873630522248512e-05,
+      "loss": 0.5714,
+      "step": 15602
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873604485477093e-05,
+      "loss": 0.428,
+      "step": 15603
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873578446040742e-05,
+      "loss": 0.4208,
+      "step": 15604
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873552403939466e-05,
+      "loss": 0.4716,
+      "step": 15605
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873526359173272e-05,
+      "loss": 0.4342,
+      "step": 15606
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873500311742167e-05,
+      "loss": 0.4636,
+      "step": 15607
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987347426164616e-05,
+      "loss": 0.4461,
+      "step": 15608
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873448208885253e-05,
+      "loss": 0.4879,
+      "step": 15609
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873422153459458e-05,
+      "loss": 0.4811,
+      "step": 15610
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987339609536878e-05,
+      "loss": 0.4948,
+      "step": 15611
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873370034613225e-05,
+      "loss": 0.5009,
+      "step": 15612
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873343971192805e-05,
+      "loss": 0.4314,
+      "step": 15613
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987331790510752e-05,
+      "loss": 0.4443,
+      "step": 15614
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987329183635738e-05,
+      "loss": 0.4923,
+      "step": 15615
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873265764942396e-05,
+      "loss": 0.5082,
+      "step": 15616
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987323969086257e-05,
+      "loss": 0.5467,
+      "step": 15617
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987321361411791e-05,
+      "loss": 0.4994,
+      "step": 15618
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.987318753470842e-05,
+      "loss": 0.549,
+      "step": 15619
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873161452634117e-05,
+      "loss": 0.4357,
+      "step": 15620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873135367894995e-05,
+      "loss": 0.4917,
+      "step": 15621
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873109280491073e-05,
+      "loss": 0.5445,
+      "step": 15622
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9873083190422352e-05,
+      "loss": 0.5187,
+      "step": 15623
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9873057097688835e-05,
+      "loss": 0.4492,
+      "step": 15624
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987303100229054e-05,
+      "loss": 0.456,
+      "step": 15625
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9873004904227463e-05,
+      "loss": 0.4484,
+      "step": 15626
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987297880349962e-05,
+      "loss": 0.4729,
+      "step": 15627
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872952700107008e-05,
+      "loss": 0.4621,
+      "step": 15628
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872926594049643e-05,
+      "loss": 0.5546,
+      "step": 15629
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872900485327528e-05,
+      "loss": 0.4773,
+      "step": 15630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872874373940675e-05,
+      "loss": 0.6215,
+      "step": 15631
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872848259889083e-05,
+      "loss": 0.4611,
+      "step": 15632
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872822143172767e-05,
+      "loss": 0.5001,
+      "step": 15633
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872796023791725e-05,
+      "loss": 0.4943,
+      "step": 15634
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872769901745972e-05,
+      "loss": 0.4881,
+      "step": 15635
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987274377703551e-05,
+      "loss": 0.4103,
+      "step": 15636
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872717649660352e-05,
+      "loss": 0.5567,
+      "step": 15637
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872691519620502e-05,
+      "loss": 0.4208,
+      "step": 15638
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872665386915964e-05,
+      "loss": 0.4474,
+      "step": 15639
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872639251546746e-05,
+      "loss": 0.5047,
+      "step": 15640
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872613113512856e-05,
+      "loss": 0.5586,
+      "step": 15641
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872586972814306e-05,
+      "loss": 0.5335,
+      "step": 15642
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872560829451095e-05,
+      "loss": 0.47,
+      "step": 15643
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872534683423237e-05,
+      "loss": 0.4422,
+      "step": 15644
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872508534730733e-05,
+      "loss": 0.4892,
+      "step": 15645
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872482383373594e-05,
+      "loss": 0.4824,
+      "step": 15646
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872456229351826e-05,
+      "loss": 0.4904,
+      "step": 15647
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872430072665434e-05,
+      "loss": 0.5486,
+      "step": 15648
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987240391331443e-05,
+      "loss": 0.4549,
+      "step": 15649
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872377751298814e-05,
+      "loss": 0.439,
+      "step": 15650
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.98723515866186e-05,
+      "loss": 0.5922,
+      "step": 15651
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872325419273792e-05,
+      "loss": 0.4567,
+      "step": 15652
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.98722992492644e-05,
+      "loss": 0.5118,
+      "step": 15653
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987227307659042e-05,
+      "loss": 0.4992,
+      "step": 15654
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872246901251874e-05,
+      "loss": 0.4908,
+      "step": 15655
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987222072324876e-05,
+      "loss": 0.5611,
+      "step": 15656
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872194542581086e-05,
+      "loss": 0.4604,
+      "step": 15657
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872168359248864e-05,
+      "loss": 0.5685,
+      "step": 15658
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.98721421732521e-05,
+      "loss": 0.4835,
+      "step": 15659
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872115984590793e-05,
+      "loss": 0.4375,
+      "step": 15660
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872089793264958e-05,
+      "loss": 0.5302,
+      "step": 15661
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872063599274602e-05,
+      "loss": 0.4507,
+      "step": 15662
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9872037402619724e-05,
+      "loss": 0.5696,
+      "step": 15663
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987201120330034e-05,
+      "loss": 0.5034,
+      "step": 15664
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871985001316456e-05,
+      "loss": 0.5177,
+      "step": 15665
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871958796668074e-05,
+      "loss": 0.5334,
+      "step": 15666
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871932589355207e-05,
+      "loss": 0.4139,
+      "step": 15667
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987190637937786e-05,
+      "loss": 0.5459,
+      "step": 15668
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871880166736037e-05,
+      "loss": 0.4438,
+      "step": 15669
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871853951429748e-05,
+      "loss": 0.448,
+      "step": 15670
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871827733458998e-05,
+      "loss": 0.5719,
+      "step": 15671
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871801512823797e-05,
+      "loss": 0.4651,
+      "step": 15672
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987177528952415e-05,
+      "loss": 0.4896,
+      "step": 15673
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871749063560064e-05,
+      "loss": 0.5045,
+      "step": 15674
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871722834931552e-05,
+      "loss": 0.4615,
+      "step": 15675
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987169660363861e-05,
+      "loss": 0.4201,
+      "step": 15676
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987167036968125e-05,
+      "loss": 0.4665,
+      "step": 15677
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871644133059486e-05,
+      "loss": 0.4481,
+      "step": 15678
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871617893773314e-05,
+      "loss": 0.4659,
+      "step": 15679
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987159165182275e-05,
+      "loss": 0.4848,
+      "step": 15680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871565407207795e-05,
+      "loss": 0.4508,
+      "step": 15681
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871539159928458e-05,
+      "loss": 0.5058,
+      "step": 15682
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871512909984748e-05,
+      "loss": 0.5296,
+      "step": 15683
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987148665737667e-05,
+      "loss": 0.5328,
+      "step": 15684
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871460402104228e-05,
+      "loss": 0.4319,
+      "step": 15685
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871434144167437e-05,
+      "loss": 0.4349,
+      "step": 15686
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.98714078835663e-05,
+      "loss": 0.5088,
+      "step": 15687
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871381620300824e-05,
+      "loss": 0.4769,
+      "step": 15688
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871355354371014e-05,
+      "loss": 0.438,
+      "step": 15689
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987132908577688e-05,
+      "loss": 0.5206,
+      "step": 15690
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871302814518425e-05,
+      "loss": 0.4384,
+      "step": 15691
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871276540595663e-05,
+      "loss": 0.4816,
+      "step": 15692
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871250264008596e-05,
+      "loss": 0.4727,
+      "step": 15693
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987122398475723e-05,
+      "loss": 0.5531,
+      "step": 15694
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987119770284158e-05,
+      "loss": 0.4766,
+      "step": 15695
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871171418261645e-05,
+      "loss": 0.4216,
+      "step": 15696
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871145131017433e-05,
+      "loss": 0.4118,
+      "step": 15697
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871118841108954e-05,
+      "loss": 0.4977,
+      "step": 15698
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871092548536214e-05,
+      "loss": 0.4181,
+      "step": 15699
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987106625329922e-05,
+      "loss": 0.5294,
+      "step": 15700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9871039955397982e-05,
+      "loss": 0.5063,
+      "step": 15701
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.98710136548325e-05,
+      "loss": 0.4714,
+      "step": 15702
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870987351602785e-05,
+      "loss": 0.4616,
+      "step": 15703
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987096104570885e-05,
+      "loss": 0.4224,
+      "step": 15704
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987093473715069e-05,
+      "loss": 0.4096,
+      "step": 15705
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987090842592832e-05,
+      "loss": 0.3988,
+      "step": 15706
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870882112041748e-05,
+      "loss": 0.4899,
+      "step": 15707
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987085579549098e-05,
+      "loss": 0.5161,
+      "step": 15708
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870829476276017e-05,
+      "loss": 0.4824,
+      "step": 15709
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870803154396872e-05,
+      "loss": 0.4026,
+      "step": 15710
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870776829853554e-05,
+      "loss": 0.5243,
+      "step": 15711
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870750502646064e-05,
+      "loss": 0.4213,
+      "step": 15712
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870724172774414e-05,
+      "loss": 0.4122,
+      "step": 15713
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987069784023861e-05,
+      "loss": 0.5132,
+      "step": 15714
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870671505038658e-05,
+      "loss": 0.506,
+      "step": 15715
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870645167174565e-05,
+      "loss": 0.5455,
+      "step": 15716
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987061882664634e-05,
+      "loss": 0.4015,
+      "step": 15717
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.987059248345399e-05,
+      "loss": 0.4923,
+      "step": 15718
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870566137597517e-05,
+      "loss": 0.4413,
+      "step": 15719
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870539789076933e-05,
+      "loss": 0.4293,
+      "step": 15720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9870513437892243e-05,
+      "loss": 0.5477,
+      "step": 15721
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870487084043463e-05,
+      "loss": 0.5109,
+      "step": 15722
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870460727530584e-05,
+      "loss": 0.4104,
+      "step": 15723
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870434368353627e-05,
+      "loss": 0.527,
+      "step": 15724
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870408006512586e-05,
+      "loss": 0.5077,
+      "step": 15725
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870381642007484e-05,
+      "loss": 0.4798,
+      "step": 15726
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870355274838314e-05,
+      "loss": 0.4598,
+      "step": 15727
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870328905005093e-05,
+      "loss": 0.4907,
+      "step": 15728
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870302532507822e-05,
+      "loss": 0.4326,
+      "step": 15729
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870276157346514e-05,
+      "loss": 0.5135,
+      "step": 15730
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870249779521168e-05,
+      "loss": 0.475,
+      "step": 15731
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870223399031795e-05,
+      "loss": 0.4637,
+      "step": 15732
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870197015878405e-05,
+      "loss": 0.5378,
+      "step": 15733
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870170630061002e-05,
+      "loss": 0.5151,
+      "step": 15734
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870144241579595e-05,
+      "loss": 0.4409,
+      "step": 15735
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.987011785043419e-05,
+      "loss": 0.4538,
+      "step": 15736
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870091456624793e-05,
+      "loss": 0.4387,
+      "step": 15737
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.987006506015141e-05,
+      "loss": 0.4834,
+      "step": 15738
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.987003866101405e-05,
+      "loss": 0.4649,
+      "step": 15739
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870012259212723e-05,
+      "loss": 0.4057,
+      "step": 15740
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869985854747432e-05,
+      "loss": 0.5851,
+      "step": 15741
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986995944761819e-05,
+      "loss": 0.5419,
+      "step": 15742
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869933037824996e-05,
+      "loss": 0.4782,
+      "step": 15743
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986990662536786e-05,
+      "loss": 0.5098,
+      "step": 15744
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986988021024679e-05,
+      "loss": 0.4719,
+      "step": 15745
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.98698537924618e-05,
+      "loss": 0.4603,
+      "step": 15746
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869827372012884e-05,
+      "loss": 0.4729,
+      "step": 15747
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869800948900058e-05,
+      "loss": 0.4619,
+      "step": 15748
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869774523123323e-05,
+      "loss": 0.5814,
+      "step": 15749
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869748094682693e-05,
+      "loss": 0.4751,
+      "step": 15750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986972166357817e-05,
+      "loss": 0.5269,
+      "step": 15751
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869695229809763e-05,
+      "loss": 0.5164,
+      "step": 15752
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869668793377482e-05,
+      "loss": 0.5371,
+      "step": 15753
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869642354281327e-05,
+      "loss": 0.474,
+      "step": 15754
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869615912521314e-05,
+      "loss": 0.5268,
+      "step": 15755
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986958946809744e-05,
+      "loss": 0.5196,
+      "step": 15756
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986956302100972e-05,
+      "loss": 0.4426,
+      "step": 15757
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869536571258165e-05,
+      "loss": 0.4544,
+      "step": 15758
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986951011884277e-05,
+      "loss": 0.4452,
+      "step": 15759
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869483663763546e-05,
+      "loss": 0.4778,
+      "step": 15760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869457206020506e-05,
+      "loss": 0.5236,
+      "step": 15761
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869430745613652e-05,
+      "loss": 0.4409,
+      "step": 15762
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869404282542992e-05,
+      "loss": 0.4433,
+      "step": 15763
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869377816808532e-05,
+      "loss": 0.4573,
+      "step": 15764
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869351348410285e-05,
+      "loss": 0.4892,
+      "step": 15765
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986932487734825e-05,
+      "loss": 0.3838,
+      "step": 15766
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869298403622442e-05,
+      "loss": 0.47,
+      "step": 15767
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869271927232863e-05,
+      "loss": 0.5385,
+      "step": 15768
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986924544817952e-05,
+      "loss": 0.5018,
+      "step": 15769
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869218966462423e-05,
+      "loss": 0.5197,
+      "step": 15770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869192482081576e-05,
+      "loss": 0.5346,
+      "step": 15771
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986916599503699e-05,
+      "loss": 0.4798,
+      "step": 15772
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869139505328668e-05,
+      "loss": 0.5183,
+      "step": 15773
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986911301295662e-05,
+      "loss": 0.5235,
+      "step": 15774
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986908651792085e-05,
+      "loss": 0.4926,
+      "step": 15775
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986906002022137e-05,
+      "loss": 0.5094,
+      "step": 15776
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869033519858187e-05,
+      "loss": 0.5632,
+      "step": 15777
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9869007016831302e-05,
+      "loss": 0.4907,
+      "step": 15778
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986898051114073e-05,
+      "loss": 0.4511,
+      "step": 15779
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868954002786467e-05,
+      "loss": 0.5288,
+      "step": 15780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986892749176853e-05,
+      "loss": 0.4013,
+      "step": 15781
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868900978086927e-05,
+      "loss": 0.4157,
+      "step": 15782
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868874461741655e-05,
+      "loss": 0.5082,
+      "step": 15783
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868847942732733e-05,
+      "loss": 0.3889,
+      "step": 15784
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986882142106016e-05,
+      "loss": 0.3946,
+      "step": 15785
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868794896723948e-05,
+      "loss": 0.5239,
+      "step": 15786
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868768369724102e-05,
+      "loss": 0.51,
+      "step": 15787
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868741840060627e-05,
+      "loss": 0.4693,
+      "step": 15788
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868715307733537e-05,
+      "loss": 0.4855,
+      "step": 15789
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986868877274283e-05,
+      "loss": 0.4012,
+      "step": 15790
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986866223508852e-05,
+      "loss": 0.5081,
+      "step": 15791
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986863569477061e-05,
+      "loss": 0.4682,
+      "step": 15792
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986860915178911e-05,
+      "loss": 0.4597,
+      "step": 15793
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986858260614403e-05,
+      "loss": 0.4697,
+      "step": 15794
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868556057835367e-05,
+      "loss": 0.41,
+      "step": 15795
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986852950686314e-05,
+      "loss": 0.4828,
+      "step": 15796
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986850295322735e-05,
+      "loss": 0.5487,
+      "step": 15797
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868476396928e-05,
+      "loss": 0.6328,
+      "step": 15798
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868449837965106e-05,
+      "loss": 0.4079,
+      "step": 15799
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868423276338672e-05,
+      "loss": 0.4476,
+      "step": 15800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868396712048703e-05,
+      "loss": 0.4682,
+      "step": 15801
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868370145095208e-05,
+      "loss": 0.4652,
+      "step": 15802
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986834357547819e-05,
+      "loss": 0.4571,
+      "step": 15803
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868317003197665e-05,
+      "loss": 0.4004,
+      "step": 15804
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868290428253634e-05,
+      "loss": 0.5034,
+      "step": 15805
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868263850646105e-05,
+      "loss": 0.4843,
+      "step": 15806
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868237270375084e-05,
+      "loss": 0.477,
+      "step": 15807
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986821068744058e-05,
+      "loss": 0.4882,
+      "step": 15808
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868184101842603e-05,
+      "loss": 0.5147,
+      "step": 15809
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868157513581155e-05,
+      "loss": 0.5051,
+      "step": 15810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868130922656244e-05,
+      "loss": 0.428,
+      "step": 15811
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986810432906788e-05,
+      "loss": 0.4805,
+      "step": 15812
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868077732816068e-05,
+      "loss": 0.5196,
+      "step": 15813
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868051133900815e-05,
+      "loss": 0.4958,
+      "step": 15814
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868024532322127e-05,
+      "loss": 0.481,
+      "step": 15815
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9867997928080016e-05,
+      "loss": 0.499,
+      "step": 15816
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9867971321174488e-05,
+      "loss": 0.4757,
+      "step": 15817
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9867944711605546e-05,
+      "loss": 0.5025,
+      "step": 15818
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9867918099373196e-05,
+      "loss": 0.3888,
+      "step": 15819
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9867891484477454e-05,
+      "loss": 0.4253,
+      "step": 15820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986786486691832e-05,
+      "loss": 0.4966,
+      "step": 15821
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867838246695802e-05,
+      "loss": 0.4492,
+      "step": 15822
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867811623809913e-05,
+      "loss": 0.4203,
+      "step": 15823
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867784998260654e-05,
+      "loss": 0.4406,
+      "step": 15824
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867758370048032e-05,
+      "loss": 0.4316,
+      "step": 15825
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867731739172054e-05,
+      "loss": 0.6358,
+      "step": 15826
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986770510563273e-05,
+      "loss": 0.502,
+      "step": 15827
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986767846943007e-05,
+      "loss": 0.484,
+      "step": 15828
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986765183056407e-05,
+      "loss": 0.5862,
+      "step": 15829
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867625189034753e-05,
+      "loss": 0.5033,
+      "step": 15830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867598544842112e-05,
+      "loss": 0.5095,
+      "step": 15831
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867571897986163e-05,
+      "loss": 0.4672,
+      "step": 15832
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867545248466912e-05,
+      "loss": 0.508,
+      "step": 15833
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986751859628436e-05,
+      "loss": 0.5364,
+      "step": 15834
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867491941438518e-05,
+      "loss": 0.4552,
+      "step": 15835
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867465283929398e-05,
+      "loss": 0.5951,
+      "step": 15836
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867438623757e-05,
+      "loss": 0.5109,
+      "step": 15837
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867411960921337e-05,
+      "loss": 0.4173,
+      "step": 15838
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867385295422413e-05,
+      "loss": 0.6012,
+      "step": 15839
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867358627260234e-05,
+      "loss": 0.4975,
+      "step": 15840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867331956434808e-05,
+      "loss": 0.4852,
+      "step": 15841
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867305282946144e-05,
+      "loss": 0.4664,
+      "step": 15842
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986727860679425e-05,
+      "loss": 0.5669,
+      "step": 15843
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867251927979127e-05,
+      "loss": 0.5388,
+      "step": 15844
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986722524650079e-05,
+      "loss": 0.443,
+      "step": 15845
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867198562359244e-05,
+      "loss": 0.4096,
+      "step": 15846
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986717187555449e-05,
+      "loss": 0.4342,
+      "step": 15847
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867145186086547e-05,
+      "loss": 0.6392,
+      "step": 15848
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986711849395541e-05,
+      "loss": 0.4576,
+      "step": 15849
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867091799161095e-05,
+      "loss": 0.4976,
+      "step": 15850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.98670651017036e-05,
+      "loss": 0.5273,
+      "step": 15851
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867038401582944e-05,
+      "loss": 0.6019,
+      "step": 15852
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9867011698799127e-05,
+      "loss": 0.4349,
+      "step": 15853
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866984993352157e-05,
+      "loss": 0.5194,
+      "step": 15854
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986695828524204e-05,
+      "loss": 0.5359,
+      "step": 15855
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866931574468786e-05,
+      "loss": 0.4101,
+      "step": 15856
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.98669048610324e-05,
+      "loss": 0.4904,
+      "step": 15857
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986687814493289e-05,
+      "loss": 0.4799,
+      "step": 15858
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986685142617027e-05,
+      "loss": 0.4637,
+      "step": 15859
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866824704744534e-05,
+      "loss": 0.4192,
+      "step": 15860
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866797980655697e-05,
+      "loss": 0.4491,
+      "step": 15861
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986677125390377e-05,
+      "loss": 0.5723,
+      "step": 15862
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866744524488748e-05,
+      "loss": 0.4418,
+      "step": 15863
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866717792410646e-05,
+      "loss": 0.4619,
+      "step": 15864
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866691057669475e-05,
+      "loss": 0.4368,
+      "step": 15865
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866664320265237e-05,
+      "loss": 0.4798,
+      "step": 15866
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986663758019794e-05,
+      "loss": 0.4315,
+      "step": 15867
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986661083746759e-05,
+      "loss": 0.4505,
+      "step": 15868
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866584092074196e-05,
+      "loss": 0.4291,
+      "step": 15869
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866557344017762e-05,
+      "loss": 0.514,
+      "step": 15870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866530593298304e-05,
+      "loss": 0.5178,
+      "step": 15871
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986650383991582e-05,
+      "loss": 0.5143,
+      "step": 15872
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986647708387032e-05,
+      "loss": 0.5189,
+      "step": 15873
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866450325161813e-05,
+      "loss": 0.4466,
+      "step": 15874
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866423563790303e-05,
+      "loss": 0.531,
+      "step": 15875
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866396799755802e-05,
+      "loss": 0.4669,
+      "step": 15876
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986637003305831e-05,
+      "loss": 0.4564,
+      "step": 15877
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866343263697844e-05,
+      "loss": 0.4168,
+      "step": 15878
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866316491674403e-05,
+      "loss": 0.4787,
+      "step": 15879
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866289716988e-05,
+      "loss": 0.4918,
+      "step": 15880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866262939638635e-05,
+      "loss": 0.4167,
+      "step": 15881
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866236159626325e-05,
+      "loss": 0.4526,
+      "step": 15882
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866209376951065e-05,
+      "loss": 0.4012,
+      "step": 15883
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866182591612873e-05,
+      "loss": 0.5191,
+      "step": 15884
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866155803611755e-05,
+      "loss": 0.551,
+      "step": 15885
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866129012947707e-05,
+      "loss": 0.4489,
+      "step": 15886
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986610221962075e-05,
+      "loss": 0.4837,
+      "step": 15887
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866075423630885e-05,
+      "loss": 0.4546,
+      "step": 15888
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866048624978124e-05,
+      "loss": 0.4099,
+      "step": 15889
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9866021823662465e-05,
+      "loss": 0.513,
+      "step": 15890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865995019683923e-05,
+      "loss": 0.4493,
+      "step": 15891
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.98659682130425e-05,
+      "loss": 0.4443,
+      "step": 15892
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865941403738208e-05,
+      "loss": 0.4634,
+      "step": 15893
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865914591771055e-05,
+      "loss": 0.461,
+      "step": 15894
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865887777141044e-05,
+      "loss": 0.5324,
+      "step": 15895
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986586095984818e-05,
+      "loss": 0.5109,
+      "step": 15896
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986583413989248e-05,
+      "loss": 0.4672,
+      "step": 15897
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986580731727394e-05,
+      "loss": 0.5567,
+      "step": 15898
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865780491992576e-05,
+      "loss": 0.4383,
+      "step": 15899
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986575366404839e-05,
+      "loss": 0.53,
+      "step": 15900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986572683344139e-05,
+      "loss": 0.416,
+      "step": 15901
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865700000171587e-05,
+      "loss": 0.4511,
+      "step": 15902
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865673164238982e-05,
+      "loss": 0.6239,
+      "step": 15903
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986564632564359e-05,
+      "loss": 0.4391,
+      "step": 15904
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986561948438541e-05,
+      "loss": 0.5226,
+      "step": 15905
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865592640464454e-05,
+      "loss": 0.452,
+      "step": 15906
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986556579388073e-05,
+      "loss": 0.4492,
+      "step": 15907
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865538944634242e-05,
+      "loss": 0.5026,
+      "step": 15908
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865512092725003e-05,
+      "loss": 0.4561,
+      "step": 15909
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986548523815301e-05,
+      "loss": 0.4414,
+      "step": 15910
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865458380918282e-05,
+      "loss": 0.4815,
+      "step": 15911
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865431521020817e-05,
+      "loss": 0.5148,
+      "step": 15912
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865404658460627e-05,
+      "loss": 0.5176,
+      "step": 15913
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865377793237717e-05,
+      "loss": 0.4439,
+      "step": 15914
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865350925352098e-05,
+      "loss": 0.5023,
+      "step": 15915
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865324054803774e-05,
+      "loss": 0.4764,
+      "step": 15916
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986529718159275e-05,
+      "loss": 0.4199,
+      "step": 15917
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.986527030571904e-05,
+      "loss": 0.509,
+      "step": 15918
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9865243427182644e-05,
+      "loss": 0.5094,
+      "step": 15919
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9865216545983578e-05,
+      "loss": 0.4119,
+      "step": 15920
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9865189662121837e-05,
+      "loss": 0.4455,
+      "step": 15921
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986516277559744e-05,
+      "loss": 0.5449,
+      "step": 15922
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9865135886410387e-05,
+      "loss": 0.5225,
+      "step": 15923
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986510899456069e-05,
+      "loss": 0.4928,
+      "step": 15924
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986508210004835e-05,
+      "loss": 0.4558,
+      "step": 15925
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9865055202873384e-05,
+      "loss": 0.5383,
+      "step": 15926
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986502830303579e-05,
+      "loss": 0.5134,
+      "step": 15927
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986500140053558e-05,
+      "loss": 0.4894,
+      "step": 15928
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986497449537276e-05,
+      "loss": 0.4672,
+      "step": 15929
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864947587547333e-05,
+      "loss": 0.3859,
+      "step": 15930
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864920677059315e-05,
+      "loss": 0.5181,
+      "step": 15931
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986489376390871e-05,
+      "loss": 0.4898,
+      "step": 15932
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986486684809552e-05,
+      "loss": 0.4399,
+      "step": 15933
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986483992961976e-05,
+      "loss": 0.4028,
+      "step": 15934
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864813008481428e-05,
+      "loss": 0.4485,
+      "step": 15935
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864786084680542e-05,
+      "loss": 0.5127,
+      "step": 15936
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864759158217102e-05,
+      "loss": 0.5293,
+      "step": 15937
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986473222909112e-05,
+      "loss": 0.5884,
+      "step": 15938
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.98647052973026e-05,
+      "loss": 0.5523,
+      "step": 15939
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864678362851547e-05,
+      "loss": 0.5193,
+      "step": 15940
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864651425737972e-05,
+      "loss": 0.5013,
+      "step": 15941
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864624485961882e-05,
+      "loss": 0.5011,
+      "step": 15942
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864597543523286e-05,
+      "loss": 0.4545,
+      "step": 15943
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864570598422185e-05,
+      "loss": 0.4242,
+      "step": 15944
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864543650658598e-05,
+      "loss": 0.4021,
+      "step": 15945
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864516700232515e-05,
+      "loss": 0.454,
+      "step": 15946
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864489747143957e-05,
+      "loss": 0.4865,
+      "step": 15947
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864462791392927e-05,
+      "loss": 0.4947,
+      "step": 15948
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864435832979436e-05,
+      "loss": 0.3873,
+      "step": 15949
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986440887190348e-05,
+      "loss": 0.5172,
+      "step": 15950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864381908165077e-05,
+      "loss": 0.5011,
+      "step": 15951
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864354941764234e-05,
+      "loss": 0.4762,
+      "step": 15952
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864327972700954e-05,
+      "loss": 0.5437,
+      "step": 15953
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864301000975245e-05,
+      "loss": 0.5417,
+      "step": 15954
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864274026587115e-05,
+      "loss": 0.5129,
+      "step": 15955
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864247049536574e-05,
+      "loss": 0.4977,
+      "step": 15956
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986422006982362e-05,
+      "loss": 0.4552,
+      "step": 15957
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864193087448274e-05,
+      "loss": 0.5301,
+      "step": 15958
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864166102410534e-05,
+      "loss": 0.4927,
+      "step": 15959
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864139114710408e-05,
+      "loss": 0.5068,
+      "step": 15960
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864112124347907e-05,
+      "loss": 0.5273,
+      "step": 15961
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864085131323033e-05,
+      "loss": 0.4218,
+      "step": 15962
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864058135635796e-05,
+      "loss": 0.4538,
+      "step": 15963
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986403113728621e-05,
+      "loss": 0.4965,
+      "step": 15964
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9864004136274268e-05,
+      "loss": 0.4231,
+      "step": 15965
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986397713259999e-05,
+      "loss": 0.4499,
+      "step": 15966
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863950126263378e-05,
+      "loss": 0.4866,
+      "step": 15967
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863923117264438e-05,
+      "loss": 0.4276,
+      "step": 15968
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986389610560318e-05,
+      "loss": 0.4255,
+      "step": 15969
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986386909127961e-05,
+      "loss": 0.4223,
+      "step": 15970
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863842074293737e-05,
+      "loss": 0.5441,
+      "step": 15971
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863815054645567e-05,
+      "loss": 0.451,
+      "step": 15972
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.98637880323351e-05,
+      "loss": 0.4667,
+      "step": 15973
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986376100736236e-05,
+      "loss": 0.5616,
+      "step": 15974
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986373397972734e-05,
+      "loss": 0.5293,
+      "step": 15975
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863706949430056e-05,
+      "loss": 0.4649,
+      "step": 15976
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986367991647051e-05,
+      "loss": 0.5179,
+      "step": 15977
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863652880848708e-05,
+      "loss": 0.4726,
+      "step": 15978
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986362584256466e-05,
+      "loss": 0.5438,
+      "step": 15979
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863598801618376e-05,
+      "loss": 0.455,
+      "step": 15980
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863571758009857e-05,
+      "loss": 0.5196,
+      "step": 15981
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863544711739116e-05,
+      "loss": 0.5326,
+      "step": 15982
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986351766280616e-05,
+      "loss": 0.4849,
+      "step": 15983
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863490611210992e-05,
+      "loss": 0.4609,
+      "step": 15984
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863463556953625e-05,
+      "loss": 0.4838,
+      "step": 15985
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863436500034058e-05,
+      "loss": 0.5643,
+      "step": 15986
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986340944045231e-05,
+      "loss": 0.5315,
+      "step": 15987
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863382378208374e-05,
+      "loss": 0.4187,
+      "step": 15988
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986335531330227e-05,
+      "loss": 0.4196,
+      "step": 15989
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863328245734e-05,
+      "loss": 0.5322,
+      "step": 15990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986330117550357e-05,
+      "loss": 0.4158,
+      "step": 15991
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863274102610986e-05,
+      "loss": 0.455,
+      "step": 15992
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863247027056264e-05,
+      "loss": 0.4738,
+      "step": 15993
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863219948839404e-05,
+      "loss": 0.4984,
+      "step": 15994
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863192867960412e-05,
+      "loss": 0.5117,
+      "step": 15995
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863165784419304e-05,
+      "loss": 0.5656,
+      "step": 15996
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863138698216076e-05,
+      "loss": 0.4846,
+      "step": 15997
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986311160935074e-05,
+      "loss": 0.4778,
+      "step": 15998
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986308451782331e-05,
+      "loss": 0.412,
+      "step": 15999
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863057423633783e-05,
+      "loss": 0.5229,
+      "step": 16000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9863030326782172e-05,
+      "loss": 0.4867,
+      "step": 16001
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986300322726848e-05,
+      "loss": 0.4801,
+      "step": 16002
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9862976125092723e-05,
+      "loss": 0.5436,
+      "step": 16003
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.98629490202549e-05,
+      "loss": 0.4435,
+      "step": 16004
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986292191275502e-05,
+      "loss": 0.4744,
+      "step": 16005
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9862894802593092e-05,
+      "loss": 0.5308,
+      "step": 16006
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986286768976912e-05,
+      "loss": 0.5122,
+      "step": 16007
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986284057428312e-05,
+      "loss": 0.398,
+      "step": 16008
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9862813456135087e-05,
+      "loss": 0.5173,
+      "step": 16009
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9862786335325037e-05,
+      "loss": 0.4963,
+      "step": 16010
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9862759211852976e-05,
+      "loss": 0.5286,
+      "step": 16011
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9862732085718907e-05,
+      "loss": 0.5226,
+      "step": 16012
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9862704956922844e-05,
+      "loss": 0.475,
+      "step": 16013
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.986267782546479e-05,
+      "loss": 0.4585,
+      "step": 16014
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9862650691344753e-05,
+      "loss": 0.4781,
+      "step": 16015
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9862623554562738e-05,
+      "loss": 0.5408,
+      "step": 16016
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9862596415118756e-05,
+      "loss": 0.4548,
+      "step": 16017
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862569273012817e-05,
+      "loss": 0.5154,
+      "step": 16018
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986254212824492e-05,
+      "loss": 0.4761,
+      "step": 16019
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986251498081508e-05,
+      "loss": 0.5064,
+      "step": 16020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862487830723297e-05,
+      "loss": 0.4246,
+      "step": 16021
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862460677969585e-05,
+      "loss": 0.4838,
+      "step": 16022
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986243352255395e-05,
+      "loss": 0.5334,
+      "step": 16023
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862406364476398e-05,
+      "loss": 0.4605,
+      "step": 16024
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862379203736934e-05,
+      "loss": 0.4921,
+      "step": 16025
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862352040335567e-05,
+      "loss": 0.388,
+      "step": 16026
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862324874272307e-05,
+      "loss": 0.5293,
+      "step": 16027
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986229770554716e-05,
+      "loss": 0.4943,
+      "step": 16028
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986227053416013e-05,
+      "loss": 0.5399,
+      "step": 16029
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986224336011123e-05,
+      "loss": 0.5557,
+      "step": 16030
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862216183400463e-05,
+      "loss": 0.4565,
+      "step": 16031
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862189004027842e-05,
+      "loss": 0.496,
+      "step": 16032
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862161821993363e-05,
+      "loss": 0.4324,
+      "step": 16033
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862134637297043e-05,
+      "loss": 0.4077,
+      "step": 16034
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862107449938887e-05,
+      "loss": 0.4699,
+      "step": 16035
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862080259918907e-05,
+      "loss": 0.3909,
+      "step": 16036
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9862053067237098e-05,
+      "loss": 0.4393,
+      "step": 16037
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986202587189348e-05,
+      "loss": 0.4157,
+      "step": 16038
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986199867388805e-05,
+      "loss": 0.4627,
+      "step": 16039
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861971473220826e-05,
+      "loss": 0.5473,
+      "step": 16040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861944269891806e-05,
+      "loss": 0.4998,
+      "step": 16041
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861917063901003e-05,
+      "loss": 0.5311,
+      "step": 16042
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861889855248422e-05,
+      "loss": 0.4897,
+      "step": 16043
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986186264393407e-05,
+      "loss": 0.4843,
+      "step": 16044
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861835429957955e-05,
+      "loss": 0.4625,
+      "step": 16045
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861808213320086e-05,
+      "loss": 0.4223,
+      "step": 16046
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986178099402047e-05,
+      "loss": 0.5302,
+      "step": 16047
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861753772059108e-05,
+      "loss": 0.5111,
+      "step": 16048
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986172654743602e-05,
+      "loss": 0.4479,
+      "step": 16049
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.98616993201512e-05,
+      "loss": 0.4778,
+      "step": 16050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861672090204665e-05,
+      "loss": 0.4702,
+      "step": 16051
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861644857596417e-05,
+      "loss": 0.5367,
+      "step": 16052
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861617622326463e-05,
+      "loss": 0.4937,
+      "step": 16053
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861590384394813e-05,
+      "loss": 0.4938,
+      "step": 16054
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861563143801476e-05,
+      "loss": 0.4138,
+      "step": 16055
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861535900546455e-05,
+      "loss": 0.5129,
+      "step": 16056
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986150865462976e-05,
+      "loss": 0.4804,
+      "step": 16057
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861481406051397e-05,
+      "loss": 0.4279,
+      "step": 16058
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986145415481138e-05,
+      "loss": 0.4615,
+      "step": 16059
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.98614269009097e-05,
+      "loss": 0.5476,
+      "step": 16060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861399644346382e-05,
+      "loss": 0.5385,
+      "step": 16061
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861372385121425e-05,
+      "loss": 0.446,
+      "step": 16062
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861345123234836e-05,
+      "loss": 0.4703,
+      "step": 16063
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861317858686625e-05,
+      "loss": 0.3601,
+      "step": 16064
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861290591476797e-05,
+      "loss": 0.4426,
+      "step": 16065
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986126332160536e-05,
+      "loss": 0.5072,
+      "step": 16066
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861236049072326e-05,
+      "loss": 0.4983,
+      "step": 16067
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861208773877694e-05,
+      "loss": 0.5398,
+      "step": 16068
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861181496021478e-05,
+      "loss": 0.4627,
+      "step": 16069
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986115421550368e-05,
+      "loss": 0.4921,
+      "step": 16070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861126932324316e-05,
+      "loss": 0.5126,
+      "step": 16071
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986109964648338e-05,
+      "loss": 0.4902,
+      "step": 16072
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861072357980893e-05,
+      "loss": 0.4937,
+      "step": 16073
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861045066816857e-05,
+      "loss": 0.4947,
+      "step": 16074
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9861017772991275e-05,
+      "loss": 0.5591,
+      "step": 16075
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986099047650416e-05,
+      "loss": 0.5704,
+      "step": 16076
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860963177355516e-05,
+      "loss": 0.4354,
+      "step": 16077
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860935875545358e-05,
+      "loss": 0.5145,
+      "step": 16078
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986090857107368e-05,
+      "loss": 0.4853,
+      "step": 16079
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.98608812639405e-05,
+      "loss": 0.4266,
+      "step": 16080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860853954145823e-05,
+      "loss": 0.4229,
+      "step": 16081
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860826641689653e-05,
+      "loss": 0.4619,
+      "step": 16082
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860799326572e-05,
+      "loss": 0.4385,
+      "step": 16083
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986077200879287e-05,
+      "loss": 0.4589,
+      "step": 16084
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860744688352277e-05,
+      "loss": 0.4045,
+      "step": 16085
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986071736525022e-05,
+      "loss": 0.4926,
+      "step": 16086
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860690039486708e-05,
+      "loss": 0.5372,
+      "step": 16087
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986066271106175e-05,
+      "loss": 0.5094,
+      "step": 16088
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986063537997535e-05,
+      "loss": 0.4395,
+      "step": 16089
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860608046227524e-05,
+      "loss": 0.432,
+      "step": 16090
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986058070981827e-05,
+      "loss": 0.4209,
+      "step": 16091
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.98605533707476e-05,
+      "loss": 0.4218,
+      "step": 16092
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860526029015522e-05,
+      "loss": 0.4845,
+      "step": 16093
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860498684622045e-05,
+      "loss": 0.4952,
+      "step": 16094
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860471337567168e-05,
+      "loss": 0.4198,
+      "step": 16095
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860443987850906e-05,
+      "loss": 0.4899,
+      "step": 16096
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860416635473263e-05,
+      "loss": 0.5428,
+      "step": 16097
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860389280434246e-05,
+      "loss": 0.4255,
+      "step": 16098
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860361922733868e-05,
+      "loss": 0.5065,
+      "step": 16099
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986033456237213e-05,
+      "loss": 0.4242,
+      "step": 16100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986030719934904e-05,
+      "loss": 0.5085,
+      "step": 16101
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986027983366461e-05,
+      "loss": 0.3898,
+      "step": 16102
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860252465318843e-05,
+      "loss": 0.4363,
+      "step": 16103
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860225094311747e-05,
+      "loss": 0.4674,
+      "step": 16104
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986019772064333e-05,
+      "loss": 0.5523,
+      "step": 16105
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860170344313602e-05,
+      "loss": 0.4276,
+      "step": 16106
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860142965322567e-05,
+      "loss": 0.5366,
+      "step": 16107
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.986011558367023e-05,
+      "loss": 0.465,
+      "step": 16108
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860088199356604e-05,
+      "loss": 0.4687,
+      "step": 16109
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860060812381694e-05,
+      "loss": 0.4619,
+      "step": 16110
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860033422745507e-05,
+      "loss": 0.5284,
+      "step": 16111
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9860006030448056e-05,
+      "loss": 0.524,
+      "step": 16112
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9859978635489338e-05,
+      "loss": 0.5011,
+      "step": 16113
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9859951237869365e-05,
+      "loss": 0.4768,
+      "step": 16114
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9859923837588146e-05,
+      "loss": 0.5123,
+      "step": 16115
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.985989643464569e-05,
+      "loss": 0.4824,
+      "step": 16116
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859869029042e-05,
+      "loss": 0.4754,
+      "step": 16117
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859841620777086e-05,
+      "loss": 0.5169,
+      "step": 16118
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859814209850953e-05,
+      "loss": 0.4744,
+      "step": 16119
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859786796263613e-05,
+      "loss": 0.4795,
+      "step": 16120
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859759380015067e-05,
+      "loss": 0.4326,
+      "step": 16121
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859731961105328e-05,
+      "loss": 0.4262,
+      "step": 16122
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.98597045395344e-05,
+      "loss": 0.5706,
+      "step": 16123
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985967711530229e-05,
+      "loss": 0.556,
+      "step": 16124
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859649688409012e-05,
+      "loss": 0.5223,
+      "step": 16125
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859622258854565e-05,
+      "loss": 0.5272,
+      "step": 16126
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985959482663896e-05,
+      "loss": 0.4816,
+      "step": 16127
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859567391762206e-05,
+      "loss": 0.4471,
+      "step": 16128
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859539954224308e-05,
+      "loss": 0.4442,
+      "step": 16129
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859512514025276e-05,
+      "loss": 0.4364,
+      "step": 16130
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859485071165114e-05,
+      "loss": 0.4085,
+      "step": 16131
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859457625643828e-05,
+      "loss": 0.4712,
+      "step": 16132
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985943017746143e-05,
+      "loss": 0.5646,
+      "step": 16133
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859402726617927e-05,
+      "loss": 0.4544,
+      "step": 16134
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859375273113322e-05,
+      "loss": 0.5021,
+      "step": 16135
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859347816947628e-05,
+      "loss": 0.5128,
+      "step": 16136
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985932035812085e-05,
+      "loss": 0.4506,
+      "step": 16137
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859292896632997e-05,
+      "loss": 0.4092,
+      "step": 16138
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985926543248407e-05,
+      "loss": 0.5341,
+      "step": 16139
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859237965674087e-05,
+      "loss": 0.4499,
+      "step": 16140
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859210496203046e-05,
+      "loss": 0.458,
+      "step": 16141
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859183024070957e-05,
+      "loss": 0.4663,
+      "step": 16142
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859155549277832e-05,
+      "loss": 0.4251,
+      "step": 16143
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859128071823672e-05,
+      "loss": 0.596,
+      "step": 16144
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985910059170849e-05,
+      "loss": 0.4754,
+      "step": 16145
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859073108932286e-05,
+      "loss": 0.437,
+      "step": 16146
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9859045623495078e-05,
+      "loss": 0.5156,
+      "step": 16147
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985901813539686e-05,
+      "loss": 0.4819,
+      "step": 16148
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858990644637653e-05,
+      "loss": 0.5257,
+      "step": 16149
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858963151217458e-05,
+      "loss": 0.4294,
+      "step": 16150
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985893565513628e-05,
+      "loss": 0.4349,
+      "step": 16151
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985890815639413e-05,
+      "loss": 0.415,
+      "step": 16152
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858880654991014e-05,
+      "loss": 0.4926,
+      "step": 16153
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985885315092694e-05,
+      "loss": 0.4207,
+      "step": 16154
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858825644201918e-05,
+      "loss": 0.4657,
+      "step": 16155
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858798134815954e-05,
+      "loss": 0.4774,
+      "step": 16156
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985877062276905e-05,
+      "loss": 0.5047,
+      "step": 16157
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985874310806122e-05,
+      "loss": 0.4227,
+      "step": 16158
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858715590692467e-05,
+      "loss": 0.4762,
+      "step": 16159
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858688070662803e-05,
+      "loss": 0.6076,
+      "step": 16160
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985866054797223e-05,
+      "loss": 0.5284,
+      "step": 16161
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858633022620764e-05,
+      "loss": 0.443,
+      "step": 16162
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858605494608403e-05,
+      "loss": 0.4752,
+      "step": 16163
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985857796393516e-05,
+      "loss": 0.3951,
+      "step": 16164
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858550430601037e-05,
+      "loss": 0.4537,
+      "step": 16165
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858522894606048e-05,
+      "loss": 0.4459,
+      "step": 16166
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.98584953559502e-05,
+      "loss": 0.5056,
+      "step": 16167
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858467814633493e-05,
+      "loss": 0.4761,
+      "step": 16168
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985844027065594e-05,
+      "loss": 0.519,
+      "step": 16169
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985841272401755e-05,
+      "loss": 0.5207,
+      "step": 16170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858385174718328e-05,
+      "loss": 0.5339,
+      "step": 16171
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858357622758283e-05,
+      "loss": 0.4318,
+      "step": 16172
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858330068137422e-05,
+      "loss": 0.4801,
+      "step": 16173
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858302510855746e-05,
+      "loss": 0.426,
+      "step": 16174
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858274950913273e-05,
+      "loss": 0.5204,
+      "step": 16175
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858247388310004e-05,
+      "loss": 0.4891,
+      "step": 16176
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858219823045948e-05,
+      "loss": 0.4281,
+      "step": 16177
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985819225512111e-05,
+      "loss": 0.6326,
+      "step": 16178
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858164684535504e-05,
+      "loss": 0.4971,
+      "step": 16179
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985813711128913e-05,
+      "loss": 0.4953,
+      "step": 16180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858109535382004e-05,
+      "loss": 0.5892,
+      "step": 16181
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858081956814123e-05,
+      "loss": 0.5078,
+      "step": 16182
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.98580543755855e-05,
+      "loss": 0.4916,
+      "step": 16183
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9858026791696146e-05,
+      "loss": 0.4575,
+      "step": 16184
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985799920514606e-05,
+      "loss": 0.4839,
+      "step": 16185
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857971615935257e-05,
+      "loss": 0.5363,
+      "step": 16186
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985794402406374e-05,
+      "loss": 0.4828,
+      "step": 16187
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857916429531518e-05,
+      "loss": 0.4517,
+      "step": 16188
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857888832338595e-05,
+      "loss": 0.4402,
+      "step": 16189
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857861232484986e-05,
+      "loss": 0.4704,
+      "step": 16190
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857833629970694e-05,
+      "loss": 0.4925,
+      "step": 16191
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857806024795722e-05,
+      "loss": 0.4703,
+      "step": 16192
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857778416960088e-05,
+      "loss": 0.4731,
+      "step": 16193
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985775080646379e-05,
+      "loss": 0.4137,
+      "step": 16194
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985772319330684e-05,
+      "loss": 0.6459,
+      "step": 16195
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857695577489243e-05,
+      "loss": 0.456,
+      "step": 16196
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857667959011012e-05,
+      "loss": 0.4692,
+      "step": 16197
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857640337872146e-05,
+      "loss": 0.4587,
+      "step": 16198
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985761271407266e-05,
+      "loss": 0.4449,
+      "step": 16199
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857585087612555e-05,
+      "loss": 0.4666,
+      "step": 16200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857557458491847e-05,
+      "loss": 0.4549,
+      "step": 16201
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857529826710533e-05,
+      "loss": 0.4595,
+      "step": 16202
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985750219226863e-05,
+      "loss": 0.4703,
+      "step": 16203
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857474555166136e-05,
+      "loss": 0.4818,
+      "step": 16204
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857446915403065e-05,
+      "loss": 0.4409,
+      "step": 16205
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857419272979423e-05,
+      "loss": 0.4901,
+      "step": 16206
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.985739162789522e-05,
+      "loss": 0.4785,
+      "step": 16207
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857363980150456e-05,
+      "loss": 0.4983,
+      "step": 16208
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857336329745148e-05,
+      "loss": 0.4566,
+      "step": 16209
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.98573086766793e-05,
+      "loss": 0.4558,
+      "step": 16210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857281020952914e-05,
+      "loss": 0.4925,
+      "step": 16211
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857253362566004e-05,
+      "loss": 0.5654,
+      "step": 16212
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857225701518574e-05,
+      "loss": 0.5066,
+      "step": 16213
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9857198037810634e-05,
+      "loss": 0.4176,
+      "step": 16214
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985717037144219e-05,
+      "loss": 0.6473,
+      "step": 16215
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9857142702413247e-05,
+      "loss": 0.459,
+      "step": 16216
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985711503072382e-05,
+      "loss": 0.6044,
+      "step": 16217
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985708735637391e-05,
+      "loss": 0.5065,
+      "step": 16218
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9857059679363525e-05,
+      "loss": 0.4187,
+      "step": 16219
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985703199969267e-05,
+      "loss": 0.4042,
+      "step": 16220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9857004317361362e-05,
+      "loss": 0.4316,
+      "step": 16221
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856976632369602e-05,
+      "loss": 0.5453,
+      "step": 16222
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856948944717392e-05,
+      "loss": 0.5595,
+      "step": 16223
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985692125440475e-05,
+      "loss": 0.4804,
+      "step": 16224
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985689356143168e-05,
+      "loss": 0.5545,
+      "step": 16225
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856865865798183e-05,
+      "loss": 0.4019,
+      "step": 16226
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985683816750428e-05,
+      "loss": 0.4255,
+      "step": 16227
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856810466549966e-05,
+      "loss": 0.4985,
+      "step": 16228
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985678276293525e-05,
+      "loss": 0.4245,
+      "step": 16229
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856755056660148e-05,
+      "loss": 0.4061,
+      "step": 16230
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856727347724656e-05,
+      "loss": 0.5769,
+      "step": 16231
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985669963612879e-05,
+      "loss": 0.456,
+      "step": 16232
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856671921872553e-05,
+      "loss": 0.512,
+      "step": 16233
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985664420495596e-05,
+      "loss": 0.4366,
+      "step": 16234
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856616485379003e-05,
+      "loss": 0.5248,
+      "step": 16235
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856588763141706e-05,
+      "loss": 0.5299,
+      "step": 16236
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856561038244065e-05,
+      "loss": 0.5076,
+      "step": 16237
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856533310686097e-05,
+      "loss": 0.5266,
+      "step": 16238
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856505580467802e-05,
+      "loss": 0.462,
+      "step": 16239
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985647784758919e-05,
+      "loss": 0.5357,
+      "step": 16240
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985645011205027e-05,
+      "loss": 0.4337,
+      "step": 16241
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985642237385105e-05,
+      "loss": 0.4678,
+      "step": 16242
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856394632991532e-05,
+      "loss": 0.4126,
+      "step": 16243
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856366889471726e-05,
+      "loss": 0.5608,
+      "step": 16244
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856339143291644e-05,
+      "loss": 0.4984,
+      "step": 16245
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856311394451285e-05,
+      "loss": 0.3906,
+      "step": 16246
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985628364295067e-05,
+      "loss": 0.4158,
+      "step": 16247
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985625588878979e-05,
+      "loss": 0.4353,
+      "step": 16248
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856228131968663e-05,
+      "loss": 0.5186,
+      "step": 16249
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856200372487297e-05,
+      "loss": 0.4935,
+      "step": 16250
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856172610345692e-05,
+      "loss": 0.5204,
+      "step": 16251
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985614484554386e-05,
+      "loss": 0.4417,
+      "step": 16252
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856117078081812e-05,
+      "loss": 0.5361,
+      "step": 16253
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985608930795955e-05,
+      "loss": 0.4599,
+      "step": 16254
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985606153517708e-05,
+      "loss": 0.5147,
+      "step": 16255
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985603375973442e-05,
+      "loss": 0.4838,
+      "step": 16256
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9856005981631566e-05,
+      "loss": 0.5359,
+      "step": 16257
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985597820086853e-05,
+      "loss": 0.51,
+      "step": 16258
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985595041744532e-05,
+      "loss": 0.4672,
+      "step": 16259
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855922631361944e-05,
+      "loss": 0.5171,
+      "step": 16260
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855894842618406e-05,
+      "loss": 0.4327,
+      "step": 16261
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855867051214716e-05,
+      "loss": 0.4432,
+      "step": 16262
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855839257150885e-05,
+      "loss": 0.4977,
+      "step": 16263
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855811460426914e-05,
+      "loss": 0.5219,
+      "step": 16264
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855783661042815e-05,
+      "loss": 0.4776,
+      "step": 16265
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855755858998593e-05,
+      "loss": 0.4927,
+      "step": 16266
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855728054294253e-05,
+      "loss": 0.6001,
+      "step": 16267
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855700246929813e-05,
+      "loss": 0.4318,
+      "step": 16268
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985567243690527e-05,
+      "loss": 0.4599,
+      "step": 16269
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855644624220633e-05,
+      "loss": 0.4466,
+      "step": 16270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855616808875916e-05,
+      "loss": 0.4472,
+      "step": 16271
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855588990871116e-05,
+      "loss": 0.3877,
+      "step": 16272
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985556117020625e-05,
+      "loss": 0.4316,
+      "step": 16273
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985553334688132e-05,
+      "loss": 0.4724,
+      "step": 16274
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985550552089634e-05,
+      "loss": 0.4254,
+      "step": 16275
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855477692251307e-05,
+      "loss": 0.5181,
+      "step": 16276
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985544986094624e-05,
+      "loss": 0.4516,
+      "step": 16277
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855422026981137e-05,
+      "loss": 0.5072,
+      "step": 16278
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985539419035601e-05,
+      "loss": 0.4368,
+      "step": 16279
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855366351070867e-05,
+      "loss": 0.4573,
+      "step": 16280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855338509125714e-05,
+      "loss": 0.4662,
+      "step": 16281
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855310664520563e-05,
+      "loss": 0.4091,
+      "step": 16282
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855282817255413e-05,
+      "loss": 0.4723,
+      "step": 16283
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855254967330278e-05,
+      "loss": 0.498,
+      "step": 16284
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855227114745164e-05,
+      "loss": 0.4709,
+      "step": 16285
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855199259500076e-05,
+      "loss": 0.5109,
+      "step": 16286
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855171401595024e-05,
+      "loss": 0.5007,
+      "step": 16287
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855143541030016e-05,
+      "loss": 0.4614,
+      "step": 16288
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855115677805058e-05,
+      "loss": 0.4461,
+      "step": 16289
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985508781192016e-05,
+      "loss": 0.4888,
+      "step": 16290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855059943375325e-05,
+      "loss": 0.4261,
+      "step": 16291
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855032072170564e-05,
+      "loss": 0.4791,
+      "step": 16292
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9855004198305886e-05,
+      "loss": 0.5458,
+      "step": 16293
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854976321781298e-05,
+      "loss": 0.4657,
+      "step": 16294
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.98549484425968e-05,
+      "loss": 0.5266,
+      "step": 16295
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985492056075241e-05,
+      "loss": 0.4407,
+      "step": 16296
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854892676248127e-05,
+      "loss": 0.6393,
+      "step": 16297
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854864789083964e-05,
+      "loss": 0.5075,
+      "step": 16298
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985483689925993e-05,
+      "loss": 0.5295,
+      "step": 16299
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854809006776023e-05,
+      "loss": 0.4535,
+      "step": 16300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985478111163226e-05,
+      "loss": 0.4924,
+      "step": 16301
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985475321382865e-05,
+      "loss": 0.5266,
+      "step": 16302
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854725313365192e-05,
+      "loss": 0.4644,
+      "step": 16303
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854697410241895e-05,
+      "loss": 0.4463,
+      "step": 16304
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854669504458776e-05,
+      "loss": 0.5085,
+      "step": 16305
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854641596015828e-05,
+      "loss": 0.475,
+      "step": 16306
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854613684913067e-05,
+      "loss": 0.5082,
+      "step": 16307
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854585771150505e-05,
+      "loss": 0.5818,
+      "step": 16308
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854557854728144e-05,
+      "loss": 0.4401,
+      "step": 16309
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854529935645987e-05,
+      "loss": 0.5345,
+      "step": 16310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985450201390405e-05,
+      "loss": 0.5437,
+      "step": 16311
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9854474089502336e-05,
+      "loss": 0.4331,
+      "step": 16312
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.985444616244085e-05,
+      "loss": 0.4549,
+      "step": 16313
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985441823271961e-05,
+      "loss": 0.3896,
+      "step": 16314
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985439030033861e-05,
+      "loss": 0.5543,
+      "step": 16315
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9854362365297867e-05,
+      "loss": 0.5164,
+      "step": 16316
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9854334427597384e-05,
+      "loss": 0.5593,
+      "step": 16317
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985430648723717e-05,
+      "loss": 0.4896,
+      "step": 16318
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9854278544217233e-05,
+      "loss": 0.4672,
+      "step": 16319
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9854250598537584e-05,
+      "loss": 0.4986,
+      "step": 16320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9854222650198222e-05,
+      "loss": 0.443,
+      "step": 16321
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985419469919916e-05,
+      "loss": 0.5126,
+      "step": 16322
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9854166745540407e-05,
+      "loss": 0.4827,
+      "step": 16323
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9854138789221967e-05,
+      "loss": 0.5317,
+      "step": 16324
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985411083024385e-05,
+      "loss": 0.4742,
+      "step": 16325
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9854082868606057e-05,
+      "loss": 0.471,
+      "step": 16326
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9854054904308605e-05,
+      "loss": 0.3891,
+      "step": 16327
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.98540269373515e-05,
+      "loss": 0.447,
+      "step": 16328
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853998967734744e-05,
+      "loss": 0.473,
+      "step": 16329
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853970995458346e-05,
+      "loss": 0.4514,
+      "step": 16330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853943020522318e-05,
+      "loss": 0.443,
+      "step": 16331
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853915042926665e-05,
+      "loss": 0.4534,
+      "step": 16332
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853887062671396e-05,
+      "loss": 0.4706,
+      "step": 16333
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985385907975651e-05,
+      "loss": 0.4746,
+      "step": 16334
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853831094182028e-05,
+      "loss": 0.4637,
+      "step": 16335
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985380310594795e-05,
+      "loss": 0.4157,
+      "step": 16336
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853775115054282e-05,
+      "loss": 0.5551,
+      "step": 16337
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853747121501033e-05,
+      "loss": 0.4742,
+      "step": 16338
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853719125288212e-05,
+      "loss": 0.5033,
+      "step": 16339
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985369112641583e-05,
+      "loss": 0.4244,
+      "step": 16340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985366312488389e-05,
+      "loss": 0.4288,
+      "step": 16341
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853635120692395e-05,
+      "loss": 0.4916,
+      "step": 16342
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853607113841365e-05,
+      "loss": 0.4866,
+      "step": 16343
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853579104330795e-05,
+      "loss": 0.4821,
+      "step": 16344
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.98535510921607e-05,
+      "loss": 0.5241,
+      "step": 16345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853523077331088e-05,
+      "loss": 0.5706,
+      "step": 16346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853495059841958e-05,
+      "loss": 0.568,
+      "step": 16347
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853467039693328e-05,
+      "loss": 0.4241,
+      "step": 16348
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.98534390168852e-05,
+      "loss": 0.457,
+      "step": 16349
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853410991417587e-05,
+      "loss": 0.54,
+      "step": 16350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853382963290485e-05,
+      "loss": 0.4708,
+      "step": 16351
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985335493250391e-05,
+      "loss": 0.4421,
+      "step": 16352
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853326899057873e-05,
+      "loss": 0.4688,
+      "step": 16353
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853298862952376e-05,
+      "loss": 0.5557,
+      "step": 16354
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853270824187425e-05,
+      "loss": 0.4796,
+      "step": 16355
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985324278276303e-05,
+      "loss": 0.4967,
+      "step": 16356
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.98532147386792e-05,
+      "loss": 0.468,
+      "step": 16357
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853186691935943e-05,
+      "loss": 0.5025,
+      "step": 16358
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853158642533264e-05,
+      "loss": 0.4432,
+      "step": 16359
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853130590471167e-05,
+      "loss": 0.493,
+      "step": 16360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985310253574967e-05,
+      "loss": 0.4318,
+      "step": 16361
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853074478368772e-05,
+      "loss": 0.4736,
+      "step": 16362
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9853046418328482e-05,
+      "loss": 0.4939,
+      "step": 16363
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985301835562881e-05,
+      "loss": 0.4828,
+      "step": 16364
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852990290269765e-05,
+      "loss": 0.5072,
+      "step": 16365
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985296222225135e-05,
+      "loss": 0.5075,
+      "step": 16366
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852934151573572e-05,
+      "loss": 0.4459,
+      "step": 16367
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852906078236443e-05,
+      "loss": 0.4641,
+      "step": 16368
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985287800223997e-05,
+      "loss": 0.4902,
+      "step": 16369
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852849923584158e-05,
+      "loss": 0.4216,
+      "step": 16370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852821842269012e-05,
+      "loss": 0.4901,
+      "step": 16371
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852793758294545e-05,
+      "loss": 0.4327,
+      "step": 16372
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852765671660767e-05,
+      "loss": 0.4505,
+      "step": 16373
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985273758236768e-05,
+      "loss": 0.5138,
+      "step": 16374
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985270949041529e-05,
+      "loss": 0.3833,
+      "step": 16375
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852681395803607e-05,
+      "loss": 0.4208,
+      "step": 16376
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852653298532644e-05,
+      "loss": 0.579,
+      "step": 16377
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.98526251986024e-05,
+      "loss": 0.544,
+      "step": 16378
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985259709601289e-05,
+      "loss": 0.533,
+      "step": 16379
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852568990764118e-05,
+      "loss": 0.4448,
+      "step": 16380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852540882856088e-05,
+      "loss": 0.4498,
+      "step": 16381
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985251277228881e-05,
+      "loss": 0.4348,
+      "step": 16382
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852484659062295e-05,
+      "loss": 0.4651,
+      "step": 16383
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852456543176554e-05,
+      "loss": 0.4869,
+      "step": 16384
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985242842463158e-05,
+      "loss": 0.5618,
+      "step": 16385
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852400303427395e-05,
+      "loss": 0.5648,
+      "step": 16386
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852372179564e-05,
+      "loss": 0.4748,
+      "step": 16387
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852344053041403e-05,
+      "loss": 0.5175,
+      "step": 16388
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852315923859612e-05,
+      "loss": 0.5192,
+      "step": 16389
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852287792018632e-05,
+      "loss": 0.5361,
+      "step": 16390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985225965751848e-05,
+      "loss": 0.4536,
+      "step": 16391
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852231520359154e-05,
+      "loss": 0.4608,
+      "step": 16392
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852203380540666e-05,
+      "loss": 0.4997,
+      "step": 16393
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985217523806302e-05,
+      "loss": 0.4619,
+      "step": 16394
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852147092926227e-05,
+      "loss": 0.5992,
+      "step": 16395
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985211894513029e-05,
+      "loss": 0.4453,
+      "step": 16396
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852090794675224e-05,
+      "loss": 0.5192,
+      "step": 16397
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852062641561032e-05,
+      "loss": 0.418,
+      "step": 16398
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852034485787722e-05,
+      "loss": 0.4349,
+      "step": 16399
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9852006327355302e-05,
+      "loss": 0.4703,
+      "step": 16400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9851978166263778e-05,
+      "loss": 0.4578,
+      "step": 16401
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985195000251316e-05,
+      "loss": 0.4275,
+      "step": 16402
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9851921836103457e-05,
+      "loss": 0.4827,
+      "step": 16403
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985189366703467e-05,
+      "loss": 0.4357,
+      "step": 16404
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9851865495306815e-05,
+      "loss": 0.5144,
+      "step": 16405
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9851837320919892e-05,
+      "loss": 0.4142,
+      "step": 16406
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9851809143873916e-05,
+      "loss": 0.5419,
+      "step": 16407
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9851780964168887e-05,
+      "loss": 0.3876,
+      "step": 16408
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9851752781804818e-05,
+      "loss": 0.4571,
+      "step": 16409
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9851724596781712e-05,
+      "loss": 0.4572,
+      "step": 16410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.985169640909958e-05,
+      "loss": 0.4455,
+      "step": 16411
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851668218758433e-05,
+      "loss": 0.5077,
+      "step": 16412
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851640025758273e-05,
+      "loss": 0.4875,
+      "step": 16413
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985161183009911e-05,
+      "loss": 0.5745,
+      "step": 16414
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985158363178095e-05,
+      "loss": 0.5147,
+      "step": 16415
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.98515554308038e-05,
+      "loss": 0.4619,
+      "step": 16416
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851527227167672e-05,
+      "loss": 0.3952,
+      "step": 16417
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851499020872567e-05,
+      "loss": 0.6538,
+      "step": 16418
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851470811918498e-05,
+      "loss": 0.4438,
+      "step": 16419
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851442600305473e-05,
+      "loss": 0.4745,
+      "step": 16420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851414386033496e-05,
+      "loss": 0.4055,
+      "step": 16421
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985138616910258e-05,
+      "loss": 0.4618,
+      "step": 16422
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851357949512723e-05,
+      "loss": 0.526,
+      "step": 16423
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985132972726394e-05,
+      "loss": 0.449,
+      "step": 16424
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851301502356234e-05,
+      "loss": 0.4577,
+      "step": 16425
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985127327478962e-05,
+      "loss": 0.5392,
+      "step": 16426
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.98512450445641e-05,
+      "loss": 0.4895,
+      "step": 16427
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851216811679687e-05,
+      "loss": 0.4521,
+      "step": 16428
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985118857613638e-05,
+      "loss": 0.5193,
+      "step": 16429
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851160337934194e-05,
+      "loss": 0.4696,
+      "step": 16430
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985113209707313e-05,
+      "loss": 0.4697,
+      "step": 16431
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851103853553202e-05,
+      "loss": 0.4545,
+      "step": 16432
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851075607374413e-05,
+      "loss": 0.5188,
+      "step": 16433
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9851047358536773e-05,
+      "loss": 0.5073,
+      "step": 16434
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985101910704029e-05,
+      "loss": 0.5635,
+      "step": 16435
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850990852884972e-05,
+      "loss": 0.4918,
+      "step": 16436
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850962596070827e-05,
+      "loss": 0.5067,
+      "step": 16437
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985093433659786e-05,
+      "loss": 0.4777,
+      "step": 16438
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850906074466078e-05,
+      "loss": 0.5155,
+      "step": 16439
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985087780967549e-05,
+      "loss": 0.4993,
+      "step": 16440
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850849542226105e-05,
+      "loss": 0.493,
+      "step": 16441
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985082127211793e-05,
+      "loss": 0.4666,
+      "step": 16442
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850792999350973e-05,
+      "loss": 0.457,
+      "step": 16443
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850764723925238e-05,
+      "loss": 0.4468,
+      "step": 16444
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850736445840736e-05,
+      "loss": 0.5283,
+      "step": 16445
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985070816509748e-05,
+      "loss": 0.4455,
+      "step": 16446
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850679881695465e-05,
+      "loss": 0.5128,
+      "step": 16447
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850651595634706e-05,
+      "loss": 0.4338,
+      "step": 16448
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850623306915215e-05,
+      "loss": 0.4779,
+      "step": 16449
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850595015536992e-05,
+      "loss": 0.5062,
+      "step": 16450
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850566721500045e-05,
+      "loss": 0.4611,
+      "step": 16451
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850538424804387e-05,
+      "loss": 0.4555,
+      "step": 16452
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985051012545002e-05,
+      "loss": 0.5679,
+      "step": 16453
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850481823436958e-05,
+      "loss": 0.4983,
+      "step": 16454
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850453518765204e-05,
+      "loss": 0.5002,
+      "step": 16455
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850425211434766e-05,
+      "loss": 0.4889,
+      "step": 16456
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985039690144565e-05,
+      "loss": 0.5391,
+      "step": 16457
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985036858879787e-05,
+      "loss": 0.5377,
+      "step": 16458
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850340273491422e-05,
+      "loss": 0.479,
+      "step": 16459
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985031195552633e-05,
+      "loss": 0.4561,
+      "step": 16460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850283634902587e-05,
+      "loss": 0.437,
+      "step": 16461
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985025531162021e-05,
+      "loss": 0.4442,
+      "step": 16462
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.98502269856792e-05,
+      "loss": 0.4646,
+      "step": 16463
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850198657079568e-05,
+      "loss": 0.4847,
+      "step": 16464
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850170325821323e-05,
+      "loss": 0.5842,
+      "step": 16465
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985014199190447e-05,
+      "loss": 0.5034,
+      "step": 16466
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985011365532902e-05,
+      "loss": 0.4968,
+      "step": 16467
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850085316094977e-05,
+      "loss": 0.4483,
+      "step": 16468
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985005697420235e-05,
+      "loss": 0.4882,
+      "step": 16469
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9850028629651145e-05,
+      "loss": 0.4881,
+      "step": 16470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.985000028244137e-05,
+      "loss": 0.5169,
+      "step": 16471
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849971932573035e-05,
+      "loss": 0.5024,
+      "step": 16472
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984994358004615e-05,
+      "loss": 0.514,
+      "step": 16473
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849915224860714e-05,
+      "loss": 0.4168,
+      "step": 16474
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849886867016744e-05,
+      "loss": 0.5323,
+      "step": 16475
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984985850651424e-05,
+      "loss": 0.4293,
+      "step": 16476
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849830143353218e-05,
+      "loss": 0.4662,
+      "step": 16477
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849801777533674e-05,
+      "loss": 0.4277,
+      "step": 16478
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984977340905563e-05,
+      "loss": 0.4244,
+      "step": 16479
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984974503791908e-05,
+      "loss": 0.4533,
+      "step": 16480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984971666412404e-05,
+      "loss": 0.4287,
+      "step": 16481
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849688287670517e-05,
+      "loss": 0.4442,
+      "step": 16482
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849659908558515e-05,
+      "loss": 0.4485,
+      "step": 16483
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849631526788044e-05,
+      "loss": 0.5424,
+      "step": 16484
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849603142359112e-05,
+      "loss": 0.457,
+      "step": 16485
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849574755271726e-05,
+      "loss": 0.474,
+      "step": 16486
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849546365525897e-05,
+      "loss": 0.4111,
+      "step": 16487
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849517973121625e-05,
+      "loss": 0.4301,
+      "step": 16488
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849489578058923e-05,
+      "loss": 0.4417,
+      "step": 16489
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849461180337797e-05,
+      "loss": 0.4954,
+      "step": 16490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849432779958256e-05,
+      "loss": 0.5372,
+      "step": 16491
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849404376920308e-05,
+      "loss": 0.4434,
+      "step": 16492
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849375971223957e-05,
+      "loss": 0.4554,
+      "step": 16493
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849347562869218e-05,
+      "loss": 0.479,
+      "step": 16494
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984931915185609e-05,
+      "loss": 0.417,
+      "step": 16495
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849290738184584e-05,
+      "loss": 0.5188,
+      "step": 16496
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984926232185471e-05,
+      "loss": 0.4984,
+      "step": 16497
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984923390286648e-05,
+      "loss": 0.4205,
+      "step": 16498
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849205481219885e-05,
+      "loss": 0.4297,
+      "step": 16499
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984917705691495e-05,
+      "loss": 0.5315,
+      "step": 16500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849148629951675e-05,
+      "loss": 0.4611,
+      "step": 16501
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984912020033007e-05,
+      "loss": 0.4888,
+      "step": 16502
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.984909176805014e-05,
+      "loss": 0.4502,
+      "step": 16503
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849063333111893e-05,
+      "loss": 0.4697,
+      "step": 16504
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849034895515337e-05,
+      "loss": 0.5055,
+      "step": 16505
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9849006455260486e-05,
+      "loss": 0.4999,
+      "step": 16506
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9848978012347334e-05,
+      "loss": 0.459,
+      "step": 16507
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9848949566775904e-05,
+      "loss": 0.4788,
+      "step": 16508
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9848921118546194e-05,
+      "loss": 0.5183,
+      "step": 16509
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9848892667658214e-05,
+      "loss": 0.5458,
+      "step": 16510
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848864214111972e-05,
+      "loss": 0.5316,
+      "step": 16511
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848835757907474e-05,
+      "loss": 0.4487,
+      "step": 16512
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984880729904473e-05,
+      "loss": 0.4974,
+      "step": 16513
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848778837523747e-05,
+      "loss": 0.5418,
+      "step": 16514
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848750373344536e-05,
+      "loss": 0.4465,
+      "step": 16515
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848721906507097e-05,
+      "loss": 0.5241,
+      "step": 16516
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848693437011443e-05,
+      "loss": 0.4759,
+      "step": 16517
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848664964857578e-05,
+      "loss": 0.5045,
+      "step": 16518
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848636490045518e-05,
+      "loss": 0.4071,
+      "step": 16519
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984860801257526e-05,
+      "loss": 0.4793,
+      "step": 16520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984857953244682e-05,
+      "loss": 0.4713,
+      "step": 16521
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848551049660198e-05,
+      "loss": 0.4356,
+      "step": 16522
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984852256421541e-05,
+      "loss": 0.5184,
+      "step": 16523
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848494076112458e-05,
+      "loss": 0.4774,
+      "step": 16524
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848465585351353e-05,
+      "loss": 0.4693,
+      "step": 16525
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848437091932097e-05,
+      "loss": 0.5138,
+      "step": 16526
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848408595854708e-05,
+      "loss": 0.5207,
+      "step": 16527
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848380097119185e-05,
+      "loss": 0.4592,
+      "step": 16528
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848351595725533e-05,
+      "loss": 0.5228,
+      "step": 16529
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984832309167377e-05,
+      "loss": 0.485,
+      "step": 16530
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848294584963902e-05,
+      "loss": 0.4812,
+      "step": 16531
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848266075595927e-05,
+      "loss": 0.4724,
+      "step": 16532
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984823756356986e-05,
+      "loss": 0.4573,
+      "step": 16533
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984820904888571e-05,
+      "loss": 0.4367,
+      "step": 16534
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984818053154348e-05,
+      "loss": 0.4255,
+      "step": 16535
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984815201154318e-05,
+      "loss": 0.5107,
+      "step": 16536
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848123488884822e-05,
+      "loss": 0.478,
+      "step": 16537
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9848094963568405e-05,
+      "loss": 0.5489,
+      "step": 16538
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984806643559394e-05,
+      "loss": 0.5054,
+      "step": 16539
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984803790496144e-05,
+      "loss": 0.556,
+      "step": 16540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984800937167091e-05,
+      "loss": 0.439,
+      "step": 16541
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847980835722353e-05,
+      "loss": 0.5094,
+      "step": 16542
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984795229711578e-05,
+      "loss": 0.4456,
+      "step": 16543
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.98479237558512e-05,
+      "loss": 0.484,
+      "step": 16544
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847895211928616e-05,
+      "loss": 0.4606,
+      "step": 16545
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984786666534804e-05,
+      "loss": 0.5066,
+      "step": 16546
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984783811610948e-05,
+      "loss": 0.5196,
+      "step": 16547
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847809564212944e-05,
+      "loss": 0.4874,
+      "step": 16548
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847781009658437e-05,
+      "loss": 0.3795,
+      "step": 16549
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847752452445964e-05,
+      "loss": 0.4352,
+      "step": 16550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984772389257554e-05,
+      "loss": 0.4364,
+      "step": 16551
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984769533004717e-05,
+      "loss": 0.4623,
+      "step": 16552
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984766676486086e-05,
+      "loss": 0.5572,
+      "step": 16553
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847638197016618e-05,
+      "loss": 0.5367,
+      "step": 16554
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847609626514456e-05,
+      "loss": 0.581,
+      "step": 16555
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847581053354376e-05,
+      "loss": 0.4718,
+      "step": 16556
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847552477536384e-05,
+      "loss": 0.4178,
+      "step": 16557
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847523899060498e-05,
+      "loss": 0.4648,
+      "step": 16558
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847495317926714e-05,
+      "loss": 0.5467,
+      "step": 16559
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984746673413505e-05,
+      "loss": 0.4585,
+      "step": 16560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847438147685503e-05,
+      "loss": 0.4304,
+      "step": 16561
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984740955857809e-05,
+      "loss": 0.5423,
+      "step": 16562
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847380966812816e-05,
+      "loss": 0.4917,
+      "step": 16563
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847352372389685e-05,
+      "loss": 0.4071,
+      "step": 16564
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984732377530871e-05,
+      "loss": 0.4844,
+      "step": 16565
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984729517556989e-05,
+      "loss": 0.4351,
+      "step": 16566
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847266573173246e-05,
+      "loss": 0.4442,
+      "step": 16567
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847237968118777e-05,
+      "loss": 0.5241,
+      "step": 16568
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847209360406495e-05,
+      "loss": 0.4592,
+      "step": 16569
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.98471807500364e-05,
+      "loss": 0.5201,
+      "step": 16570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847152137008508e-05,
+      "loss": 0.4146,
+      "step": 16571
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984712352132282e-05,
+      "loss": 0.5334,
+      "step": 16572
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847094902979354e-05,
+      "loss": 0.5077,
+      "step": 16573
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847066281978104e-05,
+      "loss": 0.4957,
+      "step": 16574
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847037658319088e-05,
+      "loss": 0.5856,
+      "step": 16575
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9847009032002314e-05,
+      "loss": 0.4352,
+      "step": 16576
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984698040302778e-05,
+      "loss": 0.4846,
+      "step": 16577
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.98469517713955e-05,
+      "loss": 0.5082,
+      "step": 16578
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846923137105487e-05,
+      "loss": 0.5436,
+      "step": 16579
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984689450015774e-05,
+      "loss": 0.4317,
+      "step": 16580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984686586055227e-05,
+      "loss": 0.5481,
+      "step": 16581
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846837218289087e-05,
+      "loss": 0.4116,
+      "step": 16582
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846808573368196e-05,
+      "loss": 0.4539,
+      "step": 16583
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846779925789607e-05,
+      "loss": 0.5363,
+      "step": 16584
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846751275553323e-05,
+      "loss": 0.4523,
+      "step": 16585
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846722622659355e-05,
+      "loss": 0.4756,
+      "step": 16586
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984669396710771e-05,
+      "loss": 0.4999,
+      "step": 16587
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.98466653088984e-05,
+      "loss": 0.4453,
+      "step": 16588
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846636648031426e-05,
+      "loss": 0.5878,
+      "step": 16589
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846607984506798e-05,
+      "loss": 0.423,
+      "step": 16590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846579318324526e-05,
+      "loss": 0.5295,
+      "step": 16591
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846550649484614e-05,
+      "loss": 0.6005,
+      "step": 16592
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846521977987075e-05,
+      "loss": 0.4878,
+      "step": 16593
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846493303831912e-05,
+      "loss": 0.4366,
+      "step": 16594
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846464627019136e-05,
+      "loss": 0.4894,
+      "step": 16595
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984643594754875e-05,
+      "loss": 0.4161,
+      "step": 16596
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846407265420772e-05,
+      "loss": 0.47,
+      "step": 16597
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846378580635193e-05,
+      "loss": 0.4208,
+      "step": 16598
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984634989319204e-05,
+      "loss": 0.5574,
+      "step": 16599
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.98463212030913e-05,
+      "loss": 0.485,
+      "step": 16600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846292510333002e-05,
+      "loss": 0.4352,
+      "step": 16601
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984626381491714e-05,
+      "loss": 0.4695,
+      "step": 16602
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846235116843725e-05,
+      "loss": 0.4804,
+      "step": 16603
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846206416112765e-05,
+      "loss": 0.4726,
+      "step": 16604
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846177712724266e-05,
+      "loss": 0.413,
+      "step": 16605
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846149006678242e-05,
+      "loss": 0.4185,
+      "step": 16606
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846120297974693e-05,
+      "loss": 0.5002,
+      "step": 16607
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984609158661363e-05,
+      "loss": 0.4441,
+      "step": 16608
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9846062872595066e-05,
+      "loss": 0.4842,
+      "step": 16609
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9846034155918995e-05,
+      "loss": 0.5423,
+      "step": 16610
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984600543658544e-05,
+      "loss": 0.5148,
+      "step": 16611
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.98459767145944e-05,
+      "loss": 0.443,
+      "step": 16612
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845947989945885e-05,
+      "loss": 0.6186,
+      "step": 16613
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.98459192626399e-05,
+      "loss": 0.466,
+      "step": 16614
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984589053267646e-05,
+      "loss": 0.4885,
+      "step": 16615
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845861800055567e-05,
+      "loss": 0.422,
+      "step": 16616
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845833064777227e-05,
+      "loss": 0.4907,
+      "step": 16617
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845804326841452e-05,
+      "loss": 0.3926,
+      "step": 16618
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984577558624825e-05,
+      "loss": 0.5127,
+      "step": 16619
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845746842997626e-05,
+      "loss": 0.5233,
+      "step": 16620
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984571809708959e-05,
+      "loss": 0.5011,
+      "step": 16621
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845689348524146e-05,
+      "loss": 0.5594,
+      "step": 16622
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845660597301308e-05,
+      "loss": 0.5362,
+      "step": 16623
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845631843421077e-05,
+      "loss": 0.4524,
+      "step": 16624
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845603086883466e-05,
+      "loss": 0.587,
+      "step": 16625
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984557432768848e-05,
+      "loss": 0.5324,
+      "step": 16626
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845545565836127e-05,
+      "loss": 0.6077,
+      "step": 16627
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845516801326418e-05,
+      "loss": 0.5094,
+      "step": 16628
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845488034159354e-05,
+      "loss": 0.4238,
+      "step": 16629
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845459264334947e-05,
+      "loss": 0.5163,
+      "step": 16630
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845430491853205e-05,
+      "loss": 0.5664,
+      "step": 16631
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845401716714137e-05,
+      "loss": 0.5241,
+      "step": 16632
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845372938917748e-05,
+      "loss": 0.4803,
+      "step": 16633
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845344158464047e-05,
+      "loss": 0.4783,
+      "step": 16634
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984531537535304e-05,
+      "loss": 0.5543,
+      "step": 16635
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984528658958474e-05,
+      "loss": 0.4598,
+      "step": 16636
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845257801159147e-05,
+      "loss": 0.5387,
+      "step": 16637
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845229010076277e-05,
+      "loss": 0.4557,
+      "step": 16638
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845200216336132e-05,
+      "loss": 0.4757,
+      "step": 16639
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984517141993872e-05,
+      "loss": 0.5161,
+      "step": 16640
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845142620884053e-05,
+      "loss": 0.5332,
+      "step": 16641
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845113819172133e-05,
+      "loss": 0.4555,
+      "step": 16642
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984508501480297e-05,
+      "loss": 0.4948,
+      "step": 16643
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845056207776576e-05,
+      "loss": 0.5201,
+      "step": 16644
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9845027398092954e-05,
+      "loss": 0.429,
+      "step": 16645
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844998585752114e-05,
+      "loss": 0.5704,
+      "step": 16646
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844969770754062e-05,
+      "loss": 0.4578,
+      "step": 16647
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844940953098806e-05,
+      "loss": 0.5541,
+      "step": 16648
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844912132786356e-05,
+      "loss": 0.4542,
+      "step": 16649
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844883309816717e-05,
+      "loss": 0.4218,
+      "step": 16650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844854484189895e-05,
+      "loss": 0.4438,
+      "step": 16651
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844825655905907e-05,
+      "loss": 0.5216,
+      "step": 16652
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984479682496475e-05,
+      "loss": 0.4382,
+      "step": 16653
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984476799136644e-05,
+      "loss": 0.5089,
+      "step": 16654
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844739155110977e-05,
+      "loss": 0.5665,
+      "step": 16655
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844710316198376e-05,
+      "loss": 0.4926,
+      "step": 16656
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984468147462864e-05,
+      "loss": 0.4334,
+      "step": 16657
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844652630401776e-05,
+      "loss": 0.5202,
+      "step": 16658
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844623783517798e-05,
+      "loss": 0.4597,
+      "step": 16659
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844594933976712e-05,
+      "loss": 0.4368,
+      "step": 16660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844566081778517e-05,
+      "loss": 0.5175,
+      "step": 16661
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844537226923234e-05,
+      "loss": 0.4899,
+      "step": 16662
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844508369410863e-05,
+      "loss": 0.5299,
+      "step": 16663
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844479509241407e-05,
+      "loss": 0.5315,
+      "step": 16664
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844450646414887e-05,
+      "loss": 0.4463,
+      "step": 16665
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844421780931302e-05,
+      "loss": 0.5304,
+      "step": 16666
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984439291279066e-05,
+      "loss": 0.423,
+      "step": 16667
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844364041992974e-05,
+      "loss": 0.5109,
+      "step": 16668
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844335168538243e-05,
+      "loss": 0.5407,
+      "step": 16669
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844306292426486e-05,
+      "loss": 0.4808,
+      "step": 16670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.98442774136577e-05,
+      "loss": 0.4778,
+      "step": 16671
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.98442485322319e-05,
+      "loss": 0.408,
+      "step": 16672
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844219648149087e-05,
+      "loss": 0.4863,
+      "step": 16673
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844190761409278e-05,
+      "loss": 0.4704,
+      "step": 16674
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844161872012476e-05,
+      "loss": 0.5437,
+      "step": 16675
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844132979958684e-05,
+      "loss": 0.3788,
+      "step": 16676
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984410408524792e-05,
+      "loss": 0.4394,
+      "step": 16677
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844075187880182e-05,
+      "loss": 0.548,
+      "step": 16678
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9844046287855485e-05,
+      "loss": 0.5465,
+      "step": 16679
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984401738517383e-05,
+      "loss": 0.552,
+      "step": 16680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843988479835232e-05,
+      "loss": 0.479,
+      "step": 16681
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843959571839694e-05,
+      "loss": 0.4097,
+      "step": 16682
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843930661187227e-05,
+      "loss": 0.5067,
+      "step": 16683
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843901747877835e-05,
+      "loss": 0.5216,
+      "step": 16684
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984387283191153e-05,
+      "loss": 0.5455,
+      "step": 16685
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843843913288317e-05,
+      "loss": 0.5155,
+      "step": 16686
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843814992008202e-05,
+      "loss": 0.4286,
+      "step": 16687
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.98437860680712e-05,
+      "loss": 0.4891,
+      "step": 16688
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843757141477315e-05,
+      "loss": 0.4446,
+      "step": 16689
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984372821222655e-05,
+      "loss": 0.5715,
+      "step": 16690
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843699280318915e-05,
+      "loss": 0.528,
+      "step": 16691
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843670345754423e-05,
+      "loss": 0.4828,
+      "step": 16692
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843641408533077e-05,
+      "loss": 0.4836,
+      "step": 16693
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843612468654883e-05,
+      "loss": 0.4714,
+      "step": 16694
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984358352611986e-05,
+      "loss": 0.6004,
+      "step": 16695
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843554580928e-05,
+      "loss": 0.4679,
+      "step": 16696
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843525633079322e-05,
+      "loss": 0.4219,
+      "step": 16697
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843496682573833e-05,
+      "loss": 0.4957,
+      "step": 16698
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843467729411533e-05,
+      "loss": 0.4392,
+      "step": 16699
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984343877359244e-05,
+      "loss": 0.5009,
+      "step": 16700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984340981511655e-05,
+      "loss": 0.4713,
+      "step": 16701
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.984338085398389e-05,
+      "loss": 0.5103,
+      "step": 16702
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843351890194446e-05,
+      "loss": 0.5416,
+      "step": 16703
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843322923748234e-05,
+      "loss": 0.4106,
+      "step": 16704
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843293954645266e-05,
+      "loss": 0.4569,
+      "step": 16705
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843264982885548e-05,
+      "loss": 0.5208,
+      "step": 16706
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9843236008469088e-05,
+      "loss": 0.5798,
+      "step": 16707
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984320703139589e-05,
+      "loss": 0.5102,
+      "step": 16708
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984317805166596e-05,
+      "loss": 0.4591,
+      "step": 16709
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9843149069279317e-05,
+      "loss": 0.5445,
+      "step": 16710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984312008423596e-05,
+      "loss": 0.5427,
+      "step": 16711
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.98430910965359e-05,
+      "loss": 0.4967,
+      "step": 16712
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984306210617914e-05,
+      "loss": 0.4544,
+      "step": 16713
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.98430331131657e-05,
+      "loss": 0.4562,
+      "step": 16714
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984300411749557e-05,
+      "loss": 0.4796,
+      "step": 16715
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842975119168772e-05,
+      "loss": 0.491,
+      "step": 16716
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842946118185306e-05,
+      "loss": 0.4369,
+      "step": 16717
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842917114545183e-05,
+      "loss": 0.4067,
+      "step": 16718
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842888108248415e-05,
+      "loss": 0.4581,
+      "step": 16719
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842859099295003e-05,
+      "loss": 0.5006,
+      "step": 16720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842830087684954e-05,
+      "loss": 0.5024,
+      "step": 16721
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842801073418287e-05,
+      "loss": 0.4395,
+      "step": 16722
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842772056494993e-05,
+      "loss": 0.4886,
+      "step": 16723
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842743036915095e-05,
+      "loss": 0.5018,
+      "step": 16724
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842714014678594e-05,
+      "loss": 0.4357,
+      "step": 16725
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842684989785495e-05,
+      "loss": 0.5429,
+      "step": 16726
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984265596223581e-05,
+      "loss": 0.4552,
+      "step": 16727
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984262693202955e-05,
+      "loss": 0.4857,
+      "step": 16728
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842597899166714e-05,
+      "loss": 0.4611,
+      "step": 16729
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842568863647317e-05,
+      "loss": 0.4006,
+      "step": 16730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842539825471366e-05,
+      "loss": 0.5287,
+      "step": 16731
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842510784638866e-05,
+      "loss": 0.4427,
+      "step": 16732
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842481741149827e-05,
+      "loss": 0.5099,
+      "step": 16733
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842452695004256e-05,
+      "loss": 0.4255,
+      "step": 16734
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842423646202162e-05,
+      "loss": 0.498,
+      "step": 16735
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984239459474355e-05,
+      "loss": 0.522,
+      "step": 16736
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984236554062843e-05,
+      "loss": 0.4634,
+      "step": 16737
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984233648385681e-05,
+      "loss": 0.477,
+      "step": 16738
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842307424428698e-05,
+      "loss": 0.5677,
+      "step": 16739
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.98422783623441e-05,
+      "loss": 0.4653,
+      "step": 16740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842249297603026e-05,
+      "loss": 0.436,
+      "step": 16741
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842220230205482e-05,
+      "loss": 0.5432,
+      "step": 16742
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842191160151477e-05,
+      "loss": 0.4934,
+      "step": 16743
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984216208744102e-05,
+      "loss": 0.4275,
+      "step": 16744
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842133012074117e-05,
+      "loss": 0.4179,
+      "step": 16745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842103934050776e-05,
+      "loss": 0.4493,
+      "step": 16746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842074853371007e-05,
+      "loss": 0.4527,
+      "step": 16747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842045770034814e-05,
+      "loss": 0.4935,
+      "step": 16748
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9842016684042203e-05,
+      "loss": 0.5805,
+      "step": 16749
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984198759539319e-05,
+      "loss": 0.5065,
+      "step": 16750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841958504087778e-05,
+      "loss": 0.4743,
+      "step": 16751
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841929410125974e-05,
+      "loss": 0.4625,
+      "step": 16752
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984190031350779e-05,
+      "loss": 0.4648,
+      "step": 16753
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984187121423323e-05,
+      "loss": 0.4487,
+      "step": 16754
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841842112302303e-05,
+      "loss": 0.5554,
+      "step": 16755
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841813007715016e-05,
+      "loss": 0.392,
+      "step": 16756
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984178390047138e-05,
+      "loss": 0.5286,
+      "step": 16757
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841754790571396e-05,
+      "loss": 0.4977,
+      "step": 16758
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984172567801508e-05,
+      "loss": 0.4936,
+      "step": 16759
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841696562802435e-05,
+      "loss": 0.568,
+      "step": 16760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984166744493347e-05,
+      "loss": 0.4593,
+      "step": 16761
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984163832440819e-05,
+      "loss": 0.5332,
+      "step": 16762
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841609201226613e-05,
+      "loss": 0.5254,
+      "step": 16763
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841580075388736e-05,
+      "loss": 0.5332,
+      "step": 16764
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984155094689457e-05,
+      "loss": 0.3897,
+      "step": 16765
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841521815744123e-05,
+      "loss": 0.5444,
+      "step": 16766
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841492681937404e-05,
+      "loss": 0.5114,
+      "step": 16767
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984146354547442e-05,
+      "loss": 0.4446,
+      "step": 16768
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984143440635518e-05,
+      "loss": 0.4185,
+      "step": 16769
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984140526457969e-05,
+      "loss": 0.4759,
+      "step": 16770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841376120147958e-05,
+      "loss": 0.4837,
+      "step": 16771
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841346973059993e-05,
+      "loss": 0.6396,
+      "step": 16772
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841317823315804e-05,
+      "loss": 0.5011,
+      "step": 16773
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841288670915393e-05,
+      "loss": 0.4991,
+      "step": 16774
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841259515858776e-05,
+      "loss": 0.4154,
+      "step": 16775
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841230358145958e-05,
+      "loss": 0.5163,
+      "step": 16776
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841201197776943e-05,
+      "loss": 0.5688,
+      "step": 16777
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841172034751745e-05,
+      "loss": 0.4324,
+      "step": 16778
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841142869070363e-05,
+      "loss": 0.5584,
+      "step": 16779
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841113700732815e-05,
+      "loss": 0.4922,
+      "step": 16780
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841084529739106e-05,
+      "loss": 0.4954,
+      "step": 16781
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841055356089238e-05,
+      "loss": 0.4905,
+      "step": 16782
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9841026179783224e-05,
+      "loss": 0.4723,
+      "step": 16783
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984099700082107e-05,
+      "loss": 0.4675,
+      "step": 16784
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984096781920279e-05,
+      "loss": 0.4938,
+      "step": 16785
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840938634928385e-05,
+      "loss": 0.3735,
+      "step": 16786
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840909447997863e-05,
+      "loss": 0.4573,
+      "step": 16787
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840880258411233e-05,
+      "loss": 0.3985,
+      "step": 16788
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840851066168504e-05,
+      "loss": 0.5033,
+      "step": 16789
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840821871269683e-05,
+      "loss": 0.5255,
+      "step": 16790
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984079267371478e-05,
+      "loss": 0.4379,
+      "step": 16791
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840763473503798e-05,
+      "loss": 0.4836,
+      "step": 16792
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984073427063675e-05,
+      "loss": 0.4632,
+      "step": 16793
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984070506511364e-05,
+      "loss": 0.4733,
+      "step": 16794
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984067585693448e-05,
+      "loss": 0.4179,
+      "step": 16795
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840646646099277e-05,
+      "loss": 0.4409,
+      "step": 16796
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840617432608035e-05,
+      "loss": 0.4734,
+      "step": 16797
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840588216460762e-05,
+      "loss": 0.5048,
+      "step": 16798
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984055899765747e-05,
+      "loss": 0.3971,
+      "step": 16799
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984052977619817e-05,
+      "loss": 0.4363,
+      "step": 16800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984050055208286e-05,
+      "loss": 0.4307,
+      "step": 16801
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840471325311553e-05,
+      "loss": 0.4644,
+      "step": 16802
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840442095884257e-05,
+      "loss": 0.4714,
+      "step": 16803
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840412863800984e-05,
+      "loss": 0.4709,
+      "step": 16804
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.984038362906173e-05,
+      "loss": 0.456,
+      "step": 16805
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9840354391666515e-05,
+      "loss": 0.5275,
+      "step": 16806
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9840325151615346e-05,
+      "loss": 0.5831,
+      "step": 16807
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9840295908908223e-05,
+      "loss": 0.4762,
+      "step": 16808
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9840266663545157e-05,
+      "loss": 0.4315,
+      "step": 16809
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.984023741552616e-05,
+      "loss": 0.466,
+      "step": 16810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9840208164851235e-05,
+      "loss": 0.435,
+      "step": 16811
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.984017891152039e-05,
+      "loss": 0.5128,
+      "step": 16812
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.984014965553364e-05,
+      "loss": 0.5401,
+      "step": 16813
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.984012039689098e-05,
+      "loss": 0.4683,
+      "step": 16814
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.984009113559243e-05,
+      "loss": 0.5662,
+      "step": 16815
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9840061871637994e-05,
+      "loss": 0.4492,
+      "step": 16816
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.984003260502768e-05,
+      "loss": 0.4411,
+      "step": 16817
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.984000333576149e-05,
+      "loss": 0.4606,
+      "step": 16818
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839974063839445e-05,
+      "loss": 0.4429,
+      "step": 16819
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983994478926154e-05,
+      "loss": 0.4394,
+      "step": 16820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983991551202779e-05,
+      "loss": 0.4836,
+      "step": 16821
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839886232138196e-05,
+      "loss": 0.4572,
+      "step": 16822
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839856949592777e-05,
+      "loss": 0.4807,
+      "step": 16823
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983982766439153e-05,
+      "loss": 0.5004,
+      "step": 16824
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983979837653447e-05,
+      "loss": 0.4995,
+      "step": 16825
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.98397690860216e-05,
+      "loss": 0.5961,
+      "step": 16826
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983973979285293e-05,
+      "loss": 0.4723,
+      "step": 16827
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839710497028474e-05,
+      "loss": 0.5248,
+      "step": 16828
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983968119854823e-05,
+      "loss": 0.5694,
+      "step": 16829
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839651897412212e-05,
+      "loss": 0.5341,
+      "step": 16830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983962259362042e-05,
+      "loss": 0.5058,
+      "step": 16831
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839593287172877e-05,
+      "loss": 0.5156,
+      "step": 16832
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839563978069575e-05,
+      "loss": 0.4507,
+      "step": 16833
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983953466631053e-05,
+      "loss": 0.4585,
+      "step": 16834
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983950535189575e-05,
+      "loss": 0.4726,
+      "step": 16835
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983947603482524e-05,
+      "loss": 0.4725,
+      "step": 16836
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983944671509901e-05,
+      "loss": 0.4158,
+      "step": 16837
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983941739271707e-05,
+      "loss": 0.4275,
+      "step": 16838
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839388067679422e-05,
+      "loss": 0.4838,
+      "step": 16839
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839358739986082e-05,
+      "loss": 0.5226,
+      "step": 16840
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839329409637047e-05,
+      "loss": 0.4869,
+      "step": 16841
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839300076632333e-05,
+      "loss": 0.6014,
+      "step": 16842
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839270740971947e-05,
+      "loss": 0.4725,
+      "step": 16843
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839241402655895e-05,
+      "loss": 0.4925,
+      "step": 16844
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839212061684186e-05,
+      "loss": 0.5397,
+      "step": 16845
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839182718056825e-05,
+      "loss": 0.4448,
+      "step": 16846
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839153371773826e-05,
+      "loss": 0.4712,
+      "step": 16847
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839124022835192e-05,
+      "loss": 0.4671,
+      "step": 16848
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839094671240934e-05,
+      "loss": 0.4351,
+      "step": 16849
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839065316991054e-05,
+      "loss": 0.4507,
+      "step": 16850
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983903596008557e-05,
+      "loss": 0.4099,
+      "step": 16851
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9839006600524483e-05,
+      "loss": 0.4807,
+      "step": 16852
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838977238307798e-05,
+      "loss": 0.4932,
+      "step": 16853
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838947873435533e-05,
+      "loss": 0.4463,
+      "step": 16854
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838918505907683e-05,
+      "loss": 0.4863,
+      "step": 16855
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983888913572427e-05,
+      "loss": 0.5771,
+      "step": 16856
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983885976288529e-05,
+      "loss": 0.4906,
+      "step": 16857
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838830387390758e-05,
+      "loss": 0.4465,
+      "step": 16858
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983880100924068e-05,
+      "loss": 0.4266,
+      "step": 16859
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838771628435063e-05,
+      "loss": 0.4571,
+      "step": 16860
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838742244973914e-05,
+      "loss": 0.4413,
+      "step": 16861
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838712858857242e-05,
+      "loss": 0.5247,
+      "step": 16862
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838683470085058e-05,
+      "loss": 0.4459,
+      "step": 16863
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838654078657368e-05,
+      "loss": 0.4331,
+      "step": 16864
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983862468457418e-05,
+      "loss": 0.4374,
+      "step": 16865
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838595287835497e-05,
+      "loss": 0.4788,
+      "step": 16866
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838565888441333e-05,
+      "loss": 0.5219,
+      "step": 16867
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838536486391694e-05,
+      "loss": 0.611,
+      "step": 16868
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983850708168659e-05,
+      "loss": 0.4494,
+      "step": 16869
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838477674326023e-05,
+      "loss": 0.4978,
+      "step": 16870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838448264310005e-05,
+      "loss": 0.4416,
+      "step": 16871
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983841885163855e-05,
+      "loss": 0.4382,
+      "step": 16872
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838389436311655e-05,
+      "loss": 0.4519,
+      "step": 16873
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838360018329332e-05,
+      "loss": 0.4894,
+      "step": 16874
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983833059769159e-05,
+      "loss": 0.4861,
+      "step": 16875
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983830117439844e-05,
+      "loss": 0.447,
+      "step": 16876
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838271748449883e-05,
+      "loss": 0.4106,
+      "step": 16877
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983824231984593e-05,
+      "loss": 0.5488,
+      "step": 16878
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838212888586593e-05,
+      "loss": 0.4721,
+      "step": 16879
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838183454671873e-05,
+      "loss": 0.4564,
+      "step": 16880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838154018101782e-05,
+      "loss": 0.5131,
+      "step": 16881
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838124578876327e-05,
+      "loss": 0.4442,
+      "step": 16882
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983809513699552e-05,
+      "loss": 0.4319,
+      "step": 16883
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983806569245936e-05,
+      "loss": 0.5087,
+      "step": 16884
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838036245267864e-05,
+      "loss": 0.4826,
+      "step": 16885
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9838006795421033e-05,
+      "loss": 0.534,
+      "step": 16886
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983797734291888e-05,
+      "loss": 0.5357,
+      "step": 16887
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9837947887761412e-05,
+      "loss": 0.4922,
+      "step": 16888
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983791842994863e-05,
+      "loss": 0.4926,
+      "step": 16889
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983788896948055e-05,
+      "loss": 0.4624,
+      "step": 16890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9837859506357183e-05,
+      "loss": 0.4902,
+      "step": 16891
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9837830040578527e-05,
+      "loss": 0.4127,
+      "step": 16892
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9837800572144598e-05,
+      "loss": 0.576,
+      "step": 16893
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9837771101055397e-05,
+      "loss": 0.4359,
+      "step": 16894
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983774162731094e-05,
+      "loss": 0.458,
+      "step": 16895
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983771215091123e-05,
+      "loss": 0.4282,
+      "step": 16896
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983768267185627e-05,
+      "loss": 0.5209,
+      "step": 16897
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983765319014608e-05,
+      "loss": 0.5013,
+      "step": 16898
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9837623705780656e-05,
+      "loss": 0.5177,
+      "step": 16899
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9837594218760013e-05,
+      "loss": 0.3848,
+      "step": 16900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9837564729084158e-05,
+      "loss": 0.4594,
+      "step": 16901
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.98375352367531e-05,
+      "loss": 0.4834,
+      "step": 16902
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.983750574176684e-05,
+      "loss": 0.5155,
+      "step": 16903
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9837476244125397e-05,
+      "loss": 0.4533,
+      "step": 16904
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983744674382877e-05,
+      "loss": 0.4609,
+      "step": 16905
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983741724087697e-05,
+      "loss": 0.5271,
+      "step": 16906
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9837387735270005e-05,
+      "loss": 0.5934,
+      "step": 16907
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9837358227007885e-05,
+      "loss": 0.5234,
+      "step": 16908
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983732871609062e-05,
+      "loss": 0.4449,
+      "step": 16909
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9837299202518206e-05,
+      "loss": 0.4729,
+      "step": 16910
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9837269686290664e-05,
+      "loss": 0.4334,
+      "step": 16911
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983724016740799e-05,
+      "loss": 0.4971,
+      "step": 16912
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983721064587021e-05,
+      "loss": 0.4709,
+      "step": 16913
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983718112167731e-05,
+      "loss": 0.4809,
+      "step": 16914
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9837151594829317e-05,
+      "loss": 0.4925,
+      "step": 16915
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9837122065326227e-05,
+      "loss": 0.5533,
+      "step": 16916
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983709253316805e-05,
+      "loss": 0.49,
+      "step": 16917
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.98370629983548e-05,
+      "loss": 0.483,
+      "step": 16918
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983703346088648e-05,
+      "loss": 0.5556,
+      "step": 16919
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9837003920763094e-05,
+      "loss": 0.4551,
+      "step": 16920
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983697437798466e-05,
+      "loss": 0.4232,
+      "step": 16921
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836944832551178e-05,
+      "loss": 0.5448,
+      "step": 16922
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983691528446266e-05,
+      "loss": 0.492,
+      "step": 16923
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983688573371911e-05,
+      "loss": 0.5343,
+      "step": 16924
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983685618032054e-05,
+      "loss": 0.4724,
+      "step": 16925
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983682662426696e-05,
+      "loss": 0.3977,
+      "step": 16926
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983679706555837e-05,
+      "loss": 0.4868,
+      "step": 16927
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836767504194782e-05,
+      "loss": 0.5008,
+      "step": 16928
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836737940176208e-05,
+      "loss": 0.5339,
+      "step": 16929
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836708373502652e-05,
+      "loss": 0.4814,
+      "step": 16930
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983667880417412e-05,
+      "loss": 0.5275,
+      "step": 16931
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836649232190623e-05,
+      "loss": 0.4541,
+      "step": 16932
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983661965755217e-05,
+      "loss": 0.4496,
+      "step": 16933
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983659008025877e-05,
+      "loss": 0.6039,
+      "step": 16934
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836560500310422e-05,
+      "loss": 0.468,
+      "step": 16935
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983653091770714e-05,
+      "loss": 0.479,
+      "step": 16936
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836501332448938e-05,
+      "loss": 0.5804,
+      "step": 16937
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836471744535818e-05,
+      "loss": 0.4377,
+      "step": 16938
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836442153967785e-05,
+      "loss": 0.4618,
+      "step": 16939
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836412560744852e-05,
+      "loss": 0.4383,
+      "step": 16940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836382964867024e-05,
+      "loss": 0.4779,
+      "step": 16941
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983635336633431e-05,
+      "loss": 0.4785,
+      "step": 16942
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983632376514672e-05,
+      "loss": 0.5699,
+      "step": 16943
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983629416130426e-05,
+      "loss": 0.448,
+      "step": 16944
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836264554806936e-05,
+      "loss": 0.4746,
+      "step": 16945
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983623494565476e-05,
+      "loss": 0.4683,
+      "step": 16946
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983620533384774e-05,
+      "loss": 0.4795,
+      "step": 16947
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836175719385878e-05,
+      "loss": 0.5112,
+      "step": 16948
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983614610226919e-05,
+      "loss": 0.4162,
+      "step": 16949
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836116482497677e-05,
+      "loss": 0.5057,
+      "step": 16950
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836086860071352e-05,
+      "loss": 0.4396,
+      "step": 16951
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836057234990223e-05,
+      "loss": 0.4504,
+      "step": 16952
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9836027607254293e-05,
+      "loss": 0.4548,
+      "step": 16953
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835997976863575e-05,
+      "loss": 0.4707,
+      "step": 16954
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835968343818073e-05,
+      "loss": 0.4764,
+      "step": 16955
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.98359387081178e-05,
+      "loss": 0.4869,
+      "step": 16956
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983590906976276e-05,
+      "loss": 0.4828,
+      "step": 16957
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835879428752964e-05,
+      "loss": 0.4734,
+      "step": 16958
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835849785088413e-05,
+      "loss": 0.3817,
+      "step": 16959
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835820138769125e-05,
+      "loss": 0.4577,
+      "step": 16960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.98357904897951e-05,
+      "loss": 0.4564,
+      "step": 16961
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983576083816635e-05,
+      "loss": 0.3977,
+      "step": 16962
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835731183882886e-05,
+      "loss": 0.5106,
+      "step": 16963
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983570152694471e-05,
+      "loss": 0.4131,
+      "step": 16964
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983567186735183e-05,
+      "loss": 0.5666,
+      "step": 16965
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835642205104256e-05,
+      "loss": 0.4868,
+      "step": 16966
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835612540201997e-05,
+      "loss": 0.5296,
+      "step": 16967
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983558287264506e-05,
+      "loss": 0.4993,
+      "step": 16968
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835553202433456e-05,
+      "loss": 0.4116,
+      "step": 16969
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835523529567187e-05,
+      "loss": 0.5046,
+      "step": 16970
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835493854046266e-05,
+      "loss": 0.3954,
+      "step": 16971
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.98354641758707e-05,
+      "loss": 0.4163,
+      "step": 16972
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835434495040494e-05,
+      "loss": 0.4622,
+      "step": 16973
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835404811555657e-05,
+      "loss": 0.5384,
+      "step": 16974
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835375125416202e-05,
+      "loss": 0.465,
+      "step": 16975
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835345436622133e-05,
+      "loss": 0.3992,
+      "step": 16976
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835315745173455e-05,
+      "loss": 0.5129,
+      "step": 16977
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835286051070183e-05,
+      "loss": 0.4027,
+      "step": 16978
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983525635431232e-05,
+      "loss": 0.4515,
+      "step": 16979
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835226654899875e-05,
+      "loss": 0.533,
+      "step": 16980
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835196952832856e-05,
+      "loss": 0.454,
+      "step": 16981
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983516724811127e-05,
+      "loss": 0.5044,
+      "step": 16982
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983513754073513e-05,
+      "loss": 0.4332,
+      "step": 16983
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835107830704437e-05,
+      "loss": 0.445,
+      "step": 16984
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835078118019204e-05,
+      "loss": 0.4691,
+      "step": 16985
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835048402679436e-05,
+      "loss": 0.463,
+      "step": 16986
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9835018684685146e-05,
+      "loss": 0.4601,
+      "step": 16987
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834988964036335e-05,
+      "loss": 0.42,
+      "step": 16988
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834959240733015e-05,
+      "loss": 0.4268,
+      "step": 16989
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834929514775195e-05,
+      "loss": 0.506,
+      "step": 16990
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834899786162882e-05,
+      "loss": 0.379,
+      "step": 16991
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834870054896083e-05,
+      "loss": 0.541,
+      "step": 16992
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834840320974805e-05,
+      "loss": 0.5422,
+      "step": 16993
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834810584399058e-05,
+      "loss": 0.4245,
+      "step": 16994
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834780845168852e-05,
+      "loss": 0.5341,
+      "step": 16995
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983475110328419e-05,
+      "loss": 0.5295,
+      "step": 16996
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834721358745083e-05,
+      "loss": 0.5197,
+      "step": 16997
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.983469161155154e-05,
+      "loss": 0.4512,
+      "step": 16998
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834661861703566e-05,
+      "loss": 0.5149,
+      "step": 16999
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834632109201174e-05,
+      "loss": 0.4309,
+      "step": 17000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834602354044366e-05,
+      "loss": 0.5231,
+      "step": 17001
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834572596233154e-05,
+      "loss": 0.5091,
+      "step": 17002
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9834542835767544e-05,
+      "loss": 0.4197,
+      "step": 17003
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834513072647546e-05,
+      "loss": 0.4554,
+      "step": 17004
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834483306873164e-05,
+      "loss": 0.5221,
+      "step": 17005
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834453538444414e-05,
+      "loss": 0.4792,
+      "step": 17006
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834423767361297e-05,
+      "loss": 0.5031,
+      "step": 17007
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834393993623822e-05,
+      "loss": 0.4571,
+      "step": 17008
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834364217231997e-05,
+      "loss": 0.5113,
+      "step": 17009
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834334438185835e-05,
+      "loss": 0.4929,
+      "step": 17010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834304656485335e-05,
+      "loss": 0.4817,
+      "step": 17011
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834274872130513e-05,
+      "loss": 0.5332,
+      "step": 17012
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834245085121374e-05,
+      "loss": 0.3894,
+      "step": 17013
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834215295457928e-05,
+      "loss": 0.4618,
+      "step": 17014
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983418550314018e-05,
+      "loss": 0.5182,
+      "step": 17015
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983415570816814e-05,
+      "loss": 0.4869,
+      "step": 17016
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834125910541813e-05,
+      "loss": 0.476,
+      "step": 17017
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834096110261212e-05,
+      "loss": 0.449,
+      "step": 17018
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9834066307326343e-05,
+      "loss": 0.4364,
+      "step": 17019
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983403650173721e-05,
+      "loss": 0.483,
+      "step": 17020
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983400669349383e-05,
+      "loss": 0.4513,
+      "step": 17021
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.98339768825962e-05,
+      "loss": 0.4969,
+      "step": 17022
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833947069044334e-05,
+      "loss": 0.5065,
+      "step": 17023
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833917252838244e-05,
+      "loss": 0.5312,
+      "step": 17024
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983388743397793e-05,
+      "loss": 0.4945,
+      "step": 17025
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833857612463404e-05,
+      "loss": 0.56,
+      "step": 17026
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983382778829468e-05,
+      "loss": 0.5806,
+      "step": 17027
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833797961471752e-05,
+      "loss": 0.4409,
+      "step": 17028
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983376813199464e-05,
+      "loss": 0.5364,
+      "step": 17029
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833738299863348e-05,
+      "loss": 0.4593,
+      "step": 17030
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833708465077885e-05,
+      "loss": 0.5061,
+      "step": 17031
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833678627638254e-05,
+      "loss": 0.5231,
+      "step": 17032
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983364878754447e-05,
+      "loss": 0.4869,
+      "step": 17033
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833618944796536e-05,
+      "loss": 0.4305,
+      "step": 17034
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833589099394464e-05,
+      "loss": 0.4527,
+      "step": 17035
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983355925133826e-05,
+      "loss": 0.4158,
+      "step": 17036
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833529400627934e-05,
+      "loss": 0.5029,
+      "step": 17037
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833499547263487e-05,
+      "loss": 0.4148,
+      "step": 17038
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983346969124494e-05,
+      "loss": 0.4547,
+      "step": 17039
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983343983257229e-05,
+      "loss": 0.475,
+      "step": 17040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983340997124555e-05,
+      "loss": 0.4898,
+      "step": 17041
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833380107264723e-05,
+      "loss": 0.434,
+      "step": 17042
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833350240629825e-05,
+      "loss": 0.5302,
+      "step": 17043
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833320371340857e-05,
+      "loss": 0.4596,
+      "step": 17044
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983329049939783e-05,
+      "loss": 0.4585,
+      "step": 17045
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833260624800754e-05,
+      "loss": 0.4847,
+      "step": 17046
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833230747549633e-05,
+      "loss": 0.4736,
+      "step": 17047
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833200867644476e-05,
+      "loss": 0.5238,
+      "step": 17048
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833170985085294e-05,
+      "loss": 0.47,
+      "step": 17049
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833141099872097e-05,
+      "loss": 0.4447,
+      "step": 17050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833111212004884e-05,
+      "loss": 0.399,
+      "step": 17051
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983308132148367e-05,
+      "loss": 0.4783,
+      "step": 17052
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833051428308462e-05,
+      "loss": 0.494,
+      "step": 17053
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9833021532479265e-05,
+      "loss": 0.4025,
+      "step": 17054
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832991633996094e-05,
+      "loss": 0.4825,
+      "step": 17055
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983296173285895e-05,
+      "loss": 0.5115,
+      "step": 17056
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832931829067842e-05,
+      "loss": 0.5108,
+      "step": 17057
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983290192262278e-05,
+      "loss": 0.5283,
+      "step": 17058
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832872013523773e-05,
+      "loss": 0.4465,
+      "step": 17059
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832842101770828e-05,
+      "loss": 0.4767,
+      "step": 17060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832812187363953e-05,
+      "loss": 0.5351,
+      "step": 17061
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832782270303153e-05,
+      "loss": 0.4134,
+      "step": 17062
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832752350588444e-05,
+      "loss": 0.5411,
+      "step": 17063
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832722428219825e-05,
+      "loss": 0.4772,
+      "step": 17064
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832692503197312e-05,
+      "loss": 0.5256,
+      "step": 17065
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832662575520907e-05,
+      "loss": 0.4885,
+      "step": 17066
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983263264519062e-05,
+      "loss": 0.4238,
+      "step": 17067
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983260271220646e-05,
+      "loss": 0.4577,
+      "step": 17068
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832572776568436e-05,
+      "loss": 0.422,
+      "step": 17069
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832542838276553e-05,
+      "loss": 0.4806,
+      "step": 17070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832512897330824e-05,
+      "loss": 0.499,
+      "step": 17071
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983248295373125e-05,
+      "loss": 0.5143,
+      "step": 17072
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832453007477842e-05,
+      "loss": 0.5087,
+      "step": 17073
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832423058570614e-05,
+      "loss": 0.5051,
+      "step": 17074
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832393107009563e-05,
+      "loss": 0.579,
+      "step": 17075
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983236315279471e-05,
+      "loss": 0.5052,
+      "step": 17076
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832333195926053e-05,
+      "loss": 0.5096,
+      "step": 17077
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832303236403604e-05,
+      "loss": 0.4366,
+      "step": 17078
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832273274227367e-05,
+      "loss": 0.56,
+      "step": 17079
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832243309397355e-05,
+      "loss": 0.6081,
+      "step": 17080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832213341913575e-05,
+      "loss": 0.5457,
+      "step": 17081
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983218337177604e-05,
+      "loss": 0.4306,
+      "step": 17082
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832153398984744e-05,
+      "loss": 0.4326,
+      "step": 17083
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832123423539705e-05,
+      "loss": 0.528,
+      "step": 17084
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832093445440936e-05,
+      "loss": 0.5193,
+      "step": 17085
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832063464688433e-05,
+      "loss": 0.45,
+      "step": 17086
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832033481282213e-05,
+      "loss": 0.5302,
+      "step": 17087
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9832003495222282e-05,
+      "loss": 0.4543,
+      "step": 17088
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831973506508645e-05,
+      "loss": 0.517,
+      "step": 17089
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831943515141315e-05,
+      "loss": 0.4185,
+      "step": 17090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831913521120295e-05,
+      "loss": 0.4281,
+      "step": 17091
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831883524445595e-05,
+      "loss": 0.445,
+      "step": 17092
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831853525117226e-05,
+      "loss": 0.478,
+      "step": 17093
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831823523135194e-05,
+      "loss": 0.4989,
+      "step": 17094
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831793518499505e-05,
+      "loss": 0.491,
+      "step": 17095
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.983176351121017e-05,
+      "loss": 0.4455,
+      "step": 17096
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831733501267195e-05,
+      "loss": 0.459,
+      "step": 17097
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831703488670594e-05,
+      "loss": 0.4526,
+      "step": 17098
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831673473420363e-05,
+      "loss": 0.5104,
+      "step": 17099
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831643455516524e-05,
+      "loss": 0.4549,
+      "step": 17100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9831613434959073e-05,
+      "loss": 0.4731,
+      "step": 17101
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831583411748027e-05,
+      "loss": 0.4349,
+      "step": 17102
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983155338588339e-05,
+      "loss": 0.5267,
+      "step": 17103
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831523357365167e-05,
+      "loss": 0.4887,
+      "step": 17104
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831493326193373e-05,
+      "loss": 0.4699,
+      "step": 17105
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831463292368015e-05,
+      "loss": 0.4808,
+      "step": 17106
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831433255889095e-05,
+      "loss": 0.45,
+      "step": 17107
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983140321675663e-05,
+      "loss": 0.5855,
+      "step": 17108
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831373174970617e-05,
+      "loss": 0.456,
+      "step": 17109
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983134313053108e-05,
+      "loss": 0.5401,
+      "step": 17110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831313083438007e-05,
+      "loss": 0.4252,
+      "step": 17111
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983128303369142e-05,
+      "loss": 0.4537,
+      "step": 17112
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831252981291328e-05,
+      "loss": 0.4867,
+      "step": 17113
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831222926237732e-05,
+      "loss": 0.4164,
+      "step": 17114
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831192868530646e-05,
+      "loss": 0.5534,
+      "step": 17115
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983116280817007e-05,
+      "loss": 0.4983,
+      "step": 17116
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983113274515602e-05,
+      "loss": 0.5691,
+      "step": 17117
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831102679488498e-05,
+      "loss": 0.4285,
+      "step": 17118
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983107261116752e-05,
+      "loss": 0.5148,
+      "step": 17119
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9831042540193085e-05,
+      "loss": 0.5271,
+      "step": 17120
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983101246656521e-05,
+      "loss": 0.4724,
+      "step": 17121
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830982390283896e-05,
+      "loss": 0.4506,
+      "step": 17122
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830952311349154e-05,
+      "loss": 0.5347,
+      "step": 17123
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830922229760994e-05,
+      "loss": 0.4759,
+      "step": 17124
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830892145519422e-05,
+      "loss": 0.493,
+      "step": 17125
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830862058624444e-05,
+      "loss": 0.5984,
+      "step": 17126
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983083196907607e-05,
+      "loss": 0.545,
+      "step": 17127
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830801876874314e-05,
+      "loss": 0.4478,
+      "step": 17128
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983077178201917e-05,
+      "loss": 0.4915,
+      "step": 17129
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830741684510664e-05,
+      "loss": 0.4855,
+      "step": 17130
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983071158434879e-05,
+      "loss": 0.4649,
+      "step": 17131
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983068148153356e-05,
+      "loss": 0.45,
+      "step": 17132
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830651376064986e-05,
+      "loss": 0.5883,
+      "step": 17133
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983062126794307e-05,
+      "loss": 0.4536,
+      "step": 17134
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830591157167826e-05,
+      "loss": 0.4561,
+      "step": 17135
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983056104373926e-05,
+      "loss": 0.5221,
+      "step": 17136
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830530927657377e-05,
+      "loss": 0.4613,
+      "step": 17137
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983050080892219e-05,
+      "loss": 0.4846,
+      "step": 17138
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830470687533703e-05,
+      "loss": 0.4777,
+      "step": 17139
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830440563491926e-05,
+      "loss": 0.4624,
+      "step": 17140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983041043679687e-05,
+      "loss": 0.4988,
+      "step": 17141
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830380307448536e-05,
+      "loss": 0.5085,
+      "step": 17142
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983035017544694e-05,
+      "loss": 0.4941,
+      "step": 17143
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830320040792084e-05,
+      "loss": 0.5363,
+      "step": 17144
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983028990348398e-05,
+      "loss": 0.4638,
+      "step": 17145
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830259763522634e-05,
+      "loss": 0.5097,
+      "step": 17146
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830229620908054e-05,
+      "loss": 0.3934,
+      "step": 17147
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983019947564025e-05,
+      "loss": 0.4483,
+      "step": 17148
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983016932771923e-05,
+      "loss": 0.5724,
+      "step": 17149
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830139177145003e-05,
+      "loss": 0.4247,
+      "step": 17150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830109023917574e-05,
+      "loss": 0.4761,
+      "step": 17151
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830078868036954e-05,
+      "loss": 0.4955,
+      "step": 17152
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830048709503148e-05,
+      "loss": 0.3841,
+      "step": 17153
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9830018548316164e-05,
+      "loss": 0.4665,
+      "step": 17154
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829988384476016e-05,
+      "loss": 0.4793,
+      "step": 17155
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829958217982706e-05,
+      "loss": 0.5059,
+      "step": 17156
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982992804883624e-05,
+      "loss": 0.3989,
+      "step": 17157
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829897877036636e-05,
+      "loss": 0.4496,
+      "step": 17158
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829867702583897e-05,
+      "loss": 0.4055,
+      "step": 17159
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982983752547803e-05,
+      "loss": 0.5071,
+      "step": 17160
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829807345719042e-05,
+      "loss": 0.4806,
+      "step": 17161
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829777163306944e-05,
+      "loss": 0.4577,
+      "step": 17162
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829746978241746e-05,
+      "loss": 0.4494,
+      "step": 17163
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829716790523454e-05,
+      "loss": 0.4443,
+      "step": 17164
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982968660015207e-05,
+      "loss": 0.5217,
+      "step": 17165
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982965640712761e-05,
+      "loss": 0.4133,
+      "step": 17166
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829626211450078e-05,
+      "loss": 0.531,
+      "step": 17167
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829596013119486e-05,
+      "loss": 0.5097,
+      "step": 17168
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829565812135842e-05,
+      "loss": 0.4375,
+      "step": 17169
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829535608499148e-05,
+      "loss": 0.4227,
+      "step": 17170
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829505402209418e-05,
+      "loss": 0.3909,
+      "step": 17171
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982947519326666e-05,
+      "loss": 0.4465,
+      "step": 17172
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982944498167088e-05,
+      "loss": 0.4775,
+      "step": 17173
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829414767422088e-05,
+      "loss": 0.4612,
+      "step": 17174
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982938455052029e-05,
+      "loss": 0.4867,
+      "step": 17175
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829354330965494e-05,
+      "loss": 0.4063,
+      "step": 17176
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829324108757708e-05,
+      "loss": 0.4998,
+      "step": 17177
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829293883896943e-05,
+      "loss": 0.4543,
+      "step": 17178
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829263656383207e-05,
+      "loss": 0.4424,
+      "step": 17179
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829233426216506e-05,
+      "loss": 0.4816,
+      "step": 17180
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829203193396847e-05,
+      "loss": 0.4494,
+      "step": 17181
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829172957924244e-05,
+      "loss": 0.4799,
+      "step": 17182
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.98291427197987e-05,
+      "loss": 0.4937,
+      "step": 17183
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829112479020224e-05,
+      "loss": 0.4588,
+      "step": 17184
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829082235588825e-05,
+      "loss": 0.4589,
+      "step": 17185
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829051989504508e-05,
+      "loss": 0.4042,
+      "step": 17186
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9829021740767284e-05,
+      "loss": 0.4887,
+      "step": 17187
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9828991489377166e-05,
+      "loss": 0.4605,
+      "step": 17188
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9828961235334155e-05,
+      "loss": 0.4826,
+      "step": 17189
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982893097863826e-05,
+      "loss": 0.4708,
+      "step": 17190
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982890071928949e-05,
+      "loss": 0.5322,
+      "step": 17191
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9828870457287857e-05,
+      "loss": 0.5684,
+      "step": 17192
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9828840192633363e-05,
+      "loss": 0.4722,
+      "step": 17193
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982880992532602e-05,
+      "loss": 0.5577,
+      "step": 17194
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9828779655365835e-05,
+      "loss": 0.4999,
+      "step": 17195
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9828749382752816e-05,
+      "loss": 0.4621,
+      "step": 17196
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9828719107486974e-05,
+      "loss": 0.4266,
+      "step": 17197
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9828688829568313e-05,
+      "loss": 0.5288,
+      "step": 17198
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9828658548996843e-05,
+      "loss": 0.5207,
+      "step": 17199
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982862826577257e-05,
+      "loss": 0.4792,
+      "step": 17200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828597979895507e-05,
+      "loss": 0.4935,
+      "step": 17201
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982856769136566e-05,
+      "loss": 0.4706,
+      "step": 17202
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828537400183036e-05,
+      "loss": 0.5207,
+      "step": 17203
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828507106347644e-05,
+      "loss": 0.5517,
+      "step": 17204
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828476809859493e-05,
+      "loss": 0.4775,
+      "step": 17205
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828446510718584e-05,
+      "loss": 0.4579,
+      "step": 17206
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828416208924937e-05,
+      "loss": 0.4195,
+      "step": 17207
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828385904478553e-05,
+      "loss": 0.5068,
+      "step": 17208
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828355597379444e-05,
+      "loss": 0.4943,
+      "step": 17209
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828325287627614e-05,
+      "loss": 0.4961,
+      "step": 17210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982829497522307e-05,
+      "loss": 0.6051,
+      "step": 17211
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828264660165827e-05,
+      "loss": 0.5582,
+      "step": 17212
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982823434245589e-05,
+      "loss": 0.472,
+      "step": 17213
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828204022093267e-05,
+      "loss": 0.4237,
+      "step": 17214
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828173699077962e-05,
+      "loss": 0.4911,
+      "step": 17215
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982814337340999e-05,
+      "loss": 0.4271,
+      "step": 17216
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828113045089356e-05,
+      "loss": 0.4488,
+      "step": 17217
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828082714116064e-05,
+      "loss": 0.4584,
+      "step": 17218
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9828052380490133e-05,
+      "loss": 0.5353,
+      "step": 17219
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982802204421156e-05,
+      "loss": 0.4401,
+      "step": 17220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827991705280363e-05,
+      "loss": 0.4139,
+      "step": 17221
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982796136369654e-05,
+      "loss": 0.5095,
+      "step": 17222
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827931019460106e-05,
+      "loss": 0.4556,
+      "step": 17223
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982790067257107e-05,
+      "loss": 0.4205,
+      "step": 17224
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827870323029433e-05,
+      "loss": 0.4923,
+      "step": 17225
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982783997083521e-05,
+      "loss": 0.4342,
+      "step": 17226
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982780961598841e-05,
+      "loss": 0.4954,
+      "step": 17227
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827779258489034e-05,
+      "loss": 0.4846,
+      "step": 17228
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827748898337097e-05,
+      "loss": 0.4629,
+      "step": 17229
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827718535532602e-05,
+      "loss": 0.495,
+      "step": 17230
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982768817007556e-05,
+      "loss": 0.5206,
+      "step": 17231
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827657801965982e-05,
+      "loss": 0.498,
+      "step": 17232
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982762743120387e-05,
+      "loss": 0.4247,
+      "step": 17233
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827597057789238e-05,
+      "loss": 0.5609,
+      "step": 17234
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982756668172209e-05,
+      "loss": 0.5081,
+      "step": 17235
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827536303002436e-05,
+      "loss": 0.4615,
+      "step": 17236
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827505921630285e-05,
+      "loss": 0.4533,
+      "step": 17237
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827475537605646e-05,
+      "loss": 0.4757,
+      "step": 17238
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982744515092852e-05,
+      "loss": 0.4803,
+      "step": 17239
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827414761598925e-05,
+      "loss": 0.4681,
+      "step": 17240
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827384369616866e-05,
+      "loss": 0.4369,
+      "step": 17241
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827353974982343e-05,
+      "loss": 0.4852,
+      "step": 17242
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827323577695378e-05,
+      "loss": 0.4821,
+      "step": 17243
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982729317775597e-05,
+      "loss": 0.4219,
+      "step": 17244
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982726277516413e-05,
+      "loss": 0.4436,
+      "step": 17245
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827232369919867e-05,
+      "loss": 0.4419,
+      "step": 17246
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827201962023187e-05,
+      "loss": 0.4796,
+      "step": 17247
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827171551474098e-05,
+      "loss": 0.5171,
+      "step": 17248
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827141138272607e-05,
+      "loss": 0.4412,
+      "step": 17249
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982711072241873e-05,
+      "loss": 0.603,
+      "step": 17250
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827080303912468e-05,
+      "loss": 0.5252,
+      "step": 17251
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982704988275383e-05,
+      "loss": 0.4371,
+      "step": 17252
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9827019458942827e-05,
+      "loss": 0.3965,
+      "step": 17253
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826989032479463e-05,
+      "loss": 0.4516,
+      "step": 17254
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826958603363754e-05,
+      "loss": 0.5748,
+      "step": 17255
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.98269281715957e-05,
+      "loss": 0.5035,
+      "step": 17256
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982689773717531e-05,
+      "loss": 0.4295,
+      "step": 17257
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826867300102595e-05,
+      "loss": 0.457,
+      "step": 17258
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826836860377565e-05,
+      "loss": 0.4624,
+      "step": 17259
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826806418000223e-05,
+      "loss": 0.4023,
+      "step": 17260
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826775972970585e-05,
+      "loss": 0.4214,
+      "step": 17261
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982674552528865e-05,
+      "loss": 0.4913,
+      "step": 17262
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982671507495443e-05,
+      "loss": 0.4773,
+      "step": 17263
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826684621967932e-05,
+      "loss": 0.4991,
+      "step": 17264
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982665416632917e-05,
+      "loss": 0.4471,
+      "step": 17265
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826623708038148e-05,
+      "loss": 0.5345,
+      "step": 17266
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826593247094874e-05,
+      "loss": 0.5487,
+      "step": 17267
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826562783499354e-05,
+      "loss": 0.4577,
+      "step": 17268
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.98265323172516e-05,
+      "loss": 0.4266,
+      "step": 17269
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826501848351622e-05,
+      "loss": 0.508,
+      "step": 17270
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982647137679942e-05,
+      "loss": 0.387,
+      "step": 17271
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982644090259501e-05,
+      "loss": 0.482,
+      "step": 17272
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826410425738398e-05,
+      "loss": 0.4339,
+      "step": 17273
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826379946229592e-05,
+      "loss": 0.4096,
+      "step": 17274
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.98263494640686e-05,
+      "loss": 0.4367,
+      "step": 17275
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982631897925543e-05,
+      "loss": 0.5224,
+      "step": 17276
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982628849179009e-05,
+      "loss": 0.4655,
+      "step": 17277
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982625800167259e-05,
+      "loss": 0.5228,
+      "step": 17278
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826227508902938e-05,
+      "loss": 0.4686,
+      "step": 17279
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826197013481138e-05,
+      "loss": 0.4505,
+      "step": 17280
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826166515407202e-05,
+      "loss": 0.4133,
+      "step": 17281
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826136014681143e-05,
+      "loss": 0.4311,
+      "step": 17282
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826105511302958e-05,
+      "loss": 0.5335,
+      "step": 17283
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826075005272663e-05,
+      "loss": 0.5225,
+      "step": 17284
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9826044496590263e-05,
+      "loss": 0.496,
+      "step": 17285
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982601398525577e-05,
+      "loss": 0.4326,
+      "step": 17286
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982598347126919e-05,
+      "loss": 0.4028,
+      "step": 17287
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982595295463053e-05,
+      "loss": 0.4407,
+      "step": 17288
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.98259224353398e-05,
+      "loss": 0.5368,
+      "step": 17289
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9825891913397008e-05,
+      "loss": 0.4906,
+      "step": 17290
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982586138880216e-05,
+      "loss": 0.465,
+      "step": 17291
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9825830861555267e-05,
+      "loss": 0.481,
+      "step": 17292
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9825800331656336e-05,
+      "loss": 0.4866,
+      "step": 17293
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9825769799105378e-05,
+      "loss": 0.4872,
+      "step": 17294
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9825739263902394e-05,
+      "loss": 0.4169,
+      "step": 17295
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.98257087260474e-05,
+      "loss": 0.4842,
+      "step": 17296
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.98256781855404e-05,
+      "loss": 0.4646,
+      "step": 17297
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9825647642381406e-05,
+      "loss": 0.4952,
+      "step": 17298
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.982561709657042e-05,
+      "loss": 0.4822,
+      "step": 17299
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982558654810746e-05,
+      "loss": 0.465,
+      "step": 17300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9825555996992527e-05,
+      "loss": 0.4883,
+      "step": 17301
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9825525443225626e-05,
+      "loss": 0.6167,
+      "step": 17302
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982549488680677e-05,
+      "loss": 0.478,
+      "step": 17303
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982546432773597e-05,
+      "loss": 0.4564,
+      "step": 17304
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982543376601323e-05,
+      "loss": 0.5737,
+      "step": 17305
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982540320163856e-05,
+      "loss": 0.4194,
+      "step": 17306
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9825372634611966e-05,
+      "loss": 0.4557,
+      "step": 17307
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982534206493346e-05,
+      "loss": 0.4089,
+      "step": 17308
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982531149260305e-05,
+      "loss": 0.5071,
+      "step": 17309
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982528091762074e-05,
+      "loss": 0.5412,
+      "step": 17310
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982525033998654e-05,
+      "loss": 0.5151,
+      "step": 17311
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982521975970046e-05,
+      "loss": 0.4555,
+      "step": 17312
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982518917676251e-05,
+      "loss": 0.4291,
+      "step": 17313
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982515859117269e-05,
+      "loss": 0.5146,
+      "step": 17314
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9825128002931016e-05,
+      "loss": 0.4281,
+      "step": 17315
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9825097412037496e-05,
+      "loss": 0.4914,
+      "step": 17316
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9825066818492136e-05,
+      "loss": 0.5794,
+      "step": 17317
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9825036222294943e-05,
+      "loss": 0.5122,
+      "step": 17318
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9825005623445927e-05,
+      "loss": 0.42,
+      "step": 17319
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824975021945098e-05,
+      "loss": 0.4694,
+      "step": 17320
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824944417792456e-05,
+      "loss": 0.5156,
+      "step": 17321
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982491381098802e-05,
+      "loss": 0.5243,
+      "step": 17322
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.98248832015318e-05,
+      "loss": 0.4775,
+      "step": 17323
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982485258942379e-05,
+      "loss": 0.4667,
+      "step": 17324
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824821974664005e-05,
+      "loss": 0.4911,
+      "step": 17325
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982479135725246e-05,
+      "loss": 0.5487,
+      "step": 17326
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824760737189153e-05,
+      "loss": 0.4887,
+      "step": 17327
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824730114474103e-05,
+      "loss": 0.4357,
+      "step": 17328
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824699489107307e-05,
+      "loss": 0.4954,
+      "step": 17329
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824668861088784e-05,
+      "loss": 0.5098,
+      "step": 17330
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824638230418532e-05,
+      "loss": 0.4514,
+      "step": 17331
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824607597096566e-05,
+      "loss": 0.547,
+      "step": 17332
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824576961122896e-05,
+      "loss": 0.3886,
+      "step": 17333
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982454632249752e-05,
+      "loss": 0.4189,
+      "step": 17334
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824515681220458e-05,
+      "loss": 0.4947,
+      "step": 17335
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824485037291712e-05,
+      "loss": 0.4835,
+      "step": 17336
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982445439071129e-05,
+      "loss": 0.5062,
+      "step": 17337
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824423741479207e-05,
+      "loss": 0.4636,
+      "step": 17338
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982439308959546e-05,
+      "loss": 0.4822,
+      "step": 17339
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824362435060064e-05,
+      "loss": 0.437,
+      "step": 17340
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824331777873033e-05,
+      "loss": 0.529,
+      "step": 17341
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824301118034362e-05,
+      "loss": 0.5175,
+      "step": 17342
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824270455544067e-05,
+      "loss": 0.5582,
+      "step": 17343
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824239790402156e-05,
+      "loss": 0.5214,
+      "step": 17344
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982420912260864e-05,
+      "loss": 0.3853,
+      "step": 17345
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982417845216352e-05,
+      "loss": 0.5114,
+      "step": 17346
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982414777906681e-05,
+      "loss": 0.4578,
+      "step": 17347
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982411710331852e-05,
+      "loss": 0.5705,
+      "step": 17348
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824086424918652e-05,
+      "loss": 0.4446,
+      "step": 17349
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824055743867213e-05,
+      "loss": 0.3866,
+      "step": 17350
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9824025060164222e-05,
+      "loss": 0.5061,
+      "step": 17351
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823994373809675e-05,
+      "loss": 0.5879,
+      "step": 17352
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982396368480359e-05,
+      "loss": 0.5478,
+      "step": 17353
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982393299314597e-05,
+      "loss": 0.4931,
+      "step": 17354
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823902298836826e-05,
+      "loss": 0.4206,
+      "step": 17355
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823871601876163e-05,
+      "loss": 0.4521,
+      "step": 17356
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823840902263992e-05,
+      "loss": 0.4134,
+      "step": 17357
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982381020000032e-05,
+      "loss": 0.4554,
+      "step": 17358
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823779495085154e-05,
+      "loss": 0.4845,
+      "step": 17359
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823748787518507e-05,
+      "loss": 0.4594,
+      "step": 17360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823718077300383e-05,
+      "loss": 0.4761,
+      "step": 17361
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982368736443079e-05,
+      "loss": 0.5024,
+      "step": 17362
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823656648909743e-05,
+      "loss": 0.4202,
+      "step": 17363
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823625930737244e-05,
+      "loss": 0.4382,
+      "step": 17364
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823595209913295e-05,
+      "loss": 0.4927,
+      "step": 17365
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982356448643792e-05,
+      "loss": 0.4503,
+      "step": 17366
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823533760311117e-05,
+      "loss": 0.4841,
+      "step": 17367
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982350303153289e-05,
+      "loss": 0.4657,
+      "step": 17368
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823472300103263e-05,
+      "loss": 0.4212,
+      "step": 17369
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823441566022228e-05,
+      "loss": 0.4764,
+      "step": 17370
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823410829289804e-05,
+      "loss": 0.5916,
+      "step": 17371
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823380089905994e-05,
+      "loss": 0.5854,
+      "step": 17372
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982334934787081e-05,
+      "loss": 0.5176,
+      "step": 17373
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823318603184254e-05,
+      "loss": 0.5304,
+      "step": 17374
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823287855846344e-05,
+      "loss": 0.4591,
+      "step": 17375
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823257105857078e-05,
+      "loss": 0.5181,
+      "step": 17376
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982322635321647e-05,
+      "loss": 0.5214,
+      "step": 17377
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823195597924528e-05,
+      "loss": 0.4997,
+      "step": 17378
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982316483998126e-05,
+      "loss": 0.4038,
+      "step": 17379
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823134079386672e-05,
+      "loss": 0.4847,
+      "step": 17380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823103316140775e-05,
+      "loss": 0.459,
+      "step": 17381
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823072550243577e-05,
+      "loss": 0.488,
+      "step": 17382
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9823041781695088e-05,
+      "loss": 0.4237,
+      "step": 17383
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982301101049531e-05,
+      "loss": 0.487,
+      "step": 17384
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982298023664426e-05,
+      "loss": 0.4967,
+      "step": 17385
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822949460141935e-05,
+      "loss": 0.4636,
+      "step": 17386
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822918680988356e-05,
+      "loss": 0.4778,
+      "step": 17387
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822887899183523e-05,
+      "loss": 0.434,
+      "step": 17388
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822857114727447e-05,
+      "loss": 0.5958,
+      "step": 17389
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822826327620134e-05,
+      "loss": 0.4852,
+      "step": 17390
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822795537861597e-05,
+      "loss": 0.4437,
+      "step": 17391
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822764745451838e-05,
+      "loss": 0.4687,
+      "step": 17392
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822733950390875e-05,
+      "loss": 0.446,
+      "step": 17393
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822703152678702e-05,
+      "loss": 0.4198,
+      "step": 17394
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.982267235231534e-05,
+      "loss": 0.5557,
+      "step": 17395
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822641549300796e-05,
+      "loss": 0.4197,
+      "step": 17396
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9822610743635073e-05,
+      "loss": 0.5118,
+      "step": 17397
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982257993531818e-05,
+      "loss": 0.5149,
+      "step": 17398
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9822549124350128e-05,
+      "loss": 0.4518,
+      "step": 17399
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9822518310730924e-05,
+      "loss": 0.5186,
+      "step": 17400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9822487494460575e-05,
+      "loss": 0.5121,
+      "step": 17401
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982245667553909e-05,
+      "loss": 0.5188,
+      "step": 17402
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982242585396648e-05,
+      "loss": 0.4287,
+      "step": 17403
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982239502974275e-05,
+      "loss": 0.472,
+      "step": 17404
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982236420286791e-05,
+      "loss": 0.4923,
+      "step": 17405
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982233337334197e-05,
+      "loss": 0.4865,
+      "step": 17406
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9822302541164936e-05,
+      "loss": 0.4396,
+      "step": 17407
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9822271706336816e-05,
+      "loss": 0.4471,
+      "step": 17408
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982224086885762e-05,
+      "loss": 0.3944,
+      "step": 17409
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982221002872735e-05,
+      "loss": 0.4789,
+      "step": 17410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9822179185946023e-05,
+      "loss": 0.4166,
+      "step": 17411
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9822148340513644e-05,
+      "loss": 0.4447,
+      "step": 17412
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9822117492430225e-05,
+      "loss": 0.5404,
+      "step": 17413
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9822086641695765e-05,
+      "loss": 0.503,
+      "step": 17414
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9822055788310282e-05,
+      "loss": 0.448,
+      "step": 17415
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982202493227378e-05,
+      "loss": 0.4659,
+      "step": 17416
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821994073586263e-05,
+      "loss": 0.4797,
+      "step": 17417
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821963212247747e-05,
+      "loss": 0.4476,
+      "step": 17418
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821932348258235e-05,
+      "loss": 0.4358,
+      "step": 17419
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982190148161774e-05,
+      "loss": 0.462,
+      "step": 17420
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982187061232627e-05,
+      "loss": 0.4627,
+      "step": 17421
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821839740383826e-05,
+      "loss": 0.4864,
+      "step": 17422
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821808865790423e-05,
+      "loss": 0.4104,
+      "step": 17423
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821777988546072e-05,
+      "loss": 0.5825,
+      "step": 17424
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821747108650772e-05,
+      "loss": 0.4117,
+      "step": 17425
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982171622610454e-05,
+      "loss": 0.449,
+      "step": 17426
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821685340907376e-05,
+      "loss": 0.5455,
+      "step": 17427
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821654453059297e-05,
+      "loss": 0.4636,
+      "step": 17428
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982162356256031e-05,
+      "loss": 0.5909,
+      "step": 17429
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821592669410418e-05,
+      "loss": 0.5427,
+      "step": 17430
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982156177360963e-05,
+      "loss": 0.5443,
+      "step": 17431
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982153087515796e-05,
+      "loss": 0.4739,
+      "step": 17432
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821499974055415e-05,
+      "loss": 0.4987,
+      "step": 17433
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821469070301998e-05,
+      "loss": 0.4941,
+      "step": 17434
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821438163897718e-05,
+      "loss": 0.5229,
+      "step": 17435
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982140725484259e-05,
+      "loss": 0.4535,
+      "step": 17436
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821376343136615e-05,
+      "loss": 0.4742,
+      "step": 17437
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821345428779807e-05,
+      "loss": 0.4938,
+      "step": 17438
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821314511772175e-05,
+      "loss": 0.5115,
+      "step": 17439
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982128359211372e-05,
+      "loss": 0.4463,
+      "step": 17440
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821252669804454e-05,
+      "loss": 0.5333,
+      "step": 17441
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982122174484439e-05,
+      "loss": 0.5584,
+      "step": 17442
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821190817233527e-05,
+      "loss": 0.5046,
+      "step": 17443
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821159886971883e-05,
+      "loss": 0.4747,
+      "step": 17444
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982112895405946e-05,
+      "loss": 0.4553,
+      "step": 17445
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982109801849627e-05,
+      "loss": 0.517,
+      "step": 17446
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821067080282317e-05,
+      "loss": 0.4903,
+      "step": 17447
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9821036139417616e-05,
+      "loss": 0.4482,
+      "step": 17448
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982100519590217e-05,
+      "loss": 0.4534,
+      "step": 17449
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820974249735987e-05,
+      "loss": 0.4776,
+      "step": 17450
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982094330091908e-05,
+      "loss": 0.5102,
+      "step": 17451
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982091234945145e-05,
+      "loss": 0.5052,
+      "step": 17452
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820881395333112e-05,
+      "loss": 0.4428,
+      "step": 17453
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820850438564073e-05,
+      "loss": 0.4523,
+      "step": 17454
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982081947914434e-05,
+      "loss": 0.3991,
+      "step": 17455
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982078851707392e-05,
+      "loss": 0.5145,
+      "step": 17456
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982075755235283e-05,
+      "loss": 0.4316,
+      "step": 17457
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820726584981065e-05,
+      "loss": 0.5081,
+      "step": 17458
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982069561495864e-05,
+      "loss": 0.5661,
+      "step": 17459
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820664642285568e-05,
+      "loss": 0.4962,
+      "step": 17460
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982063366696185e-05,
+      "loss": 0.402,
+      "step": 17461
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820602688987496e-05,
+      "loss": 0.4676,
+      "step": 17462
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820571708362516e-05,
+      "loss": 0.4753,
+      "step": 17463
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982054072508692e-05,
+      "loss": 0.3958,
+      "step": 17464
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820509739160712e-05,
+      "loss": 0.4787,
+      "step": 17465
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820478750583902e-05,
+      "loss": 0.4698,
+      "step": 17466
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820447759356502e-05,
+      "loss": 0.4619,
+      "step": 17467
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820416765478512e-05,
+      "loss": 0.5344,
+      "step": 17468
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982038576894995e-05,
+      "loss": 0.5482,
+      "step": 17469
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982035476977082e-05,
+      "loss": 0.4527,
+      "step": 17470
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820323767941127e-05,
+      "loss": 0.5219,
+      "step": 17471
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820292763460887e-05,
+      "loss": 0.4839,
+      "step": 17472
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.98202617563301e-05,
+      "loss": 0.4873,
+      "step": 17473
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.982023074654878e-05,
+      "loss": 0.4663,
+      "step": 17474
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820199734116934e-05,
+      "loss": 0.4885,
+      "step": 17475
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820168719034573e-05,
+      "loss": 0.4153,
+      "step": 17476
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.98201377013017e-05,
+      "loss": 0.4175,
+      "step": 17477
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820106680918324e-05,
+      "loss": 0.4624,
+      "step": 17478
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820075657884458e-05,
+      "loss": 0.413,
+      "step": 17479
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820044632200107e-05,
+      "loss": 0.5778,
+      "step": 17480
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9820013603865277e-05,
+      "loss": 0.5213,
+      "step": 17481
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9819982572879984e-05,
+      "loss": 0.4811,
+      "step": 17482
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.981995153924423e-05,
+      "loss": 0.4301,
+      "step": 17483
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9819920502958025e-05,
+      "loss": 0.4967,
+      "step": 17484
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.981988946402138e-05,
+      "loss": 0.5075,
+      "step": 17485
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9819858422434296e-05,
+      "loss": 0.4852,
+      "step": 17486
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.981982737819679e-05,
+      "loss": 0.4537,
+      "step": 17487
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9819796331308865e-05,
+      "loss": 0.5205,
+      "step": 17488
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.981976528177053e-05,
+      "loss": 0.5683,
+      "step": 17489
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9819734229581798e-05,
+      "loss": 0.5108,
+      "step": 17490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9819703174742674e-05,
+      "loss": 0.5005,
+      "step": 17491
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.981967211725316e-05,
+      "loss": 0.4637,
+      "step": 17492
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9819641057113278e-05,
+      "loss": 0.547,
+      "step": 17493
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9819609994323026e-05,
+      "loss": 0.4365,
+      "step": 17494
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9819578928882414e-05,
+      "loss": 0.4402,
+      "step": 17495
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9819547860791453e-05,
+      "loss": 0.4796,
+      "step": 17496
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981951679005015e-05,
+      "loss": 0.4625,
+      "step": 17497
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819485716658515e-05,
+      "loss": 0.4103,
+      "step": 17498
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819454640616554e-05,
+      "loss": 0.5007,
+      "step": 17499
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819423561924274e-05,
+      "loss": 0.4499,
+      "step": 17500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981939248058169e-05,
+      "loss": 0.4792,
+      "step": 17501
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819361396588804e-05,
+      "loss": 0.4858,
+      "step": 17502
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819330309945622e-05,
+      "loss": 0.5593,
+      "step": 17503
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819299220652163e-05,
+      "loss": 0.4132,
+      "step": 17504
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819268128708426e-05,
+      "loss": 0.5337,
+      "step": 17505
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819237034114424e-05,
+      "loss": 0.4668,
+      "step": 17506
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981920593687016e-05,
+      "loss": 0.4275,
+      "step": 17507
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819174836975655e-05,
+      "loss": 0.4717,
+      "step": 17508
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819143734430904e-05,
+      "loss": 0.456,
+      "step": 17509
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981911262923592e-05,
+      "loss": 0.4811,
+      "step": 17510
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981908152139071e-05,
+      "loss": 0.43,
+      "step": 17511
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819050410895288e-05,
+      "loss": 0.4702,
+      "step": 17512
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9819019297749655e-05,
+      "loss": 0.5315,
+      "step": 17513
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818988181953826e-05,
+      "loss": 0.5548,
+      "step": 17514
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818957063507803e-05,
+      "loss": 0.5212,
+      "step": 17515
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818925942411597e-05,
+      "loss": 0.4076,
+      "step": 17516
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981889481866522e-05,
+      "loss": 0.4291,
+      "step": 17517
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818863692268677e-05,
+      "loss": 0.4552,
+      "step": 17518
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818832563221976e-05,
+      "loss": 0.541,
+      "step": 17519
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818801431525126e-05,
+      "loss": 0.435,
+      "step": 17520
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818770297178137e-05,
+      "loss": 0.5146,
+      "step": 17521
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818739160181012e-05,
+      "loss": 0.5573,
+      "step": 17522
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818708020533765e-05,
+      "loss": 0.524,
+      "step": 17523
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818676878236406e-05,
+      "loss": 0.4465,
+      "step": 17524
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818645733288938e-05,
+      "loss": 0.4448,
+      "step": 17525
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818614585691372e-05,
+      "loss": 0.4376,
+      "step": 17526
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818583435443718e-05,
+      "loss": 0.5716,
+      "step": 17527
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981855228254598e-05,
+      "loss": 0.6035,
+      "step": 17528
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818521126998168e-05,
+      "loss": 0.4637,
+      "step": 17529
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818489968800293e-05,
+      "loss": 0.475,
+      "step": 17530
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981845880795236e-05,
+      "loss": 0.49,
+      "step": 17531
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981842764445438e-05,
+      "loss": 0.4742,
+      "step": 17532
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818396478306358e-05,
+      "loss": 0.5338,
+      "step": 17533
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981836530950831e-05,
+      "loss": 0.5065,
+      "step": 17534
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818334138060237e-05,
+      "loss": 0.5279,
+      "step": 17535
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818302963962148e-05,
+      "loss": 0.5099,
+      "step": 17536
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818271787214054e-05,
+      "loss": 0.4004,
+      "step": 17537
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818240607815964e-05,
+      "loss": 0.4278,
+      "step": 17538
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981820942576788e-05,
+      "loss": 0.4938,
+      "step": 17539
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818178241069825e-05,
+      "loss": 0.5432,
+      "step": 17540
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981814705372179e-05,
+      "loss": 0.56,
+      "step": 17541
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818115863723796e-05,
+      "loss": 0.4562,
+      "step": 17542
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818084671075844e-05,
+      "loss": 0.404,
+      "step": 17543
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818053475777943e-05,
+      "loss": 0.595,
+      "step": 17544
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9818022277830106e-05,
+      "loss": 0.4128,
+      "step": 17545
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981799107723234e-05,
+      "loss": 0.4626,
+      "step": 17546
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817959873984648e-05,
+      "loss": 0.4912,
+      "step": 17547
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817928668087048e-05,
+      "loss": 0.4169,
+      "step": 17548
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981789745953954e-05,
+      "loss": 0.5128,
+      "step": 17549
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817866248342134e-05,
+      "loss": 0.5211,
+      "step": 17550
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817835034494845e-05,
+      "loss": 0.4996,
+      "step": 17551
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817803817997674e-05,
+      "loss": 0.4449,
+      "step": 17552
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817772598850632e-05,
+      "loss": 0.4522,
+      "step": 17553
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981774137705373e-05,
+      "loss": 0.4918,
+      "step": 17554
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817710152606967e-05,
+      "loss": 0.516,
+      "step": 17555
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981767892551036e-05,
+      "loss": 0.4932,
+      "step": 17556
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981764769576392e-05,
+      "loss": 0.487,
+      "step": 17557
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817616463367648e-05,
+      "loss": 0.4354,
+      "step": 17558
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817585228321557e-05,
+      "loss": 0.5031,
+      "step": 17559
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981755399062565e-05,
+      "loss": 0.4915,
+      "step": 17560
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817522750279943e-05,
+      "loss": 0.5475,
+      "step": 17561
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817491507284437e-05,
+      "loss": 0.4322,
+      "step": 17562
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981746026163915e-05,
+      "loss": 0.4921,
+      "step": 17563
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981742901334408e-05,
+      "loss": 0.4298,
+      "step": 17564
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981739776239924e-05,
+      "loss": 0.5088,
+      "step": 17565
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981736650880464e-05,
+      "loss": 0.4576,
+      "step": 17566
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817335252560288e-05,
+      "loss": 0.4631,
+      "step": 17567
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817303993666188e-05,
+      "loss": 0.4791,
+      "step": 17568
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817272732122355e-05,
+      "loss": 0.4201,
+      "step": 17569
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817241467928792e-05,
+      "loss": 0.4396,
+      "step": 17570
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981721020108551e-05,
+      "loss": 0.5616,
+      "step": 17571
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981717893159252e-05,
+      "loss": 0.4484,
+      "step": 17572
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817147659449826e-05,
+      "loss": 0.5034,
+      "step": 17573
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981711638465744e-05,
+      "loss": 0.4584,
+      "step": 17574
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9817085107215363e-05,
+      "loss": 0.4253,
+      "step": 17575
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981705382712361e-05,
+      "loss": 0.4549,
+      "step": 17576
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981702254438219e-05,
+      "loss": 0.4381,
+      "step": 17577
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981699125899111e-05,
+      "loss": 0.4849,
+      "step": 17578
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9816959970950376e-05,
+      "loss": 0.4323,
+      "step": 17579
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981692868026e-05,
+      "loss": 0.44,
+      "step": 17580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981689738691999e-05,
+      "loss": 0.4424,
+      "step": 17581
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9816866090930353e-05,
+      "loss": 0.5351,
+      "step": 17582
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9816834792291098e-05,
+      "loss": 0.4678,
+      "step": 17583
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9816803491002236e-05,
+      "loss": 0.4592,
+      "step": 17584
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9816772187063767e-05,
+      "loss": 0.4484,
+      "step": 17585
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9816740880475708e-05,
+      "loss": 0.4711,
+      "step": 17586
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9816709571238065e-05,
+      "loss": 0.4983,
+      "step": 17587
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981667825935085e-05,
+      "loss": 0.5144,
+      "step": 17588
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981664694481406e-05,
+      "loss": 0.5383,
+      "step": 17589
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9816615627627714e-05,
+      "loss": 0.4089,
+      "step": 17590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981658430779182e-05,
+      "loss": 0.5142,
+      "step": 17591
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981655298530638e-05,
+      "loss": 0.5122,
+      "step": 17592
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.981652166017141e-05,
+      "loss": 0.4237,
+      "step": 17593
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9816490332386914e-05,
+      "loss": 0.4611,
+      "step": 17594
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816459001952902e-05,
+      "loss": 0.513,
+      "step": 17595
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981642766886938e-05,
+      "loss": 0.4678,
+      "step": 17596
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816396333136362e-05,
+      "loss": 0.4985,
+      "step": 17597
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981636499475385e-05,
+      "loss": 0.5569,
+      "step": 17598
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816333653721857e-05,
+      "loss": 0.3976,
+      "step": 17599
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816302310040385e-05,
+      "loss": 0.5299,
+      "step": 17600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816270963709452e-05,
+      "loss": 0.4345,
+      "step": 17601
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816239614729057e-05,
+      "loss": 0.4062,
+      "step": 17602
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816208263099218e-05,
+      "loss": 0.5155,
+      "step": 17603
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816176908819934e-05,
+      "loss": 0.4561,
+      "step": 17604
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816145551891223e-05,
+      "loss": 0.4689,
+      "step": 17605
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816114192313084e-05,
+      "loss": 0.4576,
+      "step": 17606
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981608283008553e-05,
+      "loss": 0.435,
+      "step": 17607
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981605146520857e-05,
+      "loss": 0.4416,
+      "step": 17608
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9816020097682217e-05,
+      "loss": 0.4011,
+      "step": 17609
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815988727506466e-05,
+      "loss": 0.4147,
+      "step": 17610
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815957354681342e-05,
+      "loss": 0.51,
+      "step": 17611
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815925979206838e-05,
+      "loss": 0.4506,
+      "step": 17612
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815894601082973e-05,
+      "loss": 0.4687,
+      "step": 17613
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815863220309753e-05,
+      "loss": 0.437,
+      "step": 17614
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815831836887182e-05,
+      "loss": 0.4904,
+      "step": 17615
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815800450815276e-05,
+      "loss": 0.443,
+      "step": 17616
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981576906209404e-05,
+      "loss": 0.5898,
+      "step": 17617
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815737670723478e-05,
+      "loss": 0.4744,
+      "step": 17618
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815706276703604e-05,
+      "loss": 0.504,
+      "step": 17619
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815674880034428e-05,
+      "loss": 0.4692,
+      "step": 17620
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815643480715953e-05,
+      "loss": 0.4296,
+      "step": 17621
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981561207874819e-05,
+      "loss": 0.4959,
+      "step": 17622
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815580674131147e-05,
+      "loss": 0.4333,
+      "step": 17623
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815549266864833e-05,
+      "loss": 0.4742,
+      "step": 17624
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815517856949257e-05,
+      "loss": 0.4942,
+      "step": 17625
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815486444384427e-05,
+      "loss": 0.5549,
+      "step": 17626
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815455029170348e-05,
+      "loss": 0.4772,
+      "step": 17627
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981542361130704e-05,
+      "loss": 0.5496,
+      "step": 17628
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815392190794495e-05,
+      "loss": 0.5281,
+      "step": 17629
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815360767632733e-05,
+      "loss": 0.4034,
+      "step": 17630
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815329341821758e-05,
+      "loss": 0.4592,
+      "step": 17631
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981529791336158e-05,
+      "loss": 0.4595,
+      "step": 17632
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815266482252208e-05,
+      "loss": 0.5801,
+      "step": 17633
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981523504849365e-05,
+      "loss": 0.5251,
+      "step": 17634
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815203612085918e-05,
+      "loss": 0.4431,
+      "step": 17635
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981517217302901e-05,
+      "loss": 0.4902,
+      "step": 17636
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815140731322943e-05,
+      "loss": 0.4333,
+      "step": 17637
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815109286967726e-05,
+      "loss": 0.4525,
+      "step": 17638
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815077839963363e-05,
+      "loss": 0.4408,
+      "step": 17639
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9815046390309867e-05,
+      "loss": 0.4099,
+      "step": 17640
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981501493800724e-05,
+      "loss": 0.4585,
+      "step": 17641
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814983483055496e-05,
+      "loss": 0.458,
+      "step": 17642
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814952025454642e-05,
+      "loss": 0.4796,
+      "step": 17643
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981492056520469e-05,
+      "loss": 0.5386,
+      "step": 17644
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981488910230564e-05,
+      "loss": 0.4542,
+      "step": 17645
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981485763675751e-05,
+      "loss": 0.4416,
+      "step": 17646
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814826168560302e-05,
+      "loss": 0.5073,
+      "step": 17647
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814794697714024e-05,
+      "loss": 0.5544,
+      "step": 17648
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814763224218692e-05,
+      "loss": 0.5217,
+      "step": 17649
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814731748074308e-05,
+      "loss": 0.5662,
+      "step": 17650
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814700269280883e-05,
+      "loss": 0.4769,
+      "step": 17651
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981466878783842e-05,
+      "loss": 0.4902,
+      "step": 17652
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814637303746936e-05,
+      "loss": 0.6659,
+      "step": 17653
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981460581700643e-05,
+      "loss": 0.5099,
+      "step": 17654
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814574327616922e-05,
+      "loss": 0.4598,
+      "step": 17655
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814542835578415e-05,
+      "loss": 0.4474,
+      "step": 17656
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814511340890915e-05,
+      "loss": 0.4674,
+      "step": 17657
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814479843554434e-05,
+      "loss": 0.5121,
+      "step": 17658
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814448343568973e-05,
+      "loss": 0.5547,
+      "step": 17659
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814416840934552e-05,
+      "loss": 0.3967,
+      "step": 17660
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814385335651175e-05,
+      "loss": 0.6112,
+      "step": 17661
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814353827718844e-05,
+      "loss": 0.4788,
+      "step": 17662
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814322317137576e-05,
+      "loss": 0.5832,
+      "step": 17663
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981429080390738e-05,
+      "loss": 0.452,
+      "step": 17664
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814259288028257e-05,
+      "loss": 0.4396,
+      "step": 17665
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981422776950022e-05,
+      "loss": 0.4822,
+      "step": 17666
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814196248323277e-05,
+      "loss": 0.4397,
+      "step": 17667
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814164724497437e-05,
+      "loss": 0.484,
+      "step": 17668
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981413319802271e-05,
+      "loss": 0.4844,
+      "step": 17669
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.98141016688991e-05,
+      "loss": 0.5591,
+      "step": 17670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9814070137126616e-05,
+      "loss": 0.4259,
+      "step": 17671
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981403860270527e-05,
+      "loss": 0.4665,
+      "step": 17672
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981400706563507e-05,
+      "loss": 0.4752,
+      "step": 17673
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813975525916027e-05,
+      "loss": 0.475,
+      "step": 17674
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813943983548144e-05,
+      "loss": 0.5127,
+      "step": 17675
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981391243853143e-05,
+      "loss": 0.4701,
+      "step": 17676
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813880890865898e-05,
+      "loss": 0.5284,
+      "step": 17677
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981384934055155e-05,
+      "loss": 0.4737,
+      "step": 17678
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813817787588402e-05,
+      "loss": 0.4625,
+      "step": 17679
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813786231976454e-05,
+      "loss": 0.426,
+      "step": 17680
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813754673715723e-05,
+      "loss": 0.3992,
+      "step": 17681
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813723112806214e-05,
+      "loss": 0.4714,
+      "step": 17682
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813691549247932e-05,
+      "loss": 0.4493,
+      "step": 17683
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813659983040893e-05,
+      "loss": 0.5215,
+      "step": 17684
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813628414185097e-05,
+      "loss": 0.5497,
+      "step": 17685
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813596842680562e-05,
+      "loss": 0.4674,
+      "step": 17686
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813565268527287e-05,
+      "loss": 0.5302,
+      "step": 17687
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813533691725285e-05,
+      "loss": 0.5026,
+      "step": 17688
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813502112274564e-05,
+      "loss": 0.4623,
+      "step": 17689
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813470530175137e-05,
+      "loss": 0.4391,
+      "step": 17690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9813438945427005e-05,
+      "loss": 0.4061,
+      "step": 17691
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981340735803018e-05,
+      "loss": 0.536,
+      "step": 17692
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.981337576798467e-05,
+      "loss": 0.4531,
+      "step": 17693
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9813344175290486e-05,
+      "loss": 0.4107,
+      "step": 17694
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981331257994763e-05,
+      "loss": 0.4515,
+      "step": 17695
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981328098195612e-05,
+      "loss": 0.4626,
+      "step": 17696
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981324938131596e-05,
+      "loss": 0.4791,
+      "step": 17697
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9813217778027157e-05,
+      "loss": 0.4838,
+      "step": 17698
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981318617208972e-05,
+      "loss": 0.5362,
+      "step": 17699
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9813154563503657e-05,
+      "loss": 0.4557,
+      "step": 17700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981312295226898e-05,
+      "loss": 0.4757,
+      "step": 17701
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9813091338385692e-05,
+      "loss": 0.4402,
+      "step": 17702
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9813059721853807e-05,
+      "loss": 0.4441,
+      "step": 17703
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9813028102673334e-05,
+      "loss": 0.466,
+      "step": 17704
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812996480844274e-05,
+      "loss": 0.445,
+      "step": 17705
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812964856366644e-05,
+      "loss": 0.4164,
+      "step": 17706
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812933229240446e-05,
+      "loss": 0.4944,
+      "step": 17707
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812901599465695e-05,
+      "loss": 0.4524,
+      "step": 17708
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981286996704239e-05,
+      "loss": 0.467,
+      "step": 17709
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981283833197055e-05,
+      "loss": 0.3994,
+      "step": 17710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981280669425018e-05,
+      "loss": 0.4537,
+      "step": 17711
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812775053881286e-05,
+      "loss": 0.429,
+      "step": 17712
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812743410863877e-05,
+      "loss": 0.4252,
+      "step": 17713
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812711765197965e-05,
+      "loss": 0.4972,
+      "step": 17714
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812680116883553e-05,
+      "loss": 0.4697,
+      "step": 17715
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981264846592066e-05,
+      "loss": 0.4811,
+      "step": 17716
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812616812309277e-05,
+      "loss": 0.4428,
+      "step": 17717
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981258515604943e-05,
+      "loss": 0.4469,
+      "step": 17718
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812553497141118e-05,
+      "loss": 0.4494,
+      "step": 17719
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812521835584353e-05,
+      "loss": 0.5381,
+      "step": 17720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812490171379143e-05,
+      "loss": 0.5259,
+      "step": 17721
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812458504525494e-05,
+      "loss": 0.4387,
+      "step": 17722
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812426835023417e-05,
+      "loss": 0.5338,
+      "step": 17723
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812395162872923e-05,
+      "loss": 0.4892,
+      "step": 17724
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812363488074014e-05,
+      "loss": 0.4709,
+      "step": 17725
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812331810626704e-05,
+      "loss": 0.5045,
+      "step": 17726
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812300130530996e-05,
+      "loss": 0.4697,
+      "step": 17727
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812268447786908e-05,
+      "loss": 0.4925,
+      "step": 17728
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812236762394442e-05,
+      "loss": 0.4406,
+      "step": 17729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812205074353606e-05,
+      "loss": 0.5072,
+      "step": 17730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981217338366441e-05,
+      "loss": 0.4085,
+      "step": 17731
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981214169032686e-05,
+      "loss": 0.4806,
+      "step": 17732
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981210999434097e-05,
+      "loss": 0.4755,
+      "step": 17733
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812078295706743e-05,
+      "loss": 0.5372,
+      "step": 17734
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812046594424194e-05,
+      "loss": 0.4941,
+      "step": 17735
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9812014890493328e-05,
+      "loss": 0.5257,
+      "step": 17736
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981198318391415e-05,
+      "loss": 0.4688,
+      "step": 17737
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981195147468667e-05,
+      "loss": 0.4572,
+      "step": 17738
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811919762810904e-05,
+      "loss": 0.4711,
+      "step": 17739
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811888048286854e-05,
+      "loss": 0.5303,
+      "step": 17740
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811856331114526e-05,
+      "loss": 0.4423,
+      "step": 17741
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811824611293935e-05,
+      "loss": 0.4274,
+      "step": 17742
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981179288882509e-05,
+      "loss": 0.566,
+      "step": 17743
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981176116370799e-05,
+      "loss": 0.4603,
+      "step": 17744
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981172943594265e-05,
+      "loss": 0.4403,
+      "step": 17745
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811697705529084e-05,
+      "loss": 0.4661,
+      "step": 17746
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981166597246729e-05,
+      "loss": 0.4402,
+      "step": 17747
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981163423675728e-05,
+      "loss": 0.5166,
+      "step": 17748
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981160249839907e-05,
+      "loss": 0.5068,
+      "step": 17749
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811570757392656e-05,
+      "loss": 0.4659,
+      "step": 17750
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981153901373806e-05,
+      "loss": 0.466,
+      "step": 17751
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811507267435282e-05,
+      "loss": 0.4422,
+      "step": 17752
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981147551848433e-05,
+      "loss": 0.3959,
+      "step": 17753
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811443766885215e-05,
+      "loss": 0.4633,
+      "step": 17754
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811412012637946e-05,
+      "loss": 0.4819,
+      "step": 17755
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981138025574253e-05,
+      "loss": 0.4707,
+      "step": 17756
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811348496198977e-05,
+      "loss": 0.5194,
+      "step": 17757
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.98113167340073e-05,
+      "loss": 0.452,
+      "step": 17758
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811284969167496e-05,
+      "loss": 0.4625,
+      "step": 17759
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811253201679584e-05,
+      "loss": 0.5583,
+      "step": 17760
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981122143154357e-05,
+      "loss": 0.455,
+      "step": 17761
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811189658759455e-05,
+      "loss": 0.5456,
+      "step": 17762
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981115788332726e-05,
+      "loss": 0.5061,
+      "step": 17763
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811126105246986e-05,
+      "loss": 0.4916,
+      "step": 17764
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811094324518642e-05,
+      "loss": 0.449,
+      "step": 17765
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811062541142242e-05,
+      "loss": 0.4848,
+      "step": 17766
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9811030755117786e-05,
+      "loss": 0.4653,
+      "step": 17767
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981099896644529e-05,
+      "loss": 0.4827,
+      "step": 17768
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810967175124757e-05,
+      "loss": 0.5458,
+      "step": 17769
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810935381156196e-05,
+      "loss": 0.4718,
+      "step": 17770
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810903584539624e-05,
+      "loss": 0.4631,
+      "step": 17771
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981087178527504e-05,
+      "loss": 0.5028,
+      "step": 17772
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810839983362456e-05,
+      "loss": 0.4743,
+      "step": 17773
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810808178801878e-05,
+      "loss": 0.4719,
+      "step": 17774
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981077637159332e-05,
+      "loss": 0.4756,
+      "step": 17775
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810744561736787e-05,
+      "loss": 0.437,
+      "step": 17776
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981071274923229e-05,
+      "loss": 0.4935,
+      "step": 17777
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810680934079835e-05,
+      "loss": 0.4717,
+      "step": 17778
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810649116279426e-05,
+      "loss": 0.5365,
+      "step": 17779
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981061729583108e-05,
+      "loss": 0.4627,
+      "step": 17780
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810585472734806e-05,
+      "loss": 0.446,
+      "step": 17781
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810553646990607e-05,
+      "loss": 0.5353,
+      "step": 17782
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810521818598495e-05,
+      "loss": 0.4304,
+      "step": 17783
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810489987558474e-05,
+      "loss": 0.5119,
+      "step": 17784
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810458153870556e-05,
+      "loss": 0.6345,
+      "step": 17785
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810426317534753e-05,
+      "loss": 0.4971,
+      "step": 17786
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981039447855107e-05,
+      "loss": 0.4941,
+      "step": 17787
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981036263691951e-05,
+      "loss": 0.4509,
+      "step": 17788
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9810330792640094e-05,
+      "loss": 0.4317,
+      "step": 17789
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.981029894571282e-05,
+      "loss": 0.43,
+      "step": 17790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.98102670961377e-05,
+      "loss": 0.4192,
+      "step": 17791
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9810235243914746e-05,
+      "loss": 0.377,
+      "step": 17792
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.981020338904396e-05,
+      "loss": 0.4915,
+      "step": 17793
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9810171531525358e-05,
+      "loss": 0.5492,
+      "step": 17794
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9810139671358943e-05,
+      "loss": 0.5294,
+      "step": 17795
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9810107808544727e-05,
+      "loss": 0.4769,
+      "step": 17796
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9810075943082716e-05,
+      "loss": 0.5819,
+      "step": 17797
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9810044074972918e-05,
+      "loss": 0.5448,
+      "step": 17798
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9810012204215345e-05,
+      "loss": 0.481,
+      "step": 17799
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809980330810002e-05,
+      "loss": 0.4614,
+      "step": 17800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809948454756905e-05,
+      "loss": 0.4266,
+      "step": 17801
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980991657605605e-05,
+      "loss": 0.5365,
+      "step": 17802
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809884694707456e-05,
+      "loss": 0.4642,
+      "step": 17803
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809852810711128e-05,
+      "loss": 0.4468,
+      "step": 17804
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809820924067073e-05,
+      "loss": 0.5244,
+      "step": 17805
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809789034775305e-05,
+      "loss": 0.4226,
+      "step": 17806
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809757142835827e-05,
+      "loss": 0.5185,
+      "step": 17807
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980972524824865e-05,
+      "loss": 0.48,
+      "step": 17808
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809693351013782e-05,
+      "loss": 0.423,
+      "step": 17809
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809661451131236e-05,
+      "loss": 0.4932,
+      "step": 17810
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980962954860101e-05,
+      "loss": 0.465,
+      "step": 17811
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809597643423123e-05,
+      "loss": 0.4736,
+      "step": 17812
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980956573559758e-05,
+      "loss": 0.4511,
+      "step": 17813
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809533825124387e-05,
+      "loss": 0.4211,
+      "step": 17814
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809501912003556e-05,
+      "loss": 0.4869,
+      "step": 17815
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809469996235093e-05,
+      "loss": 0.5015,
+      "step": 17816
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980943807781901e-05,
+      "loss": 0.4419,
+      "step": 17817
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809406156755312e-05,
+      "loss": 0.4492,
+      "step": 17818
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980937423304401e-05,
+      "loss": 0.4952,
+      "step": 17819
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809342306685116e-05,
+      "loss": 0.4597,
+      "step": 17820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809310377678633e-05,
+      "loss": 0.5099,
+      "step": 17821
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980927844602457e-05,
+      "loss": 0.4197,
+      "step": 17822
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809246511722937e-05,
+      "loss": 0.5168,
+      "step": 17823
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809214574773737e-05,
+      "loss": 0.53,
+      "step": 17824
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809182635176994e-05,
+      "loss": 0.4462,
+      "step": 17825
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.98091506929327e-05,
+      "loss": 0.4936,
+      "step": 17826
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809118748040873e-05,
+      "loss": 0.4586,
+      "step": 17827
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980908680050152e-05,
+      "loss": 0.4794,
+      "step": 17828
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809054850314645e-05,
+      "loss": 0.6313,
+      "step": 17829
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9809022897480264e-05,
+      "loss": 0.4734,
+      "step": 17830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980899094199838e-05,
+      "loss": 0.4578,
+      "step": 17831
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808958983869004e-05,
+      "loss": 0.4983,
+      "step": 17832
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808927023092144e-05,
+      "loss": 0.5328,
+      "step": 17833
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980889505966781e-05,
+      "loss": 0.6039,
+      "step": 17834
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808863093596007e-05,
+      "loss": 0.4146,
+      "step": 17835
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808831124876747e-05,
+      "loss": 0.5468,
+      "step": 17836
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808799153510037e-05,
+      "loss": 0.4877,
+      "step": 17837
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808767179495886e-05,
+      "loss": 0.5276,
+      "step": 17838
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808735202834306e-05,
+      "loss": 0.4972,
+      "step": 17839
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.98087032235253e-05,
+      "loss": 0.4824,
+      "step": 17840
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980867124156888e-05,
+      "loss": 0.5115,
+      "step": 17841
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980863925696505e-05,
+      "loss": 0.5457,
+      "step": 17842
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808607269713828e-05,
+      "loss": 0.4707,
+      "step": 17843
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808575279815213e-05,
+      "loss": 0.5147,
+      "step": 17844
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980854328726922e-05,
+      "loss": 0.5541,
+      "step": 17845
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808511292075854e-05,
+      "loss": 0.4581,
+      "step": 17846
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808479294235127e-05,
+      "loss": 0.4437,
+      "step": 17847
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808447293747045e-05,
+      "loss": 0.4717,
+      "step": 17848
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808415290611616e-05,
+      "loss": 0.5068,
+      "step": 17849
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980838328482885e-05,
+      "loss": 0.3694,
+      "step": 17850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808351276398758e-05,
+      "loss": 0.5189,
+      "step": 17851
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808319265321344e-05,
+      "loss": 0.5701,
+      "step": 17852
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808287251596615e-05,
+      "loss": 0.5103,
+      "step": 17853
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980825523522459e-05,
+      "loss": 0.5049,
+      "step": 17854
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808223216205266e-05,
+      "loss": 0.538,
+      "step": 17855
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980819119453866e-05,
+      "loss": 0.5219,
+      "step": 17856
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808159170224778e-05,
+      "loss": 0.4702,
+      "step": 17857
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808127143263624e-05,
+      "loss": 0.5294,
+      "step": 17858
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808095113655214e-05,
+      "loss": 0.5048,
+      "step": 17859
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9808063081399554e-05,
+      "loss": 0.5258,
+      "step": 17860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980803104649665e-05,
+      "loss": 0.5792,
+      "step": 17861
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807999008946512e-05,
+      "loss": 0.5316,
+      "step": 17862
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980796696874915e-05,
+      "loss": 0.5271,
+      "step": 17863
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807934925904573e-05,
+      "loss": 0.538,
+      "step": 17864
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807902880412787e-05,
+      "loss": 0.4836,
+      "step": 17865
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.98078708322738e-05,
+      "loss": 0.5,
+      "step": 17866
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807838781487628e-05,
+      "loss": 0.4607,
+      "step": 17867
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980780672805427e-05,
+      "loss": 0.4717,
+      "step": 17868
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980777467197374e-05,
+      "loss": 0.4186,
+      "step": 17869
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807742613246048e-05,
+      "loss": 0.3959,
+      "step": 17870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.98077105518712e-05,
+      "loss": 0.4413,
+      "step": 17871
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807678487849205e-05,
+      "loss": 0.5151,
+      "step": 17872
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807646421180068e-05,
+      "loss": 0.5144,
+      "step": 17873
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807614351863807e-05,
+      "loss": 0.4729,
+      "step": 17874
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807582279900422e-05,
+      "loss": 0.4962,
+      "step": 17875
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807550205289924e-05,
+      "loss": 0.5441,
+      "step": 17876
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807518128032322e-05,
+      "loss": 0.4465,
+      "step": 17877
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807486048127627e-05,
+      "loss": 0.5012,
+      "step": 17878
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807453965575846e-05,
+      "loss": 0.4891,
+      "step": 17879
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807421880376986e-05,
+      "loss": 0.5479,
+      "step": 17880
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980738979253106e-05,
+      "loss": 0.5362,
+      "step": 17881
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980735770203807e-05,
+      "loss": 0.4516,
+      "step": 17882
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980732560889803e-05,
+      "loss": 0.4886,
+      "step": 17883
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807293513110943e-05,
+      "loss": 0.5767,
+      "step": 17884
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807261414676827e-05,
+      "loss": 0.4756,
+      "step": 17885
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807229313595685e-05,
+      "loss": 0.4217,
+      "step": 17886
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807197209867525e-05,
+      "loss": 0.4545,
+      "step": 17887
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807165103492352e-05,
+      "loss": 0.4883,
+      "step": 17888
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807132994470185e-05,
+      "loss": 0.4822,
+      "step": 17889
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9807100882801023e-05,
+      "loss": 0.4871,
+      "step": 17890
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9807068768484883e-05,
+      "loss": 0.4593,
+      "step": 17891
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980703665152177e-05,
+      "loss": 0.5695,
+      "step": 17892
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9807004531911686e-05,
+      "loss": 0.468,
+      "step": 17893
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980697240965465e-05,
+      "loss": 0.4621,
+      "step": 17894
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806940284750662e-05,
+      "loss": 0.553,
+      "step": 17895
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806908157199737e-05,
+      "loss": 0.4354,
+      "step": 17896
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806876027001885e-05,
+      "loss": 0.4598,
+      "step": 17897
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980684389415711e-05,
+      "loss": 0.4781,
+      "step": 17898
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980681175866542e-05,
+      "loss": 0.4428,
+      "step": 17899
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806779620526827e-05,
+      "loss": 0.4413,
+      "step": 17900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806747479741335e-05,
+      "loss": 0.5052,
+      "step": 17901
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806715336308962e-05,
+      "loss": 0.5041,
+      "step": 17902
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806683190229707e-05,
+      "loss": 0.5104,
+      "step": 17903
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980665104150358e-05,
+      "loss": 0.4611,
+      "step": 17904
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806618890130597e-05,
+      "loss": 0.4303,
+      "step": 17905
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980658673611076e-05,
+      "loss": 0.5116,
+      "step": 17906
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980655457944408e-05,
+      "loss": 0.4712,
+      "step": 17907
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806522420130562e-05,
+      "loss": 0.5572,
+      "step": 17908
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980649025817022e-05,
+      "loss": 0.4943,
+      "step": 17909
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806458093563062e-05,
+      "loss": 0.4628,
+      "step": 17910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980642592630909e-05,
+      "loss": 0.4924,
+      "step": 17911
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806393756408325e-05,
+      "loss": 0.4366,
+      "step": 17912
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806361583860763e-05,
+      "loss": 0.4097,
+      "step": 17913
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980632940866642e-05,
+      "loss": 0.4319,
+      "step": 17914
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806297230825303e-05,
+      "loss": 0.5117,
+      "step": 17915
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980626505033742e-05,
+      "loss": 0.5584,
+      "step": 17916
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980623286720278e-05,
+      "loss": 0.5463,
+      "step": 17917
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980620068142139e-05,
+      "loss": 0.5134,
+      "step": 17918
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806168492993266e-05,
+      "loss": 0.4188,
+      "step": 17919
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980613630191841e-05,
+      "loss": 0.495,
+      "step": 17920
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980610410819683e-05,
+      "loss": 0.5352,
+      "step": 17921
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806071911828536e-05,
+      "loss": 0.5207,
+      "step": 17922
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980603971281354e-05,
+      "loss": 0.5216,
+      "step": 17923
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806007511151846e-05,
+      "loss": 0.4488,
+      "step": 17924
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805975306843463e-05,
+      "loss": 0.4774,
+      "step": 17925
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805943099888404e-05,
+      "loss": 0.5483,
+      "step": 17926
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805910890286678e-05,
+      "loss": 0.4685,
+      "step": 17927
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805878678038283e-05,
+      "loss": 0.4425,
+      "step": 17928
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805846463143245e-05,
+      "loss": 0.4431,
+      "step": 17929
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805814245601554e-05,
+      "loss": 0.5219,
+      "step": 17930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805782025413235e-05,
+      "loss": 0.4687,
+      "step": 17931
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805749802578286e-05,
+      "loss": 0.5014,
+      "step": 17932
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980571757709672e-05,
+      "loss": 0.4658,
+      "step": 17933
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805685348968543e-05,
+      "loss": 0.4649,
+      "step": 17934
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980565311819377e-05,
+      "loss": 0.4772,
+      "step": 17935
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805620884772402e-05,
+      "loss": 0.5227,
+      "step": 17936
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980558864870445e-05,
+      "loss": 0.401,
+      "step": 17937
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980555640998993e-05,
+      "loss": 0.5139,
+      "step": 17938
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805524168628838e-05,
+      "loss": 0.485,
+      "step": 17939
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805491924621193e-05,
+      "loss": 0.4962,
+      "step": 17940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805459677966997e-05,
+      "loss": 0.5752,
+      "step": 17941
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805427428666264e-05,
+      "loss": 0.4639,
+      "step": 17942
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805395176719e-05,
+      "loss": 0.4347,
+      "step": 17943
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805362922125214e-05,
+      "loss": 0.4965,
+      "step": 17944
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805330664884912e-05,
+      "loss": 0.4859,
+      "step": 17945
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805298404998107e-05,
+      "loss": 0.4988,
+      "step": 17946
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805266142464808e-05,
+      "loss": 0.4886,
+      "step": 17947
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805233877285022e-05,
+      "loss": 0.4479,
+      "step": 17948
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805201609458753e-05,
+      "loss": 0.3736,
+      "step": 17949
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805169338986018e-05,
+      "loss": 0.4345,
+      "step": 17950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805137065866823e-05,
+      "loss": 0.5018,
+      "step": 17951
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980510479010117e-05,
+      "loss": 0.4169,
+      "step": 17952
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9805072511689078e-05,
+      "loss": 0.4308,
+      "step": 17953
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980504023063055e-05,
+      "loss": 0.5223,
+      "step": 17954
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.98050079469256e-05,
+      "loss": 0.4495,
+      "step": 17955
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804975660574225e-05,
+      "loss": 0.4511,
+      "step": 17956
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980494337157645e-05,
+      "loss": 0.4662,
+      "step": 17957
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804911079932267e-05,
+      "loss": 0.4604,
+      "step": 17958
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804878785641697e-05,
+      "loss": 0.514,
+      "step": 17959
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804846488704742e-05,
+      "loss": 0.4876,
+      "step": 17960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804814189121412e-05,
+      "loss": 0.5305,
+      "step": 17961
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980478188689172e-05,
+      "loss": 0.4871,
+      "step": 17962
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804749582015672e-05,
+      "loss": 0.465,
+      "step": 17963
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980471727449327e-05,
+      "loss": 0.463,
+      "step": 17964
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804684964324537e-05,
+      "loss": 0.4242,
+      "step": 17965
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980465265150947e-05,
+      "loss": 0.5026,
+      "step": 17966
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804620336048082e-05,
+      "loss": 0.4646,
+      "step": 17967
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804588017940382e-05,
+      "loss": 0.4677,
+      "step": 17968
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804555697186375e-05,
+      "loss": 0.4118,
+      "step": 17969
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804523373786074e-05,
+      "loss": 0.4472,
+      "step": 17970
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980449104773949e-05,
+      "loss": 0.5315,
+      "step": 17971
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804458719046623e-05,
+      "loss": 0.4904,
+      "step": 17972
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980442638770749e-05,
+      "loss": 0.4117,
+      "step": 17973
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804394053722092e-05,
+      "loss": 0.5315,
+      "step": 17974
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804361717090446e-05,
+      "loss": 0.4295,
+      "step": 17975
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804329377812557e-05,
+      "loss": 0.465,
+      "step": 17976
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804297035888435e-05,
+      "loss": 0.4561,
+      "step": 17977
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804264691318084e-05,
+      "loss": 0.4118,
+      "step": 17978
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804232344101514e-05,
+      "loss": 0.4388,
+      "step": 17979
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980419999423874e-05,
+      "loss": 0.436,
+      "step": 17980
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804167641729765e-05,
+      "loss": 0.5283,
+      "step": 17981
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804135286574602e-05,
+      "loss": 0.4595,
+      "step": 17982
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804102928773258e-05,
+      "loss": 0.4473,
+      "step": 17983
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804070568325735e-05,
+      "loss": 0.557,
+      "step": 17984
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.980403820523205e-05,
+      "loss": 0.506,
+      "step": 17985
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9804005839492212e-05,
+      "loss": 0.447,
+      "step": 17986
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9803973471106222e-05,
+      "loss": 0.4834,
+      "step": 17987
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9803941100074097e-05,
+      "loss": 0.4365,
+      "step": 17988
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9803908726395842e-05,
+      "loss": 0.446,
+      "step": 17989
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803876350071466e-05,
+      "loss": 0.4635,
+      "step": 17990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980384397110098e-05,
+      "loss": 0.4951,
+      "step": 17991
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980381158948439e-05,
+      "loss": 0.5239,
+      "step": 17992
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.98037792052217e-05,
+      "loss": 0.5323,
+      "step": 17993
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980374681831293e-05,
+      "loss": 0.5849,
+      "step": 17994
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980371442875808e-05,
+      "loss": 0.5041,
+      "step": 17995
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803682036557165e-05,
+      "loss": 0.48,
+      "step": 17996
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803649641710188e-05,
+      "loss": 0.4546,
+      "step": 17997
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980361724421716e-05,
+      "loss": 0.5256,
+      "step": 17998
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980358484407809e-05,
+      "loss": 0.5364,
+      "step": 17999
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803552441292985e-05,
+      "loss": 0.4778,
+      "step": 18000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980352003586186e-05,
+      "loss": 0.4097,
+      "step": 18001
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803487627784715e-05,
+      "loss": 0.478,
+      "step": 18002
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803455217061566e-05,
+      "loss": 0.44,
+      "step": 18003
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803422803692417e-05,
+      "loss": 0.4462,
+      "step": 18004
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803390387677276e-05,
+      "loss": 0.483,
+      "step": 18005
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803357969016157e-05,
+      "loss": 0.4604,
+      "step": 18006
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803325547709063e-05,
+      "loss": 0.5098,
+      "step": 18007
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803293123756008e-05,
+      "loss": 0.49,
+      "step": 18008
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803260697157e-05,
+      "loss": 0.5212,
+      "step": 18009
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803228267912046e-05,
+      "loss": 0.457,
+      "step": 18010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803195836021153e-05,
+      "loss": 0.4944,
+      "step": 18011
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980316340148433e-05,
+      "loss": 0.4873,
+      "step": 18012
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803130964301588e-05,
+      "loss": 0.5112,
+      "step": 18013
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980309852447294e-05,
+      "loss": 0.5,
+      "step": 18014
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803066081998384e-05,
+      "loss": 0.5037,
+      "step": 18015
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803033636877934e-05,
+      "loss": 0.4893,
+      "step": 18016
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9803001189111604e-05,
+      "loss": 0.4609,
+      "step": 18017
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802968738699397e-05,
+      "loss": 0.4223,
+      "step": 18018
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980293628564132e-05,
+      "loss": 0.4745,
+      "step": 18019
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980290382993739e-05,
+      "loss": 0.4971,
+      "step": 18020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802871371587605e-05,
+      "loss": 0.4968,
+      "step": 18021
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802838910591982e-05,
+      "loss": 0.5219,
+      "step": 18022
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802806446950527e-05,
+      "loss": 0.4674,
+      "step": 18023
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980277398066325e-05,
+      "loss": 0.4093,
+      "step": 18024
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802741511730156e-05,
+      "loss": 0.4814,
+      "step": 18025
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802709040151256e-05,
+      "loss": 0.4853,
+      "step": 18026
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802676565926558e-05,
+      "loss": 0.4466,
+      "step": 18027
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802644089056078e-05,
+      "loss": 0.4533,
+      "step": 18028
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802611609539812e-05,
+      "loss": 0.5359,
+      "step": 18029
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980257912737778e-05,
+      "loss": 0.387,
+      "step": 18030
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802546642569983e-05,
+      "loss": 0.4372,
+      "step": 18031
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802514155116433e-05,
+      "loss": 0.5462,
+      "step": 18032
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802481665017138e-05,
+      "loss": 0.5581,
+      "step": 18033
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802449172272108e-05,
+      "loss": 0.4666,
+      "step": 18034
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802416676881354e-05,
+      "loss": 0.6035,
+      "step": 18035
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980238417884488e-05,
+      "loss": 0.4416,
+      "step": 18036
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802351678162694e-05,
+      "loss": 0.5413,
+      "step": 18037
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980231917483481e-05,
+      "loss": 0.4276,
+      "step": 18038
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802286668861234e-05,
+      "loss": 0.5316,
+      "step": 18039
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802254160241973e-05,
+      "loss": 0.4866,
+      "step": 18040
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980222164897704e-05,
+      "loss": 0.4798,
+      "step": 18041
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802189135066442e-05,
+      "loss": 0.4661,
+      "step": 18042
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802156618510188e-05,
+      "loss": 0.4674,
+      "step": 18043
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802124099308284e-05,
+      "loss": 0.4478,
+      "step": 18044
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980209157746074e-05,
+      "loss": 0.4764,
+      "step": 18045
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980205905296757e-05,
+      "loss": 0.4644,
+      "step": 18046
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9802026525828776e-05,
+      "loss": 0.3839,
+      "step": 18047
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801993996044368e-05,
+      "loss": 0.4726,
+      "step": 18048
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801961463614357e-05,
+      "loss": 0.4619,
+      "step": 18049
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980192892853875e-05,
+      "loss": 0.5527,
+      "step": 18050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980189639081756e-05,
+      "loss": 0.5229,
+      "step": 18051
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980186385045079e-05,
+      "loss": 0.4332,
+      "step": 18052
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801831307438447e-05,
+      "loss": 0.4165,
+      "step": 18053
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980179876178055e-05,
+      "loss": 0.4395,
+      "step": 18054
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801766213477098e-05,
+      "loss": 0.5303,
+      "step": 18055
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801733662528105e-05,
+      "loss": 0.4392,
+      "step": 18056
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801701108933577e-05,
+      "loss": 0.4651,
+      "step": 18057
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801668552693524e-05,
+      "loss": 0.5359,
+      "step": 18058
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980163599380796e-05,
+      "loss": 0.4871,
+      "step": 18059
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801603432276883e-05,
+      "loss": 0.5135,
+      "step": 18060
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801570868100308e-05,
+      "loss": 0.5002,
+      "step": 18061
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801538301278246e-05,
+      "loss": 0.445,
+      "step": 18062
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.98015057318107e-05,
+      "loss": 0.4742,
+      "step": 18063
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801473159697684e-05,
+      "loss": 0.465,
+      "step": 18064
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.98014405849392e-05,
+      "loss": 0.5063,
+      "step": 18065
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801408007535268e-05,
+      "loss": 0.4111,
+      "step": 18066
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980137542748589e-05,
+      "loss": 0.4327,
+      "step": 18067
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801342844791066e-05,
+      "loss": 0.4133,
+      "step": 18068
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801310259450824e-05,
+      "loss": 0.4529,
+      "step": 18069
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980127767146516e-05,
+      "loss": 0.4411,
+      "step": 18070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801245080834078e-05,
+      "loss": 0.5562,
+      "step": 18071
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.98012124875576e-05,
+      "loss": 0.4981,
+      "step": 18072
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801179891635732e-05,
+      "loss": 0.4855,
+      "step": 18073
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801147293068475e-05,
+      "loss": 0.4492,
+      "step": 18074
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801114691855845e-05,
+      "loss": 0.4356,
+      "step": 18075
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801082087997846e-05,
+      "loss": 0.4823,
+      "step": 18076
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801049481494493e-05,
+      "loss": 0.4597,
+      "step": 18077
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9801016872345788e-05,
+      "loss": 0.4507,
+      "step": 18078
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9800984260551744e-05,
+      "loss": 0.4375,
+      "step": 18079
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9800951646112367e-05,
+      "loss": 0.5461,
+      "step": 18080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9800919029027667e-05,
+      "loss": 0.4673,
+      "step": 18081
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9800886409297654e-05,
+      "loss": 0.5016,
+      "step": 18082
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9800853786922336e-05,
+      "loss": 0.5176,
+      "step": 18083
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980082116190172e-05,
+      "loss": 0.4978,
+      "step": 18084
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980078853423582e-05,
+      "loss": 0.4515,
+      "step": 18085
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980075590392464e-05,
+      "loss": 0.4764,
+      "step": 18086
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.980072327096819e-05,
+      "loss": 0.4654,
+      "step": 18087
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980069063536648e-05,
+      "loss": 0.5196,
+      "step": 18088
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9800657997119515e-05,
+      "loss": 0.4723,
+      "step": 18089
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980062535622731e-05,
+      "loss": 0.4223,
+      "step": 18090
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980059271268987e-05,
+      "loss": 0.5365,
+      "step": 18091
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.98005600665072e-05,
+      "loss": 0.4776,
+      "step": 18092
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9800527417679317e-05,
+      "loss": 0.4998,
+      "step": 18093
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9800494766206226e-05,
+      "loss": 0.481,
+      "step": 18094
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980046211208793e-05,
+      "loss": 0.4634,
+      "step": 18095
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9800429455324452e-05,
+      "loss": 0.4196,
+      "step": 18096
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980039679591579e-05,
+      "loss": 0.3941,
+      "step": 18097
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980036413386195e-05,
+      "loss": 0.5102,
+      "step": 18098
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980033146916295e-05,
+      "loss": 0.4346,
+      "step": 18099
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9800298801818795e-05,
+      "loss": 0.4267,
+      "step": 18100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9800266131829492e-05,
+      "loss": 0.5262,
+      "step": 18101
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980023345919505e-05,
+      "loss": 0.427,
+      "step": 18102
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980020078391548e-05,
+      "loss": 0.4704,
+      "step": 18103
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980016810599079e-05,
+      "loss": 0.5006,
+      "step": 18104
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9800135425420993e-05,
+      "loss": 0.5293,
+      "step": 18105
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.980010274220609e-05,
+      "loss": 0.4852,
+      "step": 18106
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9800070056346094e-05,
+      "loss": 0.4784,
+      "step": 18107
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9800037367841013e-05,
+      "loss": 0.5064,
+      "step": 18108
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9800004676690855e-05,
+      "loss": 0.525,
+      "step": 18109
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979997198289563e-05,
+      "loss": 0.5753,
+      "step": 18110
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799939286455352e-05,
+      "loss": 0.4767,
+      "step": 18111
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979990658737002e-05,
+      "loss": 0.4449,
+      "step": 18112
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799873885639648e-05,
+      "loss": 0.4264,
+      "step": 18113
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799841181264243e-05,
+      "loss": 0.4444,
+      "step": 18114
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799808474243818e-05,
+      "loss": 0.4951,
+      "step": 18115
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799775764578374e-05,
+      "loss": 0.4754,
+      "step": 18116
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799743052267927e-05,
+      "loss": 0.3896,
+      "step": 18117
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799710337312484e-05,
+      "loss": 0.5482,
+      "step": 18118
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799677619712055e-05,
+      "loss": 0.4791,
+      "step": 18119
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799644899466644e-05,
+      "loss": 0.482,
+      "step": 18120
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799612176576268e-05,
+      "loss": 0.4485,
+      "step": 18121
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799579451040926e-05,
+      "loss": 0.5198,
+      "step": 18122
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799546722860636e-05,
+      "loss": 0.4067,
+      "step": 18123
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.97995139920354e-05,
+      "loss": 0.4879,
+      "step": 18124
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979948125856523e-05,
+      "loss": 0.4667,
+      "step": 18125
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799448522450132e-05,
+      "loss": 0.5188,
+      "step": 18126
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979941578369012e-05,
+      "loss": 0.4348,
+      "step": 18127
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.97993830422852e-05,
+      "loss": 0.4658,
+      "step": 18128
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799350298235377e-05,
+      "loss": 0.4566,
+      "step": 18129
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799317551540667e-05,
+      "loss": 0.4713,
+      "step": 18130
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799284802201074e-05,
+      "loss": 0.611,
+      "step": 18131
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979925205021661e-05,
+      "loss": 0.4386,
+      "step": 18132
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979921929558728e-05,
+      "loss": 0.4914,
+      "step": 18133
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799186538313094e-05,
+      "loss": 0.4724,
+      "step": 18134
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799153778394063e-05,
+      "loss": 0.4727,
+      "step": 18135
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799121015830194e-05,
+      "loss": 0.4632,
+      "step": 18136
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.97990882506215e-05,
+      "loss": 0.4503,
+      "step": 18137
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979905548276798e-05,
+      "loss": 0.5036,
+      "step": 18138
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9799022712269655e-05,
+      "loss": 0.5466,
+      "step": 18139
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798989939126524e-05,
+      "loss": 0.4675,
+      "step": 18140
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.97989571633386e-05,
+      "loss": 0.4874,
+      "step": 18141
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798924384905893e-05,
+      "loss": 0.5408,
+      "step": 18142
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798891603828412e-05,
+      "loss": 0.5191,
+      "step": 18143
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798858820106162e-05,
+      "loss": 0.4162,
+      "step": 18144
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798826033739157e-05,
+      "loss": 0.5005,
+      "step": 18145
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.97987932447274e-05,
+      "loss": 0.5463,
+      "step": 18146
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.97987604530709e-05,
+      "loss": 0.4859,
+      "step": 18147
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798727658769678e-05,
+      "loss": 0.4619,
+      "step": 18148
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798694861823726e-05,
+      "loss": 0.4513,
+      "step": 18149
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798662062233066e-05,
+      "loss": 0.4585,
+      "step": 18150
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798629259997698e-05,
+      "loss": 0.5311,
+      "step": 18151
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798596455117633e-05,
+      "loss": 0.5151,
+      "step": 18152
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979856364759288e-05,
+      "loss": 0.4291,
+      "step": 18153
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798530837423452e-05,
+      "loss": 0.4499,
+      "step": 18154
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798498024609355e-05,
+      "loss": 0.4478,
+      "step": 18155
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798465209150596e-05,
+      "loss": 0.4446,
+      "step": 18156
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798432391047188e-05,
+      "loss": 0.5769,
+      "step": 18157
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798399570299132e-05,
+      "loss": 0.4751,
+      "step": 18158
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798366746906447e-05,
+      "loss": 0.4055,
+      "step": 18159
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798333920869135e-05,
+      "loss": 0.4616,
+      "step": 18160
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798301092187206e-05,
+      "loss": 0.5564,
+      "step": 18161
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979826826086067e-05,
+      "loss": 0.4096,
+      "step": 18162
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798235426889537e-05,
+      "loss": 0.4157,
+      "step": 18163
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798202590273817e-05,
+      "loss": 0.4473,
+      "step": 18164
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979816975101351e-05,
+      "loss": 0.3971,
+      "step": 18165
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798136909108636e-05,
+      "loss": 0.4126,
+      "step": 18166
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798104064559198e-05,
+      "loss": 0.4779,
+      "step": 18167
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798071217365205e-05,
+      "loss": 0.4472,
+      "step": 18168
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798038367526667e-05,
+      "loss": 0.4299,
+      "step": 18169
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9798005515043594e-05,
+      "loss": 0.4586,
+      "step": 18170
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9797972659915992e-05,
+      "loss": 0.4846,
+      "step": 18171
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979793980214387e-05,
+      "loss": 0.4123,
+      "step": 18172
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979790694172724e-05,
+      "loss": 0.4076,
+      "step": 18173
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979787407866611e-05,
+      "loss": 0.4078,
+      "step": 18174
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9797841212960486e-05,
+      "loss": 0.5421,
+      "step": 18175
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.979780834461038e-05,
+      "loss": 0.4707,
+      "step": 18176
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.97977754736158e-05,
+      "loss": 0.4544,
+      "step": 18177
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9797742599976754e-05,
+      "loss": 0.4797,
+      "step": 18178
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9797709723693254e-05,
+      "loss": 0.6057,
+      "step": 18179
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.97976768447653e-05,
+      "loss": 0.4454,
+      "step": 18180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9797643963192914e-05,
+      "loss": 0.4583,
+      "step": 18181
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9797611078976097e-05,
+      "loss": 0.5514,
+      "step": 18182
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9797578192114856e-05,
+      "loss": 0.5162,
+      "step": 18183
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9797545302609203e-05,
+      "loss": 0.4756,
+      "step": 18184
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9797512410459147e-05,
+      "loss": 0.5558,
+      "step": 18185
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.97974795156647e-05,
+      "loss": 0.5141,
+      "step": 18186
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797446618225865e-05,
+      "loss": 0.4591,
+      "step": 18187
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797413718142654e-05,
+      "loss": 0.4482,
+      "step": 18188
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797380815415074e-05,
+      "loss": 0.4779,
+      "step": 18189
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797347910043135e-05,
+      "loss": 0.5526,
+      "step": 18190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797315002026847e-05,
+      "loss": 0.49,
+      "step": 18191
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797282091366217e-05,
+      "loss": 0.4497,
+      "step": 18192
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797249178061255e-05,
+      "loss": 0.4886,
+      "step": 18193
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979721626211197e-05,
+      "loss": 0.4792,
+      "step": 18194
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979718334351837e-05,
+      "loss": 0.5819,
+      "step": 18195
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797150422280467e-05,
+      "loss": 0.4131,
+      "step": 18196
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797117498398263e-05,
+      "loss": 0.4679,
+      "step": 18197
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797084571871777e-05,
+      "loss": 0.4754,
+      "step": 18198
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9797051642701007e-05,
+      "loss": 0.4671,
+      "step": 18199
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979701871088597e-05,
+      "loss": 0.4964,
+      "step": 18200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979698577642667e-05,
+      "loss": 0.4696,
+      "step": 18201
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796952839323117e-05,
+      "loss": 0.4841,
+      "step": 18202
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979691989957532e-05,
+      "loss": 0.4093,
+      "step": 18203
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796886957183292e-05,
+      "loss": 0.4642,
+      "step": 18204
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796854012147036e-05,
+      "loss": 0.4716,
+      "step": 18205
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796821064466564e-05,
+      "loss": 0.409,
+      "step": 18206
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796788114141888e-05,
+      "loss": 0.4546,
+      "step": 18207
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796755161173006e-05,
+      "loss": 0.4564,
+      "step": 18208
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979672220555994e-05,
+      "loss": 0.4539,
+      "step": 18209
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796689247302687e-05,
+      "loss": 0.4738,
+      "step": 18210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796656286401266e-05,
+      "loss": 0.4808,
+      "step": 18211
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979662332285568e-05,
+      "loss": 0.5139,
+      "step": 18212
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796590356665942e-05,
+      "loss": 0.501,
+      "step": 18213
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796557387832055e-05,
+      "loss": 0.519,
+      "step": 18214
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796524416354032e-05,
+      "loss": 0.4776,
+      "step": 18215
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796491442231885e-05,
+      "loss": 0.5623,
+      "step": 18216
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796458465465615e-05,
+      "loss": 0.4176,
+      "step": 18217
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796425486055238e-05,
+      "loss": 0.5004,
+      "step": 18218
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796392504000755e-05,
+      "loss": 0.4756,
+      "step": 18219
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796359519302185e-05,
+      "loss": 0.3967,
+      "step": 18220
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979632653195953e-05,
+      "loss": 0.4284,
+      "step": 18221
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.97962935419728e-05,
+      "loss": 0.425,
+      "step": 18222
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796260549342008e-05,
+      "loss": 0.5064,
+      "step": 18223
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796227554067157e-05,
+      "loss": 0.4951,
+      "step": 18224
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979619455614826e-05,
+      "loss": 0.4311,
+      "step": 18225
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979616155558532e-05,
+      "loss": 0.5509,
+      "step": 18226
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796128552378352e-05,
+      "loss": 0.445,
+      "step": 18227
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796095546527367e-05,
+      "loss": 0.4612,
+      "step": 18228
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796062538032365e-05,
+      "loss": 0.5565,
+      "step": 18229
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9796029526893363e-05,
+      "loss": 0.4518,
+      "step": 18230
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795996513110364e-05,
+      "loss": 0.4402,
+      "step": 18231
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795963496683383e-05,
+      "loss": 0.5013,
+      "step": 18232
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795930477612423e-05,
+      "loss": 0.4441,
+      "step": 18233
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795897455897496e-05,
+      "loss": 0.4703,
+      "step": 18234
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979586443153861e-05,
+      "loss": 0.4388,
+      "step": 18235
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795831404535776e-05,
+      "loss": 0.5025,
+      "step": 18236
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795798374889002e-05,
+      "loss": 0.503,
+      "step": 18237
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795765342598297e-05,
+      "loss": 0.481,
+      "step": 18238
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795732307663663e-05,
+      "loss": 0.6012,
+      "step": 18239
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795699270085124e-05,
+      "loss": 0.4609,
+      "step": 18240
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795666229862673e-05,
+      "loss": 0.4329,
+      "step": 18241
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795633186996327e-05,
+      "loss": 0.4197,
+      "step": 18242
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795600141486097e-05,
+      "loss": 0.4115,
+      "step": 18243
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795567093331986e-05,
+      "loss": 0.5159,
+      "step": 18244
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795534042534007e-05,
+      "loss": 0.5552,
+      "step": 18245
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795500989092167e-05,
+      "loss": 0.4399,
+      "step": 18246
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795467933006476e-05,
+      "loss": 0.4712,
+      "step": 18247
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979543487427694e-05,
+      "loss": 0.4815,
+      "step": 18248
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979540181290357e-05,
+      "loss": 0.5006,
+      "step": 18249
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795368748886382e-05,
+      "loss": 0.4879,
+      "step": 18250
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795335682225373e-05,
+      "loss": 0.5183,
+      "step": 18251
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795302612920558e-05,
+      "loss": 0.4153,
+      "step": 18252
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795269540971946e-05,
+      "loss": 0.4658,
+      "step": 18253
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979523646637954e-05,
+      "loss": 0.4099,
+      "step": 18254
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979520338914336e-05,
+      "loss": 0.4622,
+      "step": 18255
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795170309263406e-05,
+      "loss": 0.5311,
+      "step": 18256
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795137226739692e-05,
+      "loss": 0.4189,
+      "step": 18257
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979510414157222e-05,
+      "loss": 0.5079,
+      "step": 18258
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979507105376101e-05,
+      "loss": 0.503,
+      "step": 18259
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979503796330606e-05,
+      "loss": 0.4796,
+      "step": 18260
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9795004870207383e-05,
+      "loss": 0.5355,
+      "step": 18261
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979497177446499e-05,
+      "loss": 0.5675,
+      "step": 18262
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794938676078886e-05,
+      "loss": 0.4391,
+      "step": 18263
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794905575049085e-05,
+      "loss": 0.4083,
+      "step": 18264
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794872471375596e-05,
+      "loss": 0.5906,
+      "step": 18265
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979483936505842e-05,
+      "loss": 0.4777,
+      "step": 18266
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979480625609757e-05,
+      "loss": 0.4637,
+      "step": 18267
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794773144493062e-05,
+      "loss": 0.4664,
+      "step": 18268
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794740030244895e-05,
+      "loss": 0.4834,
+      "step": 18269
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794706913353084e-05,
+      "loss": 0.5118,
+      "step": 18270
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794673793817633e-05,
+      "loss": 0.5188,
+      "step": 18271
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794640671638558e-05,
+      "loss": 0.4053,
+      "step": 18272
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979460754681586e-05,
+      "loss": 0.5179,
+      "step": 18273
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979457441934955e-05,
+      "loss": 0.4751,
+      "step": 18274
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979454128923964e-05,
+      "loss": 0.5499,
+      "step": 18275
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794508156486142e-05,
+      "loss": 0.5325,
+      "step": 18276
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794475021089057e-05,
+      "loss": 0.4555,
+      "step": 18277
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.97944418830484e-05,
+      "loss": 0.4378,
+      "step": 18278
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794408742364175e-05,
+      "loss": 0.4405,
+      "step": 18279
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979437559903639e-05,
+      "loss": 0.4436,
+      "step": 18280
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.979434245306506e-05,
+      "loss": 0.445,
+      "step": 18281
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794309304450193e-05,
+      "loss": 0.5654,
+      "step": 18282
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794276153191796e-05,
+      "loss": 0.4262,
+      "step": 18283
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9794242999289873e-05,
+      "loss": 0.5401,
+      "step": 18284
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9794209842744445e-05,
+      "loss": 0.4305,
+      "step": 18285
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979417668355551e-05,
+      "loss": 0.4823,
+      "step": 18286
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979414352172308e-05,
+      "loss": 0.4396,
+      "step": 18287
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979411035724717e-05,
+      "loss": 0.5101,
+      "step": 18288
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979407719012778e-05,
+      "loss": 0.5396,
+      "step": 18289
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9794044020364923e-05,
+      "loss": 0.4462,
+      "step": 18290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979401084795861e-05,
+      "loss": 0.4768,
+      "step": 18291
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793977672908843e-05,
+      "loss": 0.496,
+      "step": 18292
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793944495215638e-05,
+      "loss": 0.4514,
+      "step": 18293
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793911314879003e-05,
+      "loss": 0.4432,
+      "step": 18294
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793878131898944e-05,
+      "loss": 0.4303,
+      "step": 18295
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793844946275475e-05,
+      "loss": 0.515,
+      "step": 18296
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793811758008595e-05,
+      "loss": 0.4721,
+      "step": 18297
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793778567098326e-05,
+      "loss": 0.5174,
+      "step": 18298
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793745373544667e-05,
+      "loss": 0.44,
+      "step": 18299
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793712177347628e-05,
+      "loss": 0.4591,
+      "step": 18300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793678978507226e-05,
+      "loss": 0.4173,
+      "step": 18301
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979364577702346e-05,
+      "loss": 0.437,
+      "step": 18302
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793612572896347e-05,
+      "loss": 0.499,
+      "step": 18303
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793579366125887e-05,
+      "loss": 0.4605,
+      "step": 18304
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793546156712095e-05,
+      "loss": 0.5458,
+      "step": 18305
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979351294465498e-05,
+      "loss": 0.4512,
+      "step": 18306
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979347972995455e-05,
+      "loss": 0.4297,
+      "step": 18307
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979344651261082e-05,
+      "loss": 0.4622,
+      "step": 18308
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793413292623786e-05,
+      "loss": 0.4686,
+      "step": 18309
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793380069993468e-05,
+      "loss": 0.4779,
+      "step": 18310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793346844719868e-05,
+      "loss": 0.4379,
+      "step": 18311
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793313616802997e-05,
+      "loss": 0.4757,
+      "step": 18312
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793280386242868e-05,
+      "loss": 0.4076,
+      "step": 18313
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793247153039488e-05,
+      "loss": 0.4642,
+      "step": 18314
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979321391719286e-05,
+      "loss": 0.4629,
+      "step": 18315
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793180678703002e-05,
+      "loss": 0.4881,
+      "step": 18316
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793147437569916e-05,
+      "loss": 0.4491,
+      "step": 18317
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793114193793614e-05,
+      "loss": 0.429,
+      "step": 18318
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793080947374107e-05,
+      "loss": 0.5282,
+      "step": 18319
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793047698311397e-05,
+      "loss": 0.4757,
+      "step": 18320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9793014446605504e-05,
+      "loss": 0.4033,
+      "step": 18321
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792981192256428e-05,
+      "loss": 0.5067,
+      "step": 18322
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979294793526418e-05,
+      "loss": 0.521,
+      "step": 18323
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792914675628766e-05,
+      "loss": 0.5111,
+      "step": 18324
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.97928814133502e-05,
+      "loss": 0.397,
+      "step": 18325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792848148428496e-05,
+      "loss": 0.4461,
+      "step": 18326
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979281488086365e-05,
+      "loss": 0.5274,
+      "step": 18327
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979278161065568e-05,
+      "loss": 0.4716,
+      "step": 18328
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792748337804592e-05,
+      "loss": 0.5103,
+      "step": 18329
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.97927150623104e-05,
+      "loss": 0.415,
+      "step": 18330
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.97926817841731e-05,
+      "loss": 0.5366,
+      "step": 18331
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792648503392713e-05,
+      "loss": 0.429,
+      "step": 18332
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792615219969247e-05,
+      "loss": 0.5049,
+      "step": 18333
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792581933902704e-05,
+      "loss": 0.4537,
+      "step": 18334
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792548645193102e-05,
+      "loss": 0.4154,
+      "step": 18335
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979251535384044e-05,
+      "loss": 0.4843,
+      "step": 18336
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792482059844736e-05,
+      "loss": 0.4227,
+      "step": 18337
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792448763205995e-05,
+      "loss": 0.5334,
+      "step": 18338
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792415463924226e-05,
+      "loss": 0.5071,
+      "step": 18339
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792382161999437e-05,
+      "loss": 0.4387,
+      "step": 18340
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792348857431637e-05,
+      "loss": 0.4894,
+      "step": 18341
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979231555022084e-05,
+      "loss": 0.4462,
+      "step": 18342
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979228224036705e-05,
+      "loss": 0.4524,
+      "step": 18343
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979224892787028e-05,
+      "loss": 0.4972,
+      "step": 18344
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792215612730533e-05,
+      "loss": 0.4967,
+      "step": 18345
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979218229494782e-05,
+      "loss": 0.4281,
+      "step": 18346
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979214897452215e-05,
+      "loss": 0.4845,
+      "step": 18347
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792115651453542e-05,
+      "loss": 0.474,
+      "step": 18348
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792082325741987e-05,
+      "loss": 0.3687,
+      "step": 18349
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792048997387507e-05,
+      "loss": 0.5835,
+      "step": 18350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792015666390107e-05,
+      "loss": 0.4868,
+      "step": 18351
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791982332749796e-05,
+      "loss": 0.4832,
+      "step": 18352
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791948996466582e-05,
+      "loss": 0.5166,
+      "step": 18353
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791915657540476e-05,
+      "loss": 0.5411,
+      "step": 18354
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791882315971485e-05,
+      "loss": 0.5092,
+      "step": 18355
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791848971759623e-05,
+      "loss": 0.5429,
+      "step": 18356
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979181562490489e-05,
+      "loss": 0.4502,
+      "step": 18357
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791782275407304e-05,
+      "loss": 0.4822,
+      "step": 18358
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791748923266868e-05,
+      "loss": 0.4878,
+      "step": 18359
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791715568483594e-05,
+      "loss": 0.5154,
+      "step": 18360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791682211057496e-05,
+      "loss": 0.4815,
+      "step": 18361
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979164885098857e-05,
+      "loss": 0.5122,
+      "step": 18362
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791615488276835e-05,
+      "loss": 0.5288,
+      "step": 18363
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791582122922295e-05,
+      "loss": 0.4043,
+      "step": 18364
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979154875492496e-05,
+      "loss": 0.5258,
+      "step": 18365
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791515384284845e-05,
+      "loss": 0.5529,
+      "step": 18366
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979148201100195e-05,
+      "loss": 0.4981,
+      "step": 18367
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979144863507629e-05,
+      "loss": 0.4939,
+      "step": 18368
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791415256507874e-05,
+      "loss": 0.4922,
+      "step": 18369
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979138187529671e-05,
+      "loss": 0.5839,
+      "step": 18370
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.97913484914428e-05,
+      "loss": 0.5107,
+      "step": 18371
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791315104946165e-05,
+      "loss": 0.493,
+      "step": 18372
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791281715806808e-05,
+      "loss": 0.5452,
+      "step": 18373
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791248324024735e-05,
+      "loss": 0.5033,
+      "step": 18374
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979121492959996e-05,
+      "loss": 0.4902,
+      "step": 18375
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979118153253249e-05,
+      "loss": 0.5299,
+      "step": 18376
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791148132822334e-05,
+      "loss": 0.4813,
+      "step": 18377
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791114730469504e-05,
+      "loss": 0.4226,
+      "step": 18378
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791081325474005e-05,
+      "loss": 0.4464,
+      "step": 18379
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791047917835844e-05,
+      "loss": 0.4181,
+      "step": 18380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9791014507555036e-05,
+      "loss": 0.4185,
+      "step": 18381
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9790981094631587e-05,
+      "loss": 0.573,
+      "step": 18382
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.979094767906551e-05,
+      "loss": 0.4739,
+      "step": 18383
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790914260856805e-05,
+      "loss": 0.4738,
+      "step": 18384
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979088084000549e-05,
+      "loss": 0.443,
+      "step": 18385
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979084741651157e-05,
+      "loss": 0.454,
+      "step": 18386
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979081399037505e-05,
+      "loss": 0.4988,
+      "step": 18387
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979078056159595e-05,
+      "loss": 0.4672,
+      "step": 18388
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790747130174272e-05,
+      "loss": 0.4634,
+      "step": 18389
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790713696110022e-05,
+      "loss": 0.4277,
+      "step": 18390
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790680259403215e-05,
+      "loss": 0.5237,
+      "step": 18391
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979064682005386e-05,
+      "loss": 0.4806,
+      "step": 18392
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790613378061957e-05,
+      "loss": 0.5728,
+      "step": 18393
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979057993342753e-05,
+      "loss": 0.428,
+      "step": 18394
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979054648615057e-05,
+      "loss": 0.535,
+      "step": 18395
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.97905130362311e-05,
+      "loss": 0.4672,
+      "step": 18396
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790479583669125e-05,
+      "loss": 0.4745,
+      "step": 18397
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790446128464658e-05,
+      "loss": 0.4673,
+      "step": 18398
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790412670617698e-05,
+      "loss": 0.5028,
+      "step": 18399
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979037921012826e-05,
+      "loss": 0.3976,
+      "step": 18400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790345746996356e-05,
+      "loss": 0.5386,
+      "step": 18401
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979031228122199e-05,
+      "loss": 0.4619,
+      "step": 18402
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790278812805175e-05,
+      "loss": 0.5523,
+      "step": 18403
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790245341745917e-05,
+      "loss": 0.4088,
+      "step": 18404
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790211868044225e-05,
+      "loss": 0.4709,
+      "step": 18405
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979017839170011e-05,
+      "loss": 0.4366,
+      "step": 18406
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790144912713582e-05,
+      "loss": 0.4301,
+      "step": 18407
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790111431084644e-05,
+      "loss": 0.4599,
+      "step": 18408
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790077946813312e-05,
+      "loss": 0.4407,
+      "step": 18409
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.979004445989959e-05,
+      "loss": 0.5098,
+      "step": 18410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9790010970343488e-05,
+      "loss": 0.439,
+      "step": 18411
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789977478145022e-05,
+      "loss": 0.5107,
+      "step": 18412
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789943983304193e-05,
+      "loss": 0.4337,
+      "step": 18413
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789910485821008e-05,
+      "loss": 0.5023,
+      "step": 18414
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789876985695486e-05,
+      "loss": 0.4107,
+      "step": 18415
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789843482927628e-05,
+      "loss": 0.4362,
+      "step": 18416
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789809977517447e-05,
+      "loss": 0.4135,
+      "step": 18417
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789776469464947e-05,
+      "loss": 0.4513,
+      "step": 18418
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789742958770145e-05,
+      "loss": 0.4966,
+      "step": 18419
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789709445433044e-05,
+      "loss": 0.4573,
+      "step": 18420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789675929453654e-05,
+      "loss": 0.4278,
+      "step": 18421
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789642410831985e-05,
+      "loss": 0.4776,
+      "step": 18422
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789608889568045e-05,
+      "loss": 0.4902,
+      "step": 18423
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789575365661842e-05,
+      "loss": 0.5321,
+      "step": 18424
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789541839113392e-05,
+      "loss": 0.4976,
+      "step": 18425
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789508309922697e-05,
+      "loss": 0.4443,
+      "step": 18426
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789474778089764e-05,
+      "loss": 0.476,
+      "step": 18427
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978944124361461e-05,
+      "loss": 0.5585,
+      "step": 18428
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978940770649724e-05,
+      "loss": 0.453,
+      "step": 18429
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978937416673766e-05,
+      "loss": 0.4848,
+      "step": 18430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789340624335886e-05,
+      "loss": 0.5867,
+      "step": 18431
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789307079291923e-05,
+      "loss": 0.4863,
+      "step": 18432
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978927353160578e-05,
+      "loss": 0.4618,
+      "step": 18433
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789239981277465e-05,
+      "loss": 0.4655,
+      "step": 18434
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789206428306987e-05,
+      "loss": 0.501,
+      "step": 18435
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789172872694356e-05,
+      "loss": 0.4489,
+      "step": 18436
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789139314439586e-05,
+      "loss": 0.4572,
+      "step": 18437
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978910575354268e-05,
+      "loss": 0.5885,
+      "step": 18438
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789072190003646e-05,
+      "loss": 0.5075,
+      "step": 18439
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789038623822497e-05,
+      "loss": 0.5389,
+      "step": 18440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9789005054999242e-05,
+      "loss": 0.4957,
+      "step": 18441
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788971483533888e-05,
+      "loss": 0.4772,
+      "step": 18442
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788937909426442e-05,
+      "loss": 0.4092,
+      "step": 18443
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978890433267692e-05,
+      "loss": 0.4112,
+      "step": 18444
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788870753285327e-05,
+      "loss": 0.4586,
+      "step": 18445
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978883717125167e-05,
+      "loss": 0.4575,
+      "step": 18446
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788803586575958e-05,
+      "loss": 0.4961,
+      "step": 18447
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788769999258206e-05,
+      "loss": 0.4791,
+      "step": 18448
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978873640929842e-05,
+      "loss": 0.4107,
+      "step": 18449
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788702816696602e-05,
+      "loss": 0.4678,
+      "step": 18450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978866922145277e-05,
+      "loss": 0.4534,
+      "step": 18451
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788635623566934e-05,
+      "loss": 0.4474,
+      "step": 18452
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788602023039097e-05,
+      "loss": 0.4469,
+      "step": 18453
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788568419869273e-05,
+      "loss": 0.4657,
+      "step": 18454
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978853481405747e-05,
+      "loss": 0.3756,
+      "step": 18455
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978850120560369e-05,
+      "loss": 0.4536,
+      "step": 18456
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788467594507948e-05,
+      "loss": 0.5499,
+      "step": 18457
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788433980770256e-05,
+      "loss": 0.5878,
+      "step": 18458
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788400364390618e-05,
+      "loss": 0.449,
+      "step": 18459
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788366745369046e-05,
+      "loss": 0.5169,
+      "step": 18460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978833312370555e-05,
+      "loss": 0.4414,
+      "step": 18461
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788299499400135e-05,
+      "loss": 0.4673,
+      "step": 18462
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788265872452812e-05,
+      "loss": 0.5471,
+      "step": 18463
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978823224286359e-05,
+      "loss": 0.451,
+      "step": 18464
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788198610632478e-05,
+      "loss": 0.4539,
+      "step": 18465
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788164975759488e-05,
+      "loss": 0.4698,
+      "step": 18466
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788131338244626e-05,
+      "loss": 0.5233,
+      "step": 18467
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.97880976980879e-05,
+      "loss": 0.4218,
+      "step": 18468
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978806405528932e-05,
+      "loss": 0.4634,
+      "step": 18469
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9788030409848895e-05,
+      "loss": 0.4631,
+      "step": 18470
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9787996761766637e-05,
+      "loss": 0.6288,
+      "step": 18471
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9787963111042553e-05,
+      "loss": 0.5261,
+      "step": 18472
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9787929457676652e-05,
+      "loss": 0.4299,
+      "step": 18473
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9787895801668943e-05,
+      "loss": 0.509,
+      "step": 18474
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9787862143019433e-05,
+      "loss": 0.3909,
+      "step": 18475
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9787828481728136e-05,
+      "loss": 0.5448,
+      "step": 18476
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9787794817795058e-05,
+      "loss": 0.4489,
+      "step": 18477
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978776115122021e-05,
+      "loss": 0.496,
+      "step": 18478
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9787727482003598e-05,
+      "loss": 0.4847,
+      "step": 18479
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978769381014523e-05,
+      "loss": 0.5245,
+      "step": 18480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.978766013564512e-05,
+      "loss": 0.4606,
+      "step": 18481
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9787626458503276e-05,
+      "loss": 0.4664,
+      "step": 18482
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.97875927787197e-05,
+      "loss": 0.4024,
+      "step": 18483
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9787559096294413e-05,
+      "loss": 0.5168,
+      "step": 18484
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978752541122742e-05,
+      "loss": 0.5396,
+      "step": 18485
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978749172351872e-05,
+      "loss": 0.4566,
+      "step": 18486
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9787458033168338e-05,
+      "loss": 0.6664,
+      "step": 18487
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978742434017627e-05,
+      "loss": 0.4302,
+      "step": 18488
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9787390644542534e-05,
+      "loss": 0.5062,
+      "step": 18489
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9787356946267132e-05,
+      "loss": 0.4811,
+      "step": 18490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9787323245350082e-05,
+      "loss": 0.4529,
+      "step": 18491
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9787289541791383e-05,
+      "loss": 0.3978,
+      "step": 18492
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978725583559105e-05,
+      "loss": 0.5458,
+      "step": 18493
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978722212674909e-05,
+      "loss": 0.4616,
+      "step": 18494
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9787188415265515e-05,
+      "loss": 0.4395,
+      "step": 18495
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978715470114033e-05,
+      "loss": 0.4601,
+      "step": 18496
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978712098437355e-05,
+      "loss": 0.5805,
+      "step": 18497
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9787087264965174e-05,
+      "loss": 0.4655,
+      "step": 18498
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978705354291522e-05,
+      "loss": 0.3572,
+      "step": 18499
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9787019818223697e-05,
+      "loss": 0.4467,
+      "step": 18500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786986090890613e-05,
+      "loss": 0.5549,
+      "step": 18501
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786952360915973e-05,
+      "loss": 0.3936,
+      "step": 18502
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786918628299788e-05,
+      "loss": 0.5101,
+      "step": 18503
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786884893042067e-05,
+      "loss": 0.5476,
+      "step": 18504
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786851155142823e-05,
+      "loss": 0.3835,
+      "step": 18505
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978681741460206e-05,
+      "loss": 0.5096,
+      "step": 18506
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978678367141979e-05,
+      "loss": 0.4625,
+      "step": 18507
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978674992559602e-05,
+      "loss": 0.4394,
+      "step": 18508
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786716177130763e-05,
+      "loss": 0.5226,
+      "step": 18509
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786682426024025e-05,
+      "loss": 0.4055,
+      "step": 18510
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786648672275816e-05,
+      "loss": 0.521,
+      "step": 18511
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978661491588614e-05,
+      "loss": 0.477,
+      "step": 18512
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786581156855017e-05,
+      "loss": 0.4235,
+      "step": 18513
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978654739518245e-05,
+      "loss": 0.4169,
+      "step": 18514
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978651363086844e-05,
+      "loss": 0.4109,
+      "step": 18515
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786479863913013e-05,
+      "loss": 0.4341,
+      "step": 18516
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786446094316167e-05,
+      "loss": 0.3666,
+      "step": 18517
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978641232207791e-05,
+      "loss": 0.5023,
+      "step": 18518
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978637854719826e-05,
+      "loss": 0.5951,
+      "step": 18519
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786344769677217e-05,
+      "loss": 0.417,
+      "step": 18520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786310989514797e-05,
+      "loss": 0.5092,
+      "step": 18521
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786277206711002e-05,
+      "loss": 0.4419,
+      "step": 18522
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786243421265844e-05,
+      "loss": 0.5138,
+      "step": 18523
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978620963317934e-05,
+      "loss": 0.5375,
+      "step": 18524
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786175842451487e-05,
+      "loss": 0.5234,
+      "step": 18525
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786142049082298e-05,
+      "loss": 0.4913,
+      "step": 18526
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978610825307179e-05,
+      "loss": 0.5204,
+      "step": 18527
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978607445441996e-05,
+      "loss": 0.4768,
+      "step": 18528
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786040653126824e-05,
+      "loss": 0.4268,
+      "step": 18529
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978600684919239e-05,
+      "loss": 0.5718,
+      "step": 18530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978597304261667e-05,
+      "loss": 0.4228,
+      "step": 18531
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785939233399664e-05,
+      "loss": 0.4329,
+      "step": 18532
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785905421541393e-05,
+      "loss": 0.4948,
+      "step": 18533
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978587160704186e-05,
+      "loss": 0.4993,
+      "step": 18534
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978583778990107e-05,
+      "loss": 0.5858,
+      "step": 18535
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978580397011904e-05,
+      "loss": 0.5097,
+      "step": 18536
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785770147695774e-05,
+      "loss": 0.4534,
+      "step": 18537
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785736322631283e-05,
+      "loss": 0.4212,
+      "step": 18538
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785702494925578e-05,
+      "loss": 0.5535,
+      "step": 18539
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785668664578664e-05,
+      "loss": 0.5787,
+      "step": 18540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785634831590553e-05,
+      "loss": 0.4947,
+      "step": 18541
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785600995961257e-05,
+      "loss": 0.4161,
+      "step": 18542
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785567157690778e-05,
+      "loss": 0.4684,
+      "step": 18543
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785533316779127e-05,
+      "loss": 0.408,
+      "step": 18544
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785499473226316e-05,
+      "loss": 0.4645,
+      "step": 18545
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785465627032355e-05,
+      "loss": 0.5227,
+      "step": 18546
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978543177819725e-05,
+      "loss": 0.4989,
+      "step": 18547
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978539792672101e-05,
+      "loss": 0.4204,
+      "step": 18548
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785364072603646e-05,
+      "loss": 0.5498,
+      "step": 18549
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785330215845168e-05,
+      "loss": 0.5782,
+      "step": 18550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785296356445583e-05,
+      "loss": 0.3599,
+      "step": 18551
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785262494404903e-05,
+      "loss": 0.5957,
+      "step": 18552
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978522862972313e-05,
+      "loss": 0.4211,
+      "step": 18553
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978519476240028e-05,
+      "loss": 0.4647,
+      "step": 18554
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978516089243636e-05,
+      "loss": 0.4394,
+      "step": 18555
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978512701983138e-05,
+      "loss": 0.5389,
+      "step": 18556
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785093144585348e-05,
+      "loss": 0.4227,
+      "step": 18557
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785059266698274e-05,
+      "loss": 0.5409,
+      "step": 18558
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9785025386170167e-05,
+      "loss": 0.4422,
+      "step": 18559
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784991503001036e-05,
+      "loss": 0.5226,
+      "step": 18560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978495761719089e-05,
+      "loss": 0.3952,
+      "step": 18561
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784923728739738e-05,
+      "loss": 0.4155,
+      "step": 18562
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978488983764759e-05,
+      "loss": 0.4002,
+      "step": 18563
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784855943914453e-05,
+      "loss": 0.5103,
+      "step": 18564
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978482204754034e-05,
+      "loss": 0.4733,
+      "step": 18565
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784788148525257e-05,
+      "loss": 0.5067,
+      "step": 18566
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784754246869208e-05,
+      "loss": 0.4373,
+      "step": 18567
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784720342572218e-05,
+      "loss": 0.4982,
+      "step": 18568
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784686435634278e-05,
+      "loss": 0.371,
+      "step": 18569
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978465252605541e-05,
+      "loss": 0.4788,
+      "step": 18570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784618613835618e-05,
+      "loss": 0.3993,
+      "step": 18571
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.978458469897491e-05,
+      "loss": 0.454,
+      "step": 18572
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.97845507814733e-05,
+      "loss": 0.4477,
+      "step": 18573
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784516861330793e-05,
+      "loss": 0.4264,
+      "step": 18574
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784482938547397e-05,
+      "loss": 0.4252,
+      "step": 18575
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784449013123123e-05,
+      "loss": 0.4374,
+      "step": 18576
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784415085057982e-05,
+      "loss": 0.6193,
+      "step": 18577
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784381154351983e-05,
+      "loss": 0.4577,
+      "step": 18578
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784347221005128e-05,
+      "loss": 0.4155,
+      "step": 18579
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9784313285017437e-05,
+      "loss": 0.4897,
+      "step": 18580
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9784279346388915e-05,
+      "loss": 0.5254,
+      "step": 18581
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9784245405119568e-05,
+      "loss": 0.4772,
+      "step": 18582
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978421146120941e-05,
+      "loss": 0.5183,
+      "step": 18583
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9784177514658445e-05,
+      "loss": 0.5026,
+      "step": 18584
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9784143565466683e-05,
+      "loss": 0.4515,
+      "step": 18585
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978410961363414e-05,
+      "loss": 0.4929,
+      "step": 18586
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9784075659160817e-05,
+      "loss": 0.5089,
+      "step": 18587
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9784041702046725e-05,
+      "loss": 0.4658,
+      "step": 18588
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9784007742291877e-05,
+      "loss": 0.4917,
+      "step": 18589
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978397377989628e-05,
+      "loss": 0.5162,
+      "step": 18590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978393981485994e-05,
+      "loss": 0.5176,
+      "step": 18591
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783905847182868e-05,
+      "loss": 0.5487,
+      "step": 18592
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978387187686508e-05,
+      "loss": 0.4189,
+      "step": 18593
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783837903906574e-05,
+      "loss": 0.4614,
+      "step": 18594
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783803928307365e-05,
+      "loss": 0.443,
+      "step": 18595
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783769950067465e-05,
+      "loss": 0.5301,
+      "step": 18596
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783735969186876e-05,
+      "loss": 0.5322,
+      "step": 18597
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978370198566561e-05,
+      "loss": 0.4077,
+      "step": 18598
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978366799950368e-05,
+      "loss": 0.49,
+      "step": 18599
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783634010701094e-05,
+      "loss": 0.4907,
+      "step": 18600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783600019257855e-05,
+      "loss": 0.5516,
+      "step": 18601
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978356602517398e-05,
+      "loss": 0.4135,
+      "step": 18602
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783532028449474e-05,
+      "loss": 0.4889,
+      "step": 18603
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783498029084346e-05,
+      "loss": 0.4334,
+      "step": 18604
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783464027078606e-05,
+      "loss": 0.4508,
+      "step": 18605
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783430022432265e-05,
+      "loss": 0.4924,
+      "step": 18606
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978339601514533e-05,
+      "loss": 0.4711,
+      "step": 18607
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783362005217808e-05,
+      "loss": 0.4361,
+      "step": 18608
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783327992649713e-05,
+      "loss": 0.4618,
+      "step": 18609
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783293977441054e-05,
+      "loss": 0.4129,
+      "step": 18610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783259959591837e-05,
+      "loss": 0.4432,
+      "step": 18611
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978322593910207e-05,
+      "loss": 0.5075,
+      "step": 18612
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783191915971765e-05,
+      "loss": 0.4472,
+      "step": 18613
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783157890200934e-05,
+      "loss": 0.4701,
+      "step": 18614
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783123861789583e-05,
+      "loss": 0.4338,
+      "step": 18615
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9783089830737718e-05,
+      "loss": 0.5436,
+      "step": 18616
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978305579704535e-05,
+      "loss": 0.4898,
+      "step": 18617
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978302176071249e-05,
+      "loss": 0.548,
+      "step": 18618
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782987721739154e-05,
+      "loss": 0.4212,
+      "step": 18619
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782953680125337e-05,
+      "loss": 0.4614,
+      "step": 18620
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782919635871057e-05,
+      "loss": 0.4718,
+      "step": 18621
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978288558897632e-05,
+      "loss": 0.3915,
+      "step": 18622
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782851539441138e-05,
+      "loss": 0.4878,
+      "step": 18623
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782817487265514e-05,
+      "loss": 0.3809,
+      "step": 18624
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978278343244947e-05,
+      "loss": 0.4983,
+      "step": 18625
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782749374993002e-05,
+      "loss": 0.4551,
+      "step": 18626
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782715314896123e-05,
+      "loss": 0.3831,
+      "step": 18627
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978268125215885e-05,
+      "loss": 0.4857,
+      "step": 18628
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782647186781178e-05,
+      "loss": 0.4895,
+      "step": 18629
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782613118763128e-05,
+      "loss": 0.4691,
+      "step": 18630
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782579048104704e-05,
+      "loss": 0.4518,
+      "step": 18631
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782544974805915e-05,
+      "loss": 0.4431,
+      "step": 18632
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782510898866772e-05,
+      "loss": 0.5589,
+      "step": 18633
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782476820287286e-05,
+      "loss": 0.3919,
+      "step": 18634
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978244273906746e-05,
+      "loss": 0.4582,
+      "step": 18635
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782408655207307e-05,
+      "loss": 0.421,
+      "step": 18636
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978237456870684e-05,
+      "loss": 0.4879,
+      "step": 18637
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978234047956606e-05,
+      "loss": 0.5351,
+      "step": 18638
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782306387784983e-05,
+      "loss": 0.5252,
+      "step": 18639
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782272293363616e-05,
+      "loss": 0.4576,
+      "step": 18640
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978223819630197e-05,
+      "loss": 0.5029,
+      "step": 18641
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782204096600048e-05,
+      "loss": 0.4208,
+      "step": 18642
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782169994257865e-05,
+      "loss": 0.46,
+      "step": 18643
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978213588927543e-05,
+      "loss": 0.4804,
+      "step": 18644
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978210178165275e-05,
+      "loss": 0.556,
+      "step": 18645
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978206767138983e-05,
+      "loss": 0.4976,
+      "step": 18646
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9782033558486693e-05,
+      "loss": 0.5055,
+      "step": 18647
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781999442943335e-05,
+      "loss": 0.447,
+      "step": 18648
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978196532475977e-05,
+      "loss": 0.477,
+      "step": 18649
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781931203936005e-05,
+      "loss": 0.4725,
+      "step": 18650
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781897080472053e-05,
+      "loss": 0.4396,
+      "step": 18651
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978186295436792e-05,
+      "loss": 0.485,
+      "step": 18652
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781828825623618e-05,
+      "loss": 0.5324,
+      "step": 18653
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781794694239155e-05,
+      "loss": 0.4,
+      "step": 18654
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978176056021454e-05,
+      "loss": 0.3989,
+      "step": 18655
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978172642354978e-05,
+      "loss": 0.4438,
+      "step": 18656
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781692284244886e-05,
+      "loss": 0.5585,
+      "step": 18657
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978165814229987e-05,
+      "loss": 0.444,
+      "step": 18658
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978162399771474e-05,
+      "loss": 0.4811,
+      "step": 18659
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.97815898504895e-05,
+      "loss": 0.4449,
+      "step": 18660
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781555700624167e-05,
+      "loss": 0.4545,
+      "step": 18661
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978152154811874e-05,
+      "loss": 0.4819,
+      "step": 18662
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978148739297324e-05,
+      "loss": 0.5295,
+      "step": 18663
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978145323518767e-05,
+      "loss": 0.4423,
+      "step": 18664
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978141907476204e-05,
+      "loss": 0.5417,
+      "step": 18665
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978138491169636e-05,
+      "loss": 0.4133,
+      "step": 18666
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781350745990634e-05,
+      "loss": 0.4379,
+      "step": 18667
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781316577644883e-05,
+      "loss": 0.5104,
+      "step": 18668
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781282406659102e-05,
+      "loss": 0.4941,
+      "step": 18669
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781248233033313e-05,
+      "loss": 0.4463,
+      "step": 18670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781214056767516e-05,
+      "loss": 0.5037,
+      "step": 18671
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781179877861723e-05,
+      "loss": 0.434,
+      "step": 18672
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781145696315946e-05,
+      "loss": 0.426,
+      "step": 18673
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.978111151213019e-05,
+      "loss": 0.4399,
+      "step": 18674
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781077325304467e-05,
+      "loss": 0.6336,
+      "step": 18675
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781043135838786e-05,
+      "loss": 0.5515,
+      "step": 18676
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9781008943733157e-05,
+      "loss": 0.4483,
+      "step": 18677
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9780974748987587e-05,
+      "loss": 0.6336,
+      "step": 18678
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9780940551602087e-05,
+      "loss": 0.4777,
+      "step": 18679
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780906351576663e-05,
+      "loss": 0.4215,
+      "step": 18680
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.978087214891133e-05,
+      "loss": 0.4128,
+      "step": 18681
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780837943606087e-05,
+      "loss": 0.4378,
+      "step": 18682
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.978080373566096e-05,
+      "loss": 0.4431,
+      "step": 18683
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.978076952507594e-05,
+      "loss": 0.4887,
+      "step": 18684
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780735311851047e-05,
+      "loss": 0.4753,
+      "step": 18685
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.978070109598629e-05,
+      "loss": 0.4232,
+      "step": 18686
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780666877481677e-05,
+      "loss": 0.4165,
+      "step": 18687
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.978063265633721e-05,
+      "loss": 0.4282,
+      "step": 18688
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780598432552912e-05,
+      "loss": 0.4074,
+      "step": 18689
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.978056420612878e-05,
+      "loss": 0.4511,
+      "step": 18690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780529977064828e-05,
+      "loss": 0.4003,
+      "step": 18691
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780495745361066e-05,
+      "loss": 0.4615,
+      "step": 18692
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780461511017503e-05,
+      "loss": 0.5058,
+      "step": 18693
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780427274034147e-05,
+      "loss": 0.4775,
+      "step": 18694
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.978039303441101e-05,
+      "loss": 0.5135,
+      "step": 18695
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.97803587921481e-05,
+      "loss": 0.4718,
+      "step": 18696
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780324547245422e-05,
+      "loss": 0.4753,
+      "step": 18697
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.978029029970299e-05,
+      "loss": 0.4523,
+      "step": 18698
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780256049520813e-05,
+      "loss": 0.4637,
+      "step": 18699
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780221796698897e-05,
+      "loss": 0.536,
+      "step": 18700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780187541237253e-05,
+      "loss": 0.3864,
+      "step": 18701
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780153283135893e-05,
+      "loss": 0.4327,
+      "step": 18702
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780119022394825e-05,
+      "loss": 0.4666,
+      "step": 18703
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780084759014054e-05,
+      "loss": 0.5246,
+      "step": 18704
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9780050492993596e-05,
+      "loss": 0.4461,
+      "step": 18705
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.978001622433345e-05,
+      "loss": 0.4118,
+      "step": 18706
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779981953033637e-05,
+      "loss": 0.4424,
+      "step": 18707
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977994767909416e-05,
+      "loss": 0.5078,
+      "step": 18708
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977991340251503e-05,
+      "loss": 0.5431,
+      "step": 18709
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779879123296256e-05,
+      "loss": 0.5746,
+      "step": 18710
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779844841437844e-05,
+      "loss": 0.5202,
+      "step": 18711
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977981055693981e-05,
+      "loss": 0.4652,
+      "step": 18712
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977977626980216e-05,
+      "loss": 0.5481,
+      "step": 18713
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.97797419800249e-05,
+      "loss": 0.4689,
+      "step": 18714
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977970768760804e-05,
+      "loss": 0.4935,
+      "step": 18715
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779673392551598e-05,
+      "loss": 0.4366,
+      "step": 18716
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977963909485557e-05,
+      "loss": 0.4275,
+      "step": 18717
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779604794519975e-05,
+      "loss": 0.5762,
+      "step": 18718
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779570491544818e-05,
+      "loss": 0.3917,
+      "step": 18719
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779536185930106e-05,
+      "loss": 0.4444,
+      "step": 18720
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779501877675854e-05,
+      "loss": 0.5107,
+      "step": 18721
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977946756678207e-05,
+      "loss": 0.4481,
+      "step": 18722
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779433253248763e-05,
+      "loss": 0.5409,
+      "step": 18723
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977939893707594e-05,
+      "loss": 0.546,
+      "step": 18724
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977936461826361e-05,
+      "loss": 0.4295,
+      "step": 18725
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779330296811784e-05,
+      "loss": 0.4613,
+      "step": 18726
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779295972720472e-05,
+      "loss": 0.5027,
+      "step": 18727
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779261645989683e-05,
+      "loss": 0.4056,
+      "step": 18728
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779227316619424e-05,
+      "loss": 0.5028,
+      "step": 18729
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977919298460971e-05,
+      "loss": 0.5109,
+      "step": 18730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779158649960538e-05,
+      "loss": 0.4366,
+      "step": 18731
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977912431267193e-05,
+      "loss": 0.4895,
+      "step": 18732
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779089972743894e-05,
+      "loss": 0.4736,
+      "step": 18733
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977905563017643e-05,
+      "loss": 0.5307,
+      "step": 18734
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9779021284969557e-05,
+      "loss": 0.4885,
+      "step": 18735
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977898693712328e-05,
+      "loss": 0.4478,
+      "step": 18736
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778952586637604e-05,
+      "loss": 0.3966,
+      "step": 18737
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977891823351255e-05,
+      "loss": 0.5881,
+      "step": 18738
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778883877748115e-05,
+      "loss": 0.5357,
+      "step": 18739
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778849519344318e-05,
+      "loss": 0.4527,
+      "step": 18740
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977881515830116e-05,
+      "loss": 0.5686,
+      "step": 18741
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778780794618657e-05,
+      "loss": 0.5974,
+      "step": 18742
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778746428296813e-05,
+      "loss": 0.56,
+      "step": 18743
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778712059335638e-05,
+      "loss": 0.5159,
+      "step": 18744
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778677687735145e-05,
+      "loss": 0.5219,
+      "step": 18745
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778643313495342e-05,
+      "loss": 0.5274,
+      "step": 18746
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977860893661624e-05,
+      "loss": 0.549,
+      "step": 18747
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977857455709784e-05,
+      "loss": 0.4512,
+      "step": 18748
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977854017494016e-05,
+      "loss": 0.52,
+      "step": 18749
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778505790143207e-05,
+      "loss": 0.4401,
+      "step": 18750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977847140270699e-05,
+      "loss": 0.501,
+      "step": 18751
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778437012631512e-05,
+      "loss": 0.457,
+      "step": 18752
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778402619916796e-05,
+      "loss": 0.5021,
+      "step": 18753
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977836822456284e-05,
+      "loss": 0.5207,
+      "step": 18754
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778333826569655e-05,
+      "loss": 0.4394,
+      "step": 18755
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778299425937254e-05,
+      "loss": 0.4428,
+      "step": 18756
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778265022665644e-05,
+      "loss": 0.4925,
+      "step": 18757
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778230616754835e-05,
+      "loss": 0.4676,
+      "step": 18758
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778196208204834e-05,
+      "loss": 0.4912,
+      "step": 18759
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778161797015655e-05,
+      "loss": 0.5282,
+      "step": 18760
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778127383187304e-05,
+      "loss": 0.4721,
+      "step": 18761
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9778092966719788e-05,
+      "loss": 0.5889,
+      "step": 18762
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977805854761312e-05,
+      "loss": 0.5004,
+      "step": 18763
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977802412586731e-05,
+      "loss": 0.4936,
+      "step": 18764
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9777989701482366e-05,
+      "loss": 0.4797,
+      "step": 18765
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9777955274458295e-05,
+      "loss": 0.5869,
+      "step": 18766
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9777920844795107e-05,
+      "loss": 0.589,
+      "step": 18767
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9777886412492815e-05,
+      "loss": 0.5177,
+      "step": 18768
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9777851977551426e-05,
+      "loss": 0.5683,
+      "step": 18769
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9777817539970946e-05,
+      "loss": 0.4576,
+      "step": 18770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977778309975139e-05,
+      "loss": 0.5044,
+      "step": 18771
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9777748656892763e-05,
+      "loss": 0.5076,
+      "step": 18772
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9777714211395077e-05,
+      "loss": 0.461,
+      "step": 18773
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977767976325834e-05,
+      "loss": 0.5075,
+      "step": 18774
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977764531248256e-05,
+      "loss": 0.493,
+      "step": 18775
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9777610859067753e-05,
+      "loss": 0.44,
+      "step": 18776
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.977757640301392e-05,
+      "loss": 0.4586,
+      "step": 18777
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777541944321072e-05,
+      "loss": 0.4772,
+      "step": 18778
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977750748298922e-05,
+      "loss": 0.4648,
+      "step": 18779
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777473019018374e-05,
+      "loss": 0.4393,
+      "step": 18780
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977743855240854e-05,
+      "loss": 0.5238,
+      "step": 18781
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777404083159732e-05,
+      "loss": 0.3869,
+      "step": 18782
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777369611271958e-05,
+      "loss": 0.5009,
+      "step": 18783
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777335136745223e-05,
+      "loss": 0.4543,
+      "step": 18784
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777300659579543e-05,
+      "loss": 0.4443,
+      "step": 18785
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977726617977492e-05,
+      "loss": 0.5112,
+      "step": 18786
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777231697331372e-05,
+      "loss": 0.4665,
+      "step": 18787
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.97771972122489e-05,
+      "loss": 0.5132,
+      "step": 18788
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777162724527518e-05,
+      "loss": 0.5164,
+      "step": 18789
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777128234167233e-05,
+      "loss": 0.5365,
+      "step": 18790
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777093741168057e-05,
+      "loss": 0.5169,
+      "step": 18791
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777059245529995e-05,
+      "loss": 0.6142,
+      "step": 18792
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9777024747253062e-05,
+      "loss": 0.5417,
+      "step": 18793
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776990246337262e-05,
+      "loss": 0.4999,
+      "step": 18794
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977695574278261e-05,
+      "loss": 0.4768,
+      "step": 18795
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776921236589107e-05,
+      "loss": 0.5079,
+      "step": 18796
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977688672775677e-05,
+      "loss": 0.5693,
+      "step": 18797
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776852216285607e-05,
+      "loss": 0.5449,
+      "step": 18798
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776817702175622e-05,
+      "loss": 0.4358,
+      "step": 18799
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776783185426832e-05,
+      "loss": 0.4815,
+      "step": 18800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977674866603924e-05,
+      "loss": 0.436,
+      "step": 18801
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977671414401286e-05,
+      "loss": 0.4126,
+      "step": 18802
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776679619347695e-05,
+      "loss": 0.5243,
+      "step": 18803
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776645092043766e-05,
+      "loss": 0.4407,
+      "step": 18804
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977661056210107e-05,
+      "loss": 0.4912,
+      "step": 18805
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977657602951962e-05,
+      "loss": 0.4503,
+      "step": 18806
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776541494299427e-05,
+      "loss": 0.4953,
+      "step": 18807
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776506956440503e-05,
+      "loss": 0.4545,
+      "step": 18808
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977647241594285e-05,
+      "loss": 0.5231,
+      "step": 18809
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776437872806483e-05,
+      "loss": 0.4362,
+      "step": 18810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977640332703141e-05,
+      "loss": 0.495,
+      "step": 18811
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776368778617637e-05,
+      "loss": 0.4617,
+      "step": 18812
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977633422756518e-05,
+      "loss": 0.4111,
+      "step": 18813
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776299673874046e-05,
+      "loss": 0.4562,
+      "step": 18814
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977626511754424e-05,
+      "loss": 0.4341,
+      "step": 18815
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776230558575776e-05,
+      "loss": 0.4513,
+      "step": 18816
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977619599696866e-05,
+      "loss": 0.3662,
+      "step": 18817
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776161432722904e-05,
+      "loss": 0.4171,
+      "step": 18818
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776126865838515e-05,
+      "loss": 0.4921,
+      "step": 18819
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776092296315506e-05,
+      "loss": 0.455,
+      "step": 18820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776057724153882e-05,
+      "loss": 0.3926,
+      "step": 18821
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9776023149353655e-05,
+      "loss": 0.6198,
+      "step": 18822
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775988571914836e-05,
+      "loss": 0.4712,
+      "step": 18823
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775953991837427e-05,
+      "loss": 0.5048,
+      "step": 18824
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775919409121447e-05,
+      "loss": 0.43,
+      "step": 18825
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775884823766898e-05,
+      "loss": 0.433,
+      "step": 18826
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775850235773794e-05,
+      "loss": 0.4724,
+      "step": 18827
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775815645142142e-05,
+      "loss": 0.4459,
+      "step": 18828
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775781051871948e-05,
+      "loss": 0.5771,
+      "step": 18829
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977574645596323e-05,
+      "loss": 0.4878,
+      "step": 18830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977571185741599e-05,
+      "loss": 0.4286,
+      "step": 18831
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977567725623024e-05,
+      "loss": 0.461,
+      "step": 18832
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775642652405985e-05,
+      "loss": 0.5234,
+      "step": 18833
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775608045943243e-05,
+      "loss": 0.4826,
+      "step": 18834
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977557343684202e-05,
+      "loss": 0.4896,
+      "step": 18835
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977553882510232e-05,
+      "loss": 0.4508,
+      "step": 18836
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977550421072416e-05,
+      "loss": 0.4583,
+      "step": 18837
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775469593707543e-05,
+      "loss": 0.4618,
+      "step": 18838
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775434974052483e-05,
+      "loss": 0.4316,
+      "step": 18839
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775400351758987e-05,
+      "loss": 0.416,
+      "step": 18840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775365726827064e-05,
+      "loss": 0.4063,
+      "step": 18841
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775331099256725e-05,
+      "loss": 0.4401,
+      "step": 18842
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775296469047977e-05,
+      "loss": 0.4951,
+      "step": 18843
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977526183620083e-05,
+      "loss": 0.5388,
+      "step": 18844
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775227200715296e-05,
+      "loss": 0.4654,
+      "step": 18845
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977519256259138e-05,
+      "loss": 0.5129,
+      "step": 18846
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.97751579218291e-05,
+      "loss": 0.5339,
+      "step": 18847
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775123278428454e-05,
+      "loss": 0.536,
+      "step": 18848
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9775088632389456e-05,
+      "loss": 0.4752,
+      "step": 18849
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977505398371212e-05,
+      "loss": 0.4835,
+      "step": 18850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977501933239645e-05,
+      "loss": 0.3977,
+      "step": 18851
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774984678442453e-05,
+      "loss": 0.5115,
+      "step": 18852
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774950021850143e-05,
+      "loss": 0.5702,
+      "step": 18853
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977491536261953e-05,
+      "loss": 0.4504,
+      "step": 18854
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774880700750623e-05,
+      "loss": 0.5681,
+      "step": 18855
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774846036243427e-05,
+      "loss": 0.5971,
+      "step": 18856
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774811369097957e-05,
+      "loss": 0.5189,
+      "step": 18857
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774776699314216e-05,
+      "loss": 0.5337,
+      "step": 18858
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977474202689222e-05,
+      "loss": 0.4929,
+      "step": 18859
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774707351831975e-05,
+      "loss": 0.4567,
+      "step": 18860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977467267413349e-05,
+      "loss": 0.5483,
+      "step": 18861
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774637993796777e-05,
+      "loss": 0.6516,
+      "step": 18862
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774603310821844e-05,
+      "loss": 0.519,
+      "step": 18863
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774568625208697e-05,
+      "loss": 0.5526,
+      "step": 18864
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977453393695735e-05,
+      "loss": 0.464,
+      "step": 18865
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977449924606781e-05,
+      "loss": 0.5232,
+      "step": 18866
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774464552540086e-05,
+      "loss": 0.5032,
+      "step": 18867
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977442985637419e-05,
+      "loss": 0.3728,
+      "step": 18868
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977439515757013e-05,
+      "loss": 0.4576,
+      "step": 18869
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774360456127915e-05,
+      "loss": 0.5272,
+      "step": 18870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774325752047553e-05,
+      "loss": 0.503,
+      "step": 18871
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774291045329056e-05,
+      "loss": 0.4885,
+      "step": 18872
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977425633597243e-05,
+      "loss": 0.513,
+      "step": 18873
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977422162397769e-05,
+      "loss": 0.5013,
+      "step": 18874
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9774186909344837e-05,
+      "loss": 0.5179,
+      "step": 18875
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.977415219207389e-05,
+      "loss": 0.5104,
+      "step": 18876
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977411747216485e-05,
+      "loss": 0.4969,
+      "step": 18877
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9774082749617733e-05,
+      "loss": 0.4599,
+      "step": 18878
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9774048024432542e-05,
+      "loss": 0.5108,
+      "step": 18879
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9774013296609294e-05,
+      "loss": 0.5666,
+      "step": 18880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977397856614799e-05,
+      "loss": 0.4753,
+      "step": 18881
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773943833048648e-05,
+      "loss": 0.4729,
+      "step": 18882
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773909097311267e-05,
+      "loss": 0.5063,
+      "step": 18883
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773874358935866e-05,
+      "loss": 0.5023,
+      "step": 18884
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977383961792245e-05,
+      "loss": 0.448,
+      "step": 18885
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977380487427103e-05,
+      "loss": 0.456,
+      "step": 18886
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773770127981615e-05,
+      "loss": 0.553,
+      "step": 18887
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773735379054212e-05,
+      "loss": 0.4701,
+      "step": 18888
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977370062748883e-05,
+      "loss": 0.4554,
+      "step": 18889
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773665873285484e-05,
+      "loss": 0.4264,
+      "step": 18890
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773631116444182e-05,
+      "loss": 0.5174,
+      "step": 18891
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773596356964928e-05,
+      "loss": 0.4946,
+      "step": 18892
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773561594847735e-05,
+      "loss": 0.5144,
+      "step": 18893
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977352683009261e-05,
+      "loss": 0.6209,
+      "step": 18894
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773492062699565e-05,
+      "loss": 0.4182,
+      "step": 18895
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977345729266861e-05,
+      "loss": 0.402,
+      "step": 18896
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773422519999753e-05,
+      "loss": 0.481,
+      "step": 18897
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773387744693007e-05,
+      "loss": 0.4321,
+      "step": 18898
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773352966748374e-05,
+      "loss": 0.5072,
+      "step": 18899
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977331818616587e-05,
+      "loss": 0.4741,
+      "step": 18900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.97732834029455e-05,
+      "loss": 0.3987,
+      "step": 18901
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773248617087274e-05,
+      "loss": 0.4938,
+      "step": 18902
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.97732138285912e-05,
+      "loss": 0.5364,
+      "step": 18903
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773179037457296e-05,
+      "loss": 0.4539,
+      "step": 18904
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773144243685564e-05,
+      "loss": 0.4633,
+      "step": 18905
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773109447276012e-05,
+      "loss": 0.509,
+      "step": 18906
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9773074648228654e-05,
+      "loss": 0.4588,
+      "step": 18907
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.97730398465435e-05,
+      "loss": 0.5171,
+      "step": 18908
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977300504222055e-05,
+      "loss": 0.4767,
+      "step": 18909
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772970235259826e-05,
+      "loss": 0.4819,
+      "step": 18910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977293542566133e-05,
+      "loss": 0.4427,
+      "step": 18911
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772900613425074e-05,
+      "loss": 0.4976,
+      "step": 18912
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977286579855106e-05,
+      "loss": 0.57,
+      "step": 18913
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772830981039313e-05,
+      "loss": 0.4367,
+      "step": 18914
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977279616088983e-05,
+      "loss": 0.4457,
+      "step": 18915
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977276133810262e-05,
+      "loss": 0.4414,
+      "step": 18916
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772726512677702e-05,
+      "loss": 0.4725,
+      "step": 18917
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772691684615074e-05,
+      "loss": 0.4848,
+      "step": 18918
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772656853914752e-05,
+      "loss": 0.4523,
+      "step": 18919
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977262202057675e-05,
+      "loss": 0.5146,
+      "step": 18920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772587184601065e-05,
+      "loss": 0.423,
+      "step": 18921
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772552345987714e-05,
+      "loss": 0.4607,
+      "step": 18922
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977251750473671e-05,
+      "loss": 0.4254,
+      "step": 18923
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772482660848053e-05,
+      "loss": 0.492,
+      "step": 18924
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772447814321757e-05,
+      "loss": 0.4316,
+      "step": 18925
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772412965157835e-05,
+      "loss": 0.4761,
+      "step": 18926
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772378113356292e-05,
+      "loss": 0.4194,
+      "step": 18927
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772343258917136e-05,
+      "loss": 0.4626,
+      "step": 18928
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977230840184038e-05,
+      "loss": 0.548,
+      "step": 18929
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772273542126036e-05,
+      "loss": 0.4401,
+      "step": 18930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772238679774106e-05,
+      "loss": 0.5036,
+      "step": 18931
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772203814784602e-05,
+      "loss": 0.4727,
+      "step": 18932
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772168947157537e-05,
+      "loss": 0.5122,
+      "step": 18933
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977213407689292e-05,
+      "loss": 0.5216,
+      "step": 18934
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772099203990752e-05,
+      "loss": 0.4035,
+      "step": 18935
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9772064328451054e-05,
+      "loss": 0.4789,
+      "step": 18936
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977202945027383e-05,
+      "loss": 0.4622,
+      "step": 18937
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771994569459085e-05,
+      "loss": 0.4708,
+      "step": 18938
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771959686006835e-05,
+      "loss": 0.5992,
+      "step": 18939
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977192479991709e-05,
+      "loss": 0.4721,
+      "step": 18940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771889911189858e-05,
+      "loss": 0.4764,
+      "step": 18941
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771855019825145e-05,
+      "loss": 0.4672,
+      "step": 18942
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771820125822957e-05,
+      "loss": 0.4687,
+      "step": 18943
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977178522918332e-05,
+      "loss": 0.4972,
+      "step": 18944
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771750329906223e-05,
+      "loss": 0.3886,
+      "step": 18945
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771715427991688e-05,
+      "loss": 0.521,
+      "step": 18946
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977168052343972e-05,
+      "loss": 0.4874,
+      "step": 18947
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771645616250336e-05,
+      "loss": 0.477,
+      "step": 18948
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771610706423533e-05,
+      "loss": 0.4222,
+      "step": 18949
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771575793959327e-05,
+      "loss": 0.4487,
+      "step": 18950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977154087885773e-05,
+      "loss": 0.4723,
+      "step": 18951
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771505961118747e-05,
+      "loss": 0.4694,
+      "step": 18952
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977147104074239e-05,
+      "loss": 0.449,
+      "step": 18953
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771436117728666e-05,
+      "loss": 0.3554,
+      "step": 18954
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771401192077583e-05,
+      "loss": 0.5096,
+      "step": 18955
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977136626378916e-05,
+      "loss": 0.4955,
+      "step": 18956
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771331332863392e-05,
+      "loss": 0.4978,
+      "step": 18957
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.97712963993003e-05,
+      "loss": 0.4798,
+      "step": 18958
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977126146309989e-05,
+      "loss": 0.5329,
+      "step": 18959
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977122652426217e-05,
+      "loss": 0.3966,
+      "step": 18960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977119158278715e-05,
+      "loss": 0.4594,
+      "step": 18961
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977115663867484e-05,
+      "loss": 0.4285,
+      "step": 18962
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977112169192525e-05,
+      "loss": 0.5234,
+      "step": 18963
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771086742538385e-05,
+      "loss": 0.6089,
+      "step": 18964
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977105179051426e-05,
+      "loss": 0.471,
+      "step": 18965
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9771016835852883e-05,
+      "loss": 0.3885,
+      "step": 18966
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9770981878554264e-05,
+      "loss": 0.441,
+      "step": 18967
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977094691861841e-05,
+      "loss": 0.4633,
+      "step": 18968
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9770911956045333e-05,
+      "loss": 0.5384,
+      "step": 18969
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977087699083504e-05,
+      "loss": 0.6057,
+      "step": 18970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977084202298754e-05,
+      "loss": 0.5294,
+      "step": 18971
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9770807052502846e-05,
+      "loss": 0.395,
+      "step": 18972
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9770772079380967e-05,
+      "loss": 0.6078,
+      "step": 18973
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.977073710362191e-05,
+      "loss": 0.5264,
+      "step": 18974
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770702125225685e-05,
+      "loss": 0.426,
+      "step": 18975
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.97706671441923e-05,
+      "loss": 0.4703,
+      "step": 18976
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770632160521767e-05,
+      "loss": 0.4407,
+      "step": 18977
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770597174214096e-05,
+      "loss": 0.4614,
+      "step": 18978
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770562185269293e-05,
+      "loss": 0.4668,
+      "step": 18979
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770527193687372e-05,
+      "loss": 0.4423,
+      "step": 18980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.977049219946834e-05,
+      "loss": 0.4952,
+      "step": 18981
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770457202612204e-05,
+      "loss": 0.451,
+      "step": 18982
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770422203118978e-05,
+      "loss": 0.5192,
+      "step": 18983
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.977038720098867e-05,
+      "loss": 0.4781,
+      "step": 18984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770352196221286e-05,
+      "loss": 0.4611,
+      "step": 18985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770317188816843e-05,
+      "loss": 0.4987,
+      "step": 18986
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770282178775343e-05,
+      "loss": 0.5161,
+      "step": 18987
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770247166096796e-05,
+      "loss": 0.4307,
+      "step": 18988
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770212150781215e-05,
+      "loss": 0.444,
+      "step": 18989
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.977017713282861e-05,
+      "loss": 0.431,
+      "step": 18990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770142112238985e-05,
+      "loss": 0.4495,
+      "step": 18991
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770107089012356e-05,
+      "loss": 0.4599,
+      "step": 18992
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770072063148728e-05,
+      "loss": 0.4778,
+      "step": 18993
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770037034648113e-05,
+      "loss": 0.4076,
+      "step": 18994
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.977000200351052e-05,
+      "loss": 0.4597,
+      "step": 18995
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769966969735955e-05,
+      "loss": 0.4415,
+      "step": 18996
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769931933324432e-05,
+      "loss": 0.5267,
+      "step": 18997
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976989689427596e-05,
+      "loss": 0.4674,
+      "step": 18998
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769861852590546e-05,
+      "loss": 0.5277,
+      "step": 18999
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.97698268082682e-05,
+      "loss": 0.5019,
+      "step": 19000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976979176130893e-05,
+      "loss": 0.5639,
+      "step": 19001
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976975671171275e-05,
+      "loss": 0.4735,
+      "step": 19002
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769721659479672e-05,
+      "loss": 0.4545,
+      "step": 19003
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976968660460969e-05,
+      "loss": 0.4429,
+      "step": 19004
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769651547102834e-05,
+      "loss": 0.5022,
+      "step": 19005
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769616486959097e-05,
+      "loss": 0.5473,
+      "step": 19006
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769581424178497e-05,
+      "loss": 0.5013,
+      "step": 19007
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976954635876104e-05,
+      "loss": 0.5266,
+      "step": 19008
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769511290706737e-05,
+      "loss": 0.3871,
+      "step": 19009
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769476220015597e-05,
+      "loss": 0.4759,
+      "step": 19010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976944114668763e-05,
+      "loss": 0.5163,
+      "step": 19011
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769406070722847e-05,
+      "loss": 0.4568,
+      "step": 19012
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769370992121257e-05,
+      "loss": 0.5447,
+      "step": 19013
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769335910882863e-05,
+      "loss": 0.4519,
+      "step": 19014
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769300827007683e-05,
+      "loss": 0.4946,
+      "step": 19015
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769265740495723e-05,
+      "loss": 0.4862,
+      "step": 19016
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976923065134699e-05,
+      "loss": 0.4363,
+      "step": 19017
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769195559561498e-05,
+      "loss": 0.4594,
+      "step": 19018
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769160465139254e-05,
+      "loss": 0.4998,
+      "step": 19019
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769125368080266e-05,
+      "loss": 0.5183,
+      "step": 19020
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769090268384547e-05,
+      "loss": 0.4806,
+      "step": 19021
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769055166052106e-05,
+      "loss": 0.3984,
+      "step": 19022
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9769020061082952e-05,
+      "loss": 0.439,
+      "step": 19023
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976898495347709e-05,
+      "loss": 0.5113,
+      "step": 19024
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768949843234537e-05,
+      "loss": 0.4905,
+      "step": 19025
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.97689147303553e-05,
+      "loss": 0.4437,
+      "step": 19026
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768879614839382e-05,
+      "loss": 0.4259,
+      "step": 19027
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.97688444966868e-05,
+      "loss": 0.486,
+      "step": 19028
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768809375897563e-05,
+      "loss": 0.462,
+      "step": 19029
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768774252471677e-05,
+      "loss": 0.4512,
+      "step": 19030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768739126409153e-05,
+      "loss": 0.4508,
+      "step": 19031
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768703997710003e-05,
+      "loss": 0.4335,
+      "step": 19032
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976866886637423e-05,
+      "loss": 0.58,
+      "step": 19033
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976863373240185e-05,
+      "loss": 0.4298,
+      "step": 19034
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768598595792872e-05,
+      "loss": 0.4923,
+      "step": 19035
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.97685634565473e-05,
+      "loss": 0.5188,
+      "step": 19036
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976852831466515e-05,
+      "loss": 0.4413,
+      "step": 19037
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768493170146425e-05,
+      "loss": 0.4616,
+      "step": 19038
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768458022991138e-05,
+      "loss": 0.518,
+      "step": 19039
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.97684228731993e-05,
+      "loss": 0.4615,
+      "step": 19040
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768387720770923e-05,
+      "loss": 0.3986,
+      "step": 19041
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768352565706008e-05,
+      "loss": 0.5086,
+      "step": 19042
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976831740800457e-05,
+      "loss": 0.4412,
+      "step": 19043
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976828224766662e-05,
+      "loss": 0.4457,
+      "step": 19044
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976824708469216e-05,
+      "loss": 0.4287,
+      "step": 19045
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976821191908121e-05,
+      "loss": 0.4605,
+      "step": 19046
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768176750833768e-05,
+      "loss": 0.4967,
+      "step": 19047
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768141579949852e-05,
+      "loss": 0.4969,
+      "step": 19048
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768106406429472e-05,
+      "loss": 0.446,
+      "step": 19049
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976807123027263e-05,
+      "loss": 0.4341,
+      "step": 19050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768036051479342e-05,
+      "loss": 0.4964,
+      "step": 19051
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9768000870049615e-05,
+      "loss": 0.4513,
+      "step": 19052
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767965685983458e-05,
+      "loss": 0.451,
+      "step": 19053
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767930499280884e-05,
+      "loss": 0.5755,
+      "step": 19054
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.97678953099419e-05,
+      "loss": 0.4581,
+      "step": 19055
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976786011796651e-05,
+      "loss": 0.4515,
+      "step": 19056
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767824923354735e-05,
+      "loss": 0.5485,
+      "step": 19057
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767789726106574e-05,
+      "loss": 0.4322,
+      "step": 19058
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767754526222042e-05,
+      "loss": 0.5152,
+      "step": 19059
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767719323701148e-05,
+      "loss": 0.6052,
+      "step": 19060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.97676841185439e-05,
+      "loss": 0.4765,
+      "step": 19061
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767648910750315e-05,
+      "loss": 0.4441,
+      "step": 19062
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767613700320386e-05,
+      "loss": 0.4819,
+      "step": 19063
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767578487254138e-05,
+      "loss": 0.4549,
+      "step": 19064
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767543271551574e-05,
+      "loss": 0.4638,
+      "step": 19065
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767508053212706e-05,
+      "loss": 0.4785,
+      "step": 19066
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767472832237538e-05,
+      "loss": 0.4443,
+      "step": 19067
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767437608626086e-05,
+      "loss": 0.5335,
+      "step": 19068
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976740238237836e-05,
+      "loss": 0.5508,
+      "step": 19069
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976736715349436e-05,
+      "loss": 0.4631,
+      "step": 19070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9767331921974104e-05,
+      "loss": 0.4745,
+      "step": 19071
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.97672966878176e-05,
+      "loss": 0.4108,
+      "step": 19072
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.976726145102486e-05,
+      "loss": 0.6219,
+      "step": 19073
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9767226211595885e-05,
+      "loss": 0.4379,
+      "step": 19074
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9767190969530695e-05,
+      "loss": 0.384,
+      "step": 19075
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976715572482929e-05,
+      "loss": 0.4517,
+      "step": 19076
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9767120477491687e-05,
+      "loss": 0.5433,
+      "step": 19077
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976708522751789e-05,
+      "loss": 0.5019,
+      "step": 19078
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9767049974907915e-05,
+      "loss": 0.4798,
+      "step": 19079
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9767014719661764e-05,
+      "loss": 0.5307,
+      "step": 19080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976697946177945e-05,
+      "loss": 0.3874,
+      "step": 19081
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766944201260987e-05,
+      "loss": 0.4626,
+      "step": 19082
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766908938106378e-05,
+      "loss": 0.4742,
+      "step": 19083
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766873672315632e-05,
+      "loss": 0.579,
+      "step": 19084
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766838403888764e-05,
+      "loss": 0.4797,
+      "step": 19085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766803132825778e-05,
+      "loss": 0.4719,
+      "step": 19086
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976676785912669e-05,
+      "loss": 0.5198,
+      "step": 19087
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.97667325827915e-05,
+      "loss": 0.4948,
+      "step": 19088
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976669730382023e-05,
+      "loss": 0.4853,
+      "step": 19089
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766662022212878e-05,
+      "loss": 0.4599,
+      "step": 19090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976662673796946e-05,
+      "loss": 0.5124,
+      "step": 19091
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976659145108998e-05,
+      "loss": 0.4745,
+      "step": 19092
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766556161574456e-05,
+      "loss": 0.5282,
+      "step": 19093
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976652086942289e-05,
+      "loss": 0.4586,
+      "step": 19094
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.97664855746353e-05,
+      "loss": 0.4077,
+      "step": 19095
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766450277211682e-05,
+      "loss": 0.5064,
+      "step": 19096
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976641497715206e-05,
+      "loss": 0.4138,
+      "step": 19097
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766379674456435e-05,
+      "loss": 0.534,
+      "step": 19098
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766344369124817e-05,
+      "loss": 0.5432,
+      "step": 19099
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766309061157217e-05,
+      "loss": 0.4641,
+      "step": 19100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766273750553645e-05,
+      "loss": 0.4195,
+      "step": 19101
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766238437314113e-05,
+      "loss": 0.4994,
+      "step": 19102
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766203121438625e-05,
+      "loss": 0.5354,
+      "step": 19103
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976616780292719e-05,
+      "loss": 0.484,
+      "step": 19104
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766132481779823e-05,
+      "loss": 0.4748,
+      "step": 19105
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766097157996533e-05,
+      "loss": 0.4887,
+      "step": 19106
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766061831577327e-05,
+      "loss": 0.4778,
+      "step": 19107
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9766026502522213e-05,
+      "loss": 0.4835,
+      "step": 19108
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765991170831206e-05,
+      "loss": 0.5074,
+      "step": 19109
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765955836504308e-05,
+      "loss": 0.5699,
+      "step": 19110
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976592049954154e-05,
+      "loss": 0.4139,
+      "step": 19111
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.97658851599429e-05,
+      "loss": 0.429,
+      "step": 19112
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765849817708403e-05,
+      "loss": 0.5167,
+      "step": 19113
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765814472838055e-05,
+      "loss": 0.5513,
+      "step": 19114
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765779125331866e-05,
+      "loss": 0.4811,
+      "step": 19115
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765743775189853e-05,
+      "loss": 0.422,
+      "step": 19116
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976570842241202e-05,
+      "loss": 0.4601,
+      "step": 19117
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765673066998374e-05,
+      "loss": 0.4507,
+      "step": 19118
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765637708948932e-05,
+      "loss": 0.5168,
+      "step": 19119
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765602348263693e-05,
+      "loss": 0.4778,
+      "step": 19120
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765566984942674e-05,
+      "loss": 0.4185,
+      "step": 19121
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976553161898588e-05,
+      "loss": 0.4659,
+      "step": 19122
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976549625039333e-05,
+      "loss": 0.4683,
+      "step": 19123
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765460879165023e-05,
+      "loss": 0.5292,
+      "step": 19124
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976542550530097e-05,
+      "loss": 0.511,
+      "step": 19125
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765390128801187e-05,
+      "loss": 0.4116,
+      "step": 19126
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765354749665676e-05,
+      "loss": 0.437,
+      "step": 19127
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765319367894453e-05,
+      "loss": 0.4918,
+      "step": 19128
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765283983487528e-05,
+      "loss": 0.4799,
+      "step": 19129
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.97652485964449e-05,
+      "loss": 0.5486,
+      "step": 19130
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976521320676659e-05,
+      "loss": 0.5073,
+      "step": 19131
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765177814452603e-05,
+      "loss": 0.3447,
+      "step": 19132
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765142419502947e-05,
+      "loss": 0.4284,
+      "step": 19133
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765107021917636e-05,
+      "loss": 0.5298,
+      "step": 19134
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765071621696673e-05,
+      "loss": 0.4947,
+      "step": 19135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765036218840076e-05,
+      "loss": 0.5266,
+      "step": 19136
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9765000813347848e-05,
+      "loss": 0.5483,
+      "step": 19137
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764965405220002e-05,
+      "loss": 0.4798,
+      "step": 19138
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764929994456542e-05,
+      "loss": 0.4628,
+      "step": 19139
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764894581057485e-05,
+      "loss": 0.4617,
+      "step": 19140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764859165022837e-05,
+      "loss": 0.4219,
+      "step": 19141
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976482374635261e-05,
+      "loss": 0.495,
+      "step": 19142
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764788325046808e-05,
+      "loss": 0.462,
+      "step": 19143
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764752901105446e-05,
+      "loss": 0.4727,
+      "step": 19144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764717474528532e-05,
+      "loss": 0.4312,
+      "step": 19145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764682045316074e-05,
+      "loss": 0.4593,
+      "step": 19146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976464661346808e-05,
+      "loss": 0.4495,
+      "step": 19147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764611178984568e-05,
+      "loss": 0.4836,
+      "step": 19148
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976457574186554e-05,
+      "loss": 0.4937,
+      "step": 19149
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764540302111004e-05,
+      "loss": 0.5105,
+      "step": 19150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764504859720973e-05,
+      "loss": 0.4809,
+      "step": 19151
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976446941469546e-05,
+      "loss": 0.4511,
+      "step": 19152
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976443396703447e-05,
+      "loss": 0.4152,
+      "step": 19153
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764398516738014e-05,
+      "loss": 0.4615,
+      "step": 19154
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.97643630638061e-05,
+      "loss": 0.4645,
+      "step": 19155
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764327608238737e-05,
+      "loss": 0.4477,
+      "step": 19156
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976429215003594e-05,
+      "loss": 0.5408,
+      "step": 19157
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976425668919771e-05,
+      "loss": 0.4051,
+      "step": 19158
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976422122572407e-05,
+      "loss": 0.4935,
+      "step": 19159
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764185759615016e-05,
+      "loss": 0.4757,
+      "step": 19160
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976415029087056e-05,
+      "loss": 0.441,
+      "step": 19161
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764114819490717e-05,
+      "loss": 0.4997,
+      "step": 19162
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764079345475495e-05,
+      "loss": 0.4606,
+      "step": 19163
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.97640438688249e-05,
+      "loss": 0.4952,
+      "step": 19164
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9764008389538944e-05,
+      "loss": 0.4458,
+      "step": 19165
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9763972907617635e-05,
+      "loss": 0.4673,
+      "step": 19166
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9763937423060987e-05,
+      "loss": 0.5186,
+      "step": 19167
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976390193586901e-05,
+      "loss": 0.4787,
+      "step": 19168
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9763866446041705e-05,
+      "loss": 0.5725,
+      "step": 19169
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9763830953579085e-05,
+      "loss": 0.4818,
+      "step": 19170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.976379545848117e-05,
+      "loss": 0.4473,
+      "step": 19171
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763759960747954e-05,
+      "loss": 0.4564,
+      "step": 19172
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763724460379453e-05,
+      "loss": 0.4007,
+      "step": 19173
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763688957375678e-05,
+      "loss": 0.4273,
+      "step": 19174
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976365345173664e-05,
+      "loss": 0.504,
+      "step": 19175
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763617943462344e-05,
+      "loss": 0.4624,
+      "step": 19176
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763582432552803e-05,
+      "loss": 0.4525,
+      "step": 19177
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763546919008024e-05,
+      "loss": 0.4269,
+      "step": 19178
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976351140282802e-05,
+      "loss": 0.4758,
+      "step": 19179
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.97634758840128e-05,
+      "loss": 0.4633,
+      "step": 19180
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976344036256237e-05,
+      "loss": 0.4822,
+      "step": 19181
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976340483847674e-05,
+      "loss": 0.4703,
+      "step": 19182
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763369311755924e-05,
+      "loss": 0.5083,
+      "step": 19183
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976333378239993e-05,
+      "loss": 0.4757,
+      "step": 19184
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763298250408766e-05,
+      "loss": 0.5629,
+      "step": 19185
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763262715782438e-05,
+      "loss": 0.4906,
+      "step": 19186
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763227178520963e-05,
+      "loss": 0.5478,
+      "step": 19187
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976319163862435e-05,
+      "loss": 0.4478,
+      "step": 19188
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763156096092602e-05,
+      "loss": 0.4883,
+      "step": 19189
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763120550925735e-05,
+      "loss": 0.4796,
+      "step": 19190
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763085003123752e-05,
+      "loss": 0.4076,
+      "step": 19191
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976304945268667e-05,
+      "loss": 0.4406,
+      "step": 19192
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9763013899614495e-05,
+      "loss": 0.4341,
+      "step": 19193
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976297834390724e-05,
+      "loss": 0.4644,
+      "step": 19194
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762942785564906e-05,
+      "loss": 0.5534,
+      "step": 19195
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976290722458751e-05,
+      "loss": 0.4954,
+      "step": 19196
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976287166097506e-05,
+      "loss": 0.5527,
+      "step": 19197
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762836094727565e-05,
+      "loss": 0.4484,
+      "step": 19198
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762800525845037e-05,
+      "loss": 0.4409,
+      "step": 19199
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762764954327483e-05,
+      "loss": 0.5395,
+      "step": 19200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976272938017491e-05,
+      "loss": 0.4184,
+      "step": 19201
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762693803387332e-05,
+      "loss": 0.4657,
+      "step": 19202
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762658223964756e-05,
+      "loss": 0.4269,
+      "step": 19203
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762622641907198e-05,
+      "loss": 0.46,
+      "step": 19204
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976258705721466e-05,
+      "loss": 0.6095,
+      "step": 19205
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762551469887152e-05,
+      "loss": 0.4698,
+      "step": 19206
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762515879924687e-05,
+      "loss": 0.5087,
+      "step": 19207
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762480287327275e-05,
+      "loss": 0.492,
+      "step": 19208
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762444692094923e-05,
+      "loss": 0.4275,
+      "step": 19209
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976240909422764e-05,
+      "loss": 0.5825,
+      "step": 19210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762373493725437e-05,
+      "loss": 0.6063,
+      "step": 19211
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762337890588327e-05,
+      "loss": 0.4389,
+      "step": 19212
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762302284816314e-05,
+      "loss": 0.4904,
+      "step": 19213
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976226667640941e-05,
+      "loss": 0.4567,
+      "step": 19214
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762231065367623e-05,
+      "loss": 0.5172,
+      "step": 19215
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762195451690966e-05,
+      "loss": 0.4331,
+      "step": 19216
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762159835379447e-05,
+      "loss": 0.4518,
+      "step": 19217
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762124216433077e-05,
+      "loss": 0.4746,
+      "step": 19218
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976208859485186e-05,
+      "loss": 0.6059,
+      "step": 19219
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762052970635814e-05,
+      "loss": 0.5293,
+      "step": 19220
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9762017343784942e-05,
+      "loss": 0.5344,
+      "step": 19221
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761981714299255e-05,
+      "loss": 0.539,
+      "step": 19222
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761946082178768e-05,
+      "loss": 0.4268,
+      "step": 19223
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976191044742348e-05,
+      "loss": 0.4882,
+      "step": 19224
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976187481003341e-05,
+      "loss": 0.4836,
+      "step": 19225
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761839170008567e-05,
+      "loss": 0.5551,
+      "step": 19226
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976180352734895e-05,
+      "loss": 0.4596,
+      "step": 19227
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761767882054582e-05,
+      "loss": 0.4517,
+      "step": 19228
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976173223412547e-05,
+      "loss": 0.4584,
+      "step": 19229
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761696583561618e-05,
+      "loss": 0.4677,
+      "step": 19230
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761660930363037e-05,
+      "loss": 0.436,
+      "step": 19231
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976162527452974e-05,
+      "loss": 0.5242,
+      "step": 19232
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976158961606173e-05,
+      "loss": 0.5596,
+      "step": 19233
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761553954959028e-05,
+      "loss": 0.4377,
+      "step": 19234
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761518291221633e-05,
+      "loss": 0.4651,
+      "step": 19235
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976148262484956e-05,
+      "loss": 0.4734,
+      "step": 19236
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761446955842816e-05,
+      "loss": 0.4384,
+      "step": 19237
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761411284201414e-05,
+      "loss": 0.5087,
+      "step": 19238
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761375609925357e-05,
+      "loss": 0.5198,
+      "step": 19239
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761339933014665e-05,
+      "loss": 0.4363,
+      "step": 19240
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976130425346934e-05,
+      "loss": 0.4766,
+      "step": 19241
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976126857128939e-05,
+      "loss": 0.4978,
+      "step": 19242
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976123288647483e-05,
+      "loss": 0.4602,
+      "step": 19243
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761197199025666e-05,
+      "loss": 0.439,
+      "step": 19244
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761161508941915e-05,
+      "loss": 0.55,
+      "step": 19245
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761125816223575e-05,
+      "loss": 0.5369,
+      "step": 19246
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761090120870662e-05,
+      "loss": 0.4653,
+      "step": 19247
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976105442288319e-05,
+      "loss": 0.4168,
+      "step": 19248
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9761018722261157e-05,
+      "loss": 0.581,
+      "step": 19249
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760983019004586e-05,
+      "loss": 0.443,
+      "step": 19250
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760947313113475e-05,
+      "loss": 0.4937,
+      "step": 19251
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976091160458784e-05,
+      "loss": 0.5534,
+      "step": 19252
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976087589342769e-05,
+      "loss": 0.4846,
+      "step": 19253
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760840179633037e-05,
+      "loss": 0.474,
+      "step": 19254
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760804463203884e-05,
+      "loss": 0.4679,
+      "step": 19255
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760768744140245e-05,
+      "loss": 0.4931,
+      "step": 19256
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976073302244213e-05,
+      "loss": 0.4876,
+      "step": 19257
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760697298109543e-05,
+      "loss": 0.4092,
+      "step": 19258
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760661571142504e-05,
+      "loss": 0.5177,
+      "step": 19259
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760625841541016e-05,
+      "loss": 0.4823,
+      "step": 19260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760590109305083e-05,
+      "loss": 0.4086,
+      "step": 19261
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976055437443473e-05,
+      "loss": 0.4475,
+      "step": 19262
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976051863692995e-05,
+      "loss": 0.5114,
+      "step": 19263
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760482896790765e-05,
+      "loss": 0.4271,
+      "step": 19264
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976044715401718e-05,
+      "loss": 0.4407,
+      "step": 19265
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760411408609203e-05,
+      "loss": 0.4516,
+      "step": 19266
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760375660566846e-05,
+      "loss": 0.4616,
+      "step": 19267
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760339909890118e-05,
+      "loss": 0.5258,
+      "step": 19268
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.976030415657903e-05,
+      "loss": 0.4557,
+      "step": 19269
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9760268400633587e-05,
+      "loss": 0.4184,
+      "step": 19270
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9760232642053805e-05,
+      "loss": 0.4776,
+      "step": 19271
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.976019688083969e-05,
+      "loss": 0.5804,
+      "step": 19272
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.976016111699125e-05,
+      "loss": 0.4565,
+      "step": 19273
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9760125350508497e-05,
+      "loss": 0.4587,
+      "step": 19274
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9760089581391445e-05,
+      "loss": 0.5072,
+      "step": 19275
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9760053809640097e-05,
+      "loss": 0.4382,
+      "step": 19276
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9760018035254463e-05,
+      "loss": 0.4642,
+      "step": 19277
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759982258234553e-05,
+      "loss": 0.4817,
+      "step": 19278
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759946478580384e-05,
+      "loss": 0.4508,
+      "step": 19279
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759910696291956e-05,
+      "loss": 0.5102,
+      "step": 19280
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975987491136928e-05,
+      "loss": 0.5396,
+      "step": 19281
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759839123812375e-05,
+      "loss": 0.3814,
+      "step": 19282
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759803333621238e-05,
+      "loss": 0.5296,
+      "step": 19283
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975976754079589e-05,
+      "loss": 0.5342,
+      "step": 19284
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759731745336326e-05,
+      "loss": 0.4277,
+      "step": 19285
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759695947242572e-05,
+      "loss": 0.4735,
+      "step": 19286
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975966014651463e-05,
+      "loss": 0.4335,
+      "step": 19287
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975962434315251e-05,
+      "loss": 0.5849,
+      "step": 19288
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975958853715622e-05,
+      "loss": 0.4467,
+      "step": 19289
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759552728525772e-05,
+      "loss": 0.4791,
+      "step": 19290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759516917261177e-05,
+      "loss": 0.5698,
+      "step": 19291
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975948110336244e-05,
+      "loss": 0.4255,
+      "step": 19292
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759445286829573e-05,
+      "loss": 0.4554,
+      "step": 19293
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759409467662588e-05,
+      "loss": 0.6114,
+      "step": 19294
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759373645861493e-05,
+      "loss": 0.4749,
+      "step": 19295
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759337821426298e-05,
+      "loss": 0.4949,
+      "step": 19296
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975930199435701e-05,
+      "loss": 0.5037,
+      "step": 19297
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975926616465364e-05,
+      "loss": 0.497,
+      "step": 19298
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.97592303323162e-05,
+      "loss": 0.5864,
+      "step": 19299
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.97591944973447e-05,
+      "loss": 0.5204,
+      "step": 19300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759158659739147e-05,
+      "loss": 0.4205,
+      "step": 19301
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9759122819499547e-05,
+      "loss": 0.4286,
+      "step": 19302
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975908697662592e-05,
+      "loss": 0.4524,
+      "step": 19303
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975905113111827e-05,
+      "loss": 0.4712,
+      "step": 19304
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.97590152829766e-05,
+      "loss": 0.4803,
+      "step": 19305
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975897943220093e-05,
+      "loss": 0.4924,
+      "step": 19306
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758943578791266e-05,
+      "loss": 0.506,
+      "step": 19307
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758907722747617e-05,
+      "loss": 0.4885,
+      "step": 19308
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758871864069994e-05,
+      "loss": 0.5456,
+      "step": 19309
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758836002758407e-05,
+      "loss": 0.4049,
+      "step": 19310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758800138812862e-05,
+      "loss": 0.4579,
+      "step": 19311
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758764272233373e-05,
+      "loss": 0.457,
+      "step": 19312
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758728403019948e-05,
+      "loss": 0.5228,
+      "step": 19313
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758692531172592e-05,
+      "loss": 0.4454,
+      "step": 19314
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758656656691323e-05,
+      "loss": 0.4669,
+      "step": 19315
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975862077957615e-05,
+      "loss": 0.5774,
+      "step": 19316
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758584899827075e-05,
+      "loss": 0.4822,
+      "step": 19317
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758549017444113e-05,
+      "loss": 0.4975,
+      "step": 19318
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758513132427275e-05,
+      "loss": 0.4713,
+      "step": 19319
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758477244776565e-05,
+      "loss": 0.4405,
+      "step": 19320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758441354492e-05,
+      "loss": 0.5061,
+      "step": 19321
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758405461573583e-05,
+      "loss": 0.4727,
+      "step": 19322
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758369566021333e-05,
+      "loss": 0.4364,
+      "step": 19323
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758333667835247e-05,
+      "loss": 0.4959,
+      "step": 19324
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758297767015343e-05,
+      "loss": 0.4879,
+      "step": 19325
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975826186356163e-05,
+      "loss": 0.5103,
+      "step": 19326
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758225957474116e-05,
+      "loss": 0.4273,
+      "step": 19327
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975819004875281e-05,
+      "loss": 0.4241,
+      "step": 19328
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758154137397723e-05,
+      "loss": 0.4517,
+      "step": 19329
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758118223408866e-05,
+      "loss": 0.4571,
+      "step": 19330
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758082306786247e-05,
+      "loss": 0.4824,
+      "step": 19331
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9758046387529878e-05,
+      "loss": 0.4099,
+      "step": 19332
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975801046563976e-05,
+      "loss": 0.4149,
+      "step": 19333
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757974541115918e-05,
+      "loss": 0.4598,
+      "step": 19334
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757938613958347e-05,
+      "loss": 0.411,
+      "step": 19335
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757902684167063e-05,
+      "loss": 0.5124,
+      "step": 19336
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757866751742076e-05,
+      "loss": 0.4487,
+      "step": 19337
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757830816683395e-05,
+      "loss": 0.4977,
+      "step": 19338
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757794878991032e-05,
+      "loss": 0.486,
+      "step": 19339
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757758938664996e-05,
+      "loss": 0.4609,
+      "step": 19340
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975772299570529e-05,
+      "loss": 0.4824,
+      "step": 19341
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975768705011193e-05,
+      "loss": 0.4388,
+      "step": 19342
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975765110188493e-05,
+      "loss": 0.571,
+      "step": 19343
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975761515102429e-05,
+      "loss": 0.4665,
+      "step": 19344
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757579197530023e-05,
+      "loss": 0.4424,
+      "step": 19345
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757543241402143e-05,
+      "loss": 0.4671,
+      "step": 19346
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757507282640652e-05,
+      "loss": 0.4278,
+      "step": 19347
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975747132124557e-05,
+      "loss": 0.4709,
+      "step": 19348
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757435357216898e-05,
+      "loss": 0.4535,
+      "step": 19349
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975739939055465e-05,
+      "loss": 0.4565,
+      "step": 19350
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975736342125883e-05,
+      "loss": 0.5025,
+      "step": 19351
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757327449329456e-05,
+      "loss": 0.4336,
+      "step": 19352
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975729147476653e-05,
+      "loss": 0.3971,
+      "step": 19353
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975725549757007e-05,
+      "loss": 0.4189,
+      "step": 19354
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975721951774008e-05,
+      "loss": 0.472,
+      "step": 19355
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757183535276567e-05,
+      "loss": 0.4007,
+      "step": 19356
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975714755017955e-05,
+      "loss": 0.5151,
+      "step": 19357
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975711156244903e-05,
+      "loss": 0.5999,
+      "step": 19358
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757075572085022e-05,
+      "loss": 0.4353,
+      "step": 19359
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9757039579087533e-05,
+      "loss": 0.4594,
+      "step": 19360
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975700358345657e-05,
+      "loss": 0.4408,
+      "step": 19361
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975696758519215e-05,
+      "loss": 0.4211,
+      "step": 19362
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9756931584294276e-05,
+      "loss": 0.4788,
+      "step": 19363
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9756895580762963e-05,
+      "loss": 0.477,
+      "step": 19364
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975685957459822e-05,
+      "loss": 0.5744,
+      "step": 19365
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975682356580005e-05,
+      "loss": 0.5253,
+      "step": 19366
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9756787554368473e-05,
+      "loss": 0.4731,
+      "step": 19367
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.975675154030349e-05,
+      "loss": 0.4292,
+      "step": 19368
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9756715523605113e-05,
+      "loss": 0.5378,
+      "step": 19369
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9756679504273356e-05,
+      "loss": 0.4358,
+      "step": 19370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9756643482308225e-05,
+      "loss": 0.6008,
+      "step": 19371
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975660745770973e-05,
+      "loss": 0.5344,
+      "step": 19372
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975657143047788e-05,
+      "loss": 0.4329,
+      "step": 19373
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9756535400612688e-05,
+      "loss": 0.4953,
+      "step": 19374
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975649936811416e-05,
+      "loss": 0.5192,
+      "step": 19375
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9756463332982304e-05,
+      "loss": 0.3919,
+      "step": 19376
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975642729521714e-05,
+      "loss": 0.4235,
+      "step": 19377
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9756391254818666e-05,
+      "loss": 0.4752,
+      "step": 19378
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9756355211786897e-05,
+      "loss": 0.4487,
+      "step": 19379
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975631916612184e-05,
+      "loss": 0.4968,
+      "step": 19380
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9756283117823512e-05,
+      "loss": 0.3429,
+      "step": 19381
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975624706689191e-05,
+      "loss": 0.4153,
+      "step": 19382
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975621101332706e-05,
+      "loss": 0.401,
+      "step": 19383
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9756174957128956e-05,
+      "loss": 0.4364,
+      "step": 19384
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975613889829762e-05,
+      "loss": 0.4033,
+      "step": 19385
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9756102836833055e-05,
+      "loss": 0.4462,
+      "step": 19386
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9756066772735272e-05,
+      "loss": 0.51,
+      "step": 19387
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975603070600428e-05,
+      "loss": 0.5072,
+      "step": 19388
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975599463664009e-05,
+      "loss": 0.4544,
+      "step": 19389
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975595856464271e-05,
+      "loss": 0.4237,
+      "step": 19390
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755922490012153e-05,
+      "loss": 0.4324,
+      "step": 19391
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975588641274843e-05,
+      "loss": 0.5347,
+      "step": 19392
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755850332851542e-05,
+      "loss": 0.4618,
+      "step": 19393
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975581425032151e-05,
+      "loss": 0.4839,
+      "step": 19394
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975577816515833e-05,
+      "loss": 0.4752,
+      "step": 19395
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755742077362026e-05,
+      "loss": 0.458,
+      "step": 19396
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755705986932604e-05,
+      "loss": 0.409,
+      "step": 19397
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755669893870064e-05,
+      "loss": 0.4601,
+      "step": 19398
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755633798174428e-05,
+      "loss": 0.4404,
+      "step": 19399
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755597699845702e-05,
+      "loss": 0.4284,
+      "step": 19400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975556159888389e-05,
+      "loss": 0.4187,
+      "step": 19401
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975552549528901e-05,
+      "loss": 0.4823,
+      "step": 19402
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755489389061068e-05,
+      "loss": 0.4397,
+      "step": 19403
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755453280200073e-05,
+      "loss": 0.5282,
+      "step": 19404
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755417168706033e-05,
+      "loss": 0.4631,
+      "step": 19405
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755381054578963e-05,
+      "loss": 0.4405,
+      "step": 19406
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755344937818868e-05,
+      "loss": 0.528,
+      "step": 19407
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755308818425764e-05,
+      "loss": 0.4244,
+      "step": 19408
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975527269639965e-05,
+      "loss": 0.4863,
+      "step": 19409
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755236571740546e-05,
+      "loss": 0.3836,
+      "step": 19410
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975520044444846e-05,
+      "loss": 0.4631,
+      "step": 19411
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.97551643145234e-05,
+      "loss": 0.5602,
+      "step": 19412
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975512818196537e-05,
+      "loss": 0.4419,
+      "step": 19413
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755092046774387e-05,
+      "loss": 0.4138,
+      "step": 19414
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9755055908950463e-05,
+      "loss": 0.6009,
+      "step": 19415
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.97550197684936e-05,
+      "loss": 0.4717,
+      "step": 19416
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754983625403815e-05,
+      "loss": 0.4305,
+      "step": 19417
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754947479681116e-05,
+      "loss": 0.4953,
+      "step": 19418
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754911331325506e-05,
+      "loss": 0.4895,
+      "step": 19419
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754875180337002e-05,
+      "loss": 0.548,
+      "step": 19420
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754839026715608e-05,
+      "loss": 0.4444,
+      "step": 19421
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975480287046134e-05,
+      "loss": 0.497,
+      "step": 19422
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754766711574208e-05,
+      "loss": 0.4682,
+      "step": 19423
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975473055005422e-05,
+      "loss": 0.5074,
+      "step": 19424
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754694385901376e-05,
+      "loss": 0.4678,
+      "step": 19425
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.97546582191157e-05,
+      "loss": 0.3988,
+      "step": 19426
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754622049697197e-05,
+      "loss": 0.4186,
+      "step": 19427
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754585877645875e-05,
+      "loss": 0.5749,
+      "step": 19428
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754549702961743e-05,
+      "loss": 0.4629,
+      "step": 19429
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754513525644812e-05,
+      "loss": 0.5231,
+      "step": 19430
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754477345695094e-05,
+      "loss": 0.4442,
+      "step": 19431
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.97544411631126e-05,
+      "loss": 0.5186,
+      "step": 19432
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975440497789733e-05,
+      "loss": 0.4431,
+      "step": 19433
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754368790049305e-05,
+      "loss": 0.4069,
+      "step": 19434
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975433259956853e-05,
+      "loss": 0.5093,
+      "step": 19435
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754296406455015e-05,
+      "loss": 0.443,
+      "step": 19436
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975426021070877e-05,
+      "loss": 0.4458,
+      "step": 19437
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.97542240123298e-05,
+      "loss": 0.4678,
+      "step": 19438
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754187811318124e-05,
+      "loss": 0.4604,
+      "step": 19439
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754151607673746e-05,
+      "loss": 0.5825,
+      "step": 19440
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754115401396675e-05,
+      "loss": 0.4916,
+      "step": 19441
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754079192486926e-05,
+      "loss": 0.4466,
+      "step": 19442
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754042980944505e-05,
+      "loss": 0.5225,
+      "step": 19443
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9754006766769424e-05,
+      "loss": 0.4091,
+      "step": 19444
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753970549961686e-05,
+      "loss": 0.4392,
+      "step": 19445
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753934330521306e-05,
+      "loss": 0.5021,
+      "step": 19446
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753898108448297e-05,
+      "loss": 0.5384,
+      "step": 19447
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753861883742665e-05,
+      "loss": 0.506,
+      "step": 19448
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975382565640442e-05,
+      "loss": 0.4976,
+      "step": 19449
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975378942643357e-05,
+      "loss": 0.5329,
+      "step": 19450
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975375319383013e-05,
+      "loss": 0.4528,
+      "step": 19451
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.97537169585941e-05,
+      "loss": 0.5253,
+      "step": 19452
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.97536807207255e-05,
+      "loss": 0.471,
+      "step": 19453
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753644480224337e-05,
+      "loss": 0.5408,
+      "step": 19454
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975360823709062e-05,
+      "loss": 0.4423,
+      "step": 19455
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753571991324353e-05,
+      "loss": 0.4497,
+      "step": 19456
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753535742925556e-05,
+      "loss": 0.4273,
+      "step": 19457
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753499491894235e-05,
+      "loss": 0.5363,
+      "step": 19458
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753463238230397e-05,
+      "loss": 0.4994,
+      "step": 19459
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753426981934058e-05,
+      "loss": 0.5159,
+      "step": 19460
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975339072300522e-05,
+      "loss": 0.5009,
+      "step": 19461
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753354461443896e-05,
+      "loss": 0.5767,
+      "step": 19462
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753318197250096e-05,
+      "loss": 0.5113,
+      "step": 19463
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975328193042383e-05,
+      "loss": 0.4509,
+      "step": 19464
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753245660965107e-05,
+      "loss": 0.4699,
+      "step": 19465
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9753209388873938e-05,
+      "loss": 0.4566,
+      "step": 19466
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.975317311415033e-05,
+      "loss": 0.4303,
+      "step": 19467
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.97531368367943e-05,
+      "loss": 0.5206,
+      "step": 19468
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9753100556805845e-05,
+      "loss": 0.4884,
+      "step": 19469
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975306427418499e-05,
+      "loss": 0.5313,
+      "step": 19470
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9753027988931734e-05,
+      "loss": 0.561,
+      "step": 19471
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975299170104609e-05,
+      "loss": 0.5064,
+      "step": 19472
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752955410528073e-05,
+      "loss": 0.4331,
+      "step": 19473
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975291911737768e-05,
+      "loss": 0.4883,
+      "step": 19474
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975288282159493e-05,
+      "loss": 0.5108,
+      "step": 19475
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752846523179835e-05,
+      "loss": 0.4534,
+      "step": 19476
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.97528102221324e-05,
+      "loss": 0.4693,
+      "step": 19477
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752773918452635e-05,
+      "loss": 0.5017,
+      "step": 19478
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752737612140553e-05,
+      "loss": 0.4916,
+      "step": 19479
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752701303196163e-05,
+      "loss": 0.4721,
+      "step": 19480
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975266499161947e-05,
+      "loss": 0.5019,
+      "step": 19481
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752628677410484e-05,
+      "loss": 0.5265,
+      "step": 19482
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752592360569223e-05,
+      "loss": 0.4246,
+      "step": 19483
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752556041095688e-05,
+      "loss": 0.4324,
+      "step": 19484
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752519718989897e-05,
+      "loss": 0.5082,
+      "step": 19485
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752483394251852e-05,
+      "loss": 0.4522,
+      "step": 19486
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752447066881565e-05,
+      "loss": 0.443,
+      "step": 19487
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975241073687905e-05,
+      "loss": 0.3638,
+      "step": 19488
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752374404244312e-05,
+      "loss": 0.4639,
+      "step": 19489
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752338068977367e-05,
+      "loss": 0.5195,
+      "step": 19490
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752301731078213e-05,
+      "loss": 0.4893,
+      "step": 19491
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752265390546873e-05,
+      "loss": 0.4944,
+      "step": 19492
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752229047383348e-05,
+      "loss": 0.4049,
+      "step": 19493
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752192701587655e-05,
+      "loss": 0.5139,
+      "step": 19494
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752156353159793e-05,
+      "loss": 0.4819,
+      "step": 19495
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752120002099783e-05,
+      "loss": 0.4508,
+      "step": 19496
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752083648407628e-05,
+      "loss": 0.502,
+      "step": 19497
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752047292083342e-05,
+      "loss": 0.4533,
+      "step": 19498
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9752010933126932e-05,
+      "loss": 0.4422,
+      "step": 19499
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751974571538407e-05,
+      "loss": 0.44,
+      "step": 19500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975193820731778e-05,
+      "loss": 0.4589,
+      "step": 19501
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751901840465056e-05,
+      "loss": 0.4647,
+      "step": 19502
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751865470980253e-05,
+      "loss": 0.5336,
+      "step": 19503
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751829098863374e-05,
+      "loss": 0.5077,
+      "step": 19504
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975179272411443e-05,
+      "loss": 0.4347,
+      "step": 19505
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751756346733434e-05,
+      "loss": 0.4493,
+      "step": 19506
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975171996672039e-05,
+      "loss": 0.4978,
+      "step": 19507
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751683584075316e-05,
+      "loss": 0.4665,
+      "step": 19508
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751647198798214e-05,
+      "loss": 0.4153,
+      "step": 19509
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751610810889096e-05,
+      "loss": 0.4033,
+      "step": 19510
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751574420347975e-05,
+      "loss": 0.4204,
+      "step": 19511
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751538027174856e-05,
+      "loss": 0.4981,
+      "step": 19512
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751501631369755e-05,
+      "loss": 0.4815,
+      "step": 19513
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975146523293267e-05,
+      "loss": 0.4446,
+      "step": 19514
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975142883186363e-05,
+      "loss": 0.5453,
+      "step": 19515
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751392428162626e-05,
+      "loss": 0.5755,
+      "step": 19516
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751356021829677e-05,
+      "loss": 0.451,
+      "step": 19517
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975131961286479e-05,
+      "loss": 0.5294,
+      "step": 19518
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975128320126798e-05,
+      "loss": 0.4684,
+      "step": 19519
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751246787039254e-05,
+      "loss": 0.4292,
+      "step": 19520
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751210370178618e-05,
+      "loss": 0.4987,
+      "step": 19521
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751173950686084e-05,
+      "loss": 0.5011,
+      "step": 19522
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751137528561664e-05,
+      "loss": 0.4704,
+      "step": 19523
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975110110380537e-05,
+      "loss": 0.4481,
+      "step": 19524
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.97510646764172e-05,
+      "loss": 0.4964,
+      "step": 19525
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9751028246397176e-05,
+      "loss": 0.4871,
+      "step": 19526
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750991813745304e-05,
+      "loss": 0.5039,
+      "step": 19527
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750955378461597e-05,
+      "loss": 0.5022,
+      "step": 19528
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975091894054606e-05,
+      "loss": 0.4462,
+      "step": 19529
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750882499998702e-05,
+      "loss": 0.4403,
+      "step": 19530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750846056819536e-05,
+      "loss": 0.4936,
+      "step": 19531
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975080961100857e-05,
+      "loss": 0.4373,
+      "step": 19532
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750773162565816e-05,
+      "loss": 0.4863,
+      "step": 19533
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750736711491286e-05,
+      "loss": 0.4689,
+      "step": 19534
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975070025778498e-05,
+      "loss": 0.4242,
+      "step": 19535
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750663801446917e-05,
+      "loss": 0.4659,
+      "step": 19536
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975062734247711e-05,
+      "loss": 0.4877,
+      "step": 19537
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750590880875555e-05,
+      "loss": 0.493,
+      "step": 19538
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750554416642274e-05,
+      "loss": 0.474,
+      "step": 19539
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750517949777275e-05,
+      "loss": 0.4634,
+      "step": 19540
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975048148028056e-05,
+      "loss": 0.4687,
+      "step": 19541
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975044500815215e-05,
+      "loss": 0.4461,
+      "step": 19542
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750408533392044e-05,
+      "loss": 0.4474,
+      "step": 19543
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975037205600026e-05,
+      "loss": 0.4414,
+      "step": 19544
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750335575976806e-05,
+      "loss": 0.5357,
+      "step": 19545
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975029909332169e-05,
+      "loss": 0.5348,
+      "step": 19546
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975026260803492e-05,
+      "loss": 0.4532,
+      "step": 19547
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750226120116513e-05,
+      "loss": 0.4724,
+      "step": 19548
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975018962956647e-05,
+      "loss": 0.4106,
+      "step": 19549
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750153136384808e-05,
+      "loss": 0.4181,
+      "step": 19550
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750116640571532e-05,
+      "loss": 0.4786,
+      "step": 19551
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9750080142126658e-05,
+      "loss": 0.4417,
+      "step": 19552
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975004364105019e-05,
+      "loss": 0.5572,
+      "step": 19553
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.975000713734214e-05,
+      "loss": 0.4749,
+      "step": 19554
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9749970631002514e-05,
+      "loss": 0.4889,
+      "step": 19555
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9749934122031327e-05,
+      "loss": 0.5083,
+      "step": 19556
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.974989761042859e-05,
+      "loss": 0.4832,
+      "step": 19557
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.974986109619431e-05,
+      "loss": 0.4773,
+      "step": 19558
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.974982457932849e-05,
+      "loss": 0.5105,
+      "step": 19559
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.974978805983115e-05,
+      "loss": 0.4366,
+      "step": 19560
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9749751537702303e-05,
+      "loss": 0.5363,
+      "step": 19561
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9749715012941948e-05,
+      "loss": 0.3542,
+      "step": 19562
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.97496784855501e-05,
+      "loss": 0.4999,
+      "step": 19563
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9749641955526765e-05,
+      "loss": 0.5433,
+      "step": 19564
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9749605422871958e-05,
+      "loss": 0.4427,
+      "step": 19565
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9749568887585687e-05,
+      "loss": 0.463,
+      "step": 19566
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9749532349667962e-05,
+      "loss": 0.4847,
+      "step": 19567
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974949580911879e-05,
+      "loss": 0.4554,
+      "step": 19568
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974945926593819e-05,
+      "loss": 0.4605,
+      "step": 19569
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9749422720126163e-05,
+      "loss": 0.4557,
+      "step": 19570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974938617168272e-05,
+      "loss": 0.5464,
+      "step": 19571
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974934962060787e-05,
+      "loss": 0.518,
+      "step": 19572
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9749313066901626e-05,
+      "loss": 0.4474,
+      "step": 19573
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9749276510563998e-05,
+      "loss": 0.5009,
+      "step": 19574
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9749239951594994e-05,
+      "loss": 0.4572,
+      "step": 19575
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9749203389994625e-05,
+      "loss": 0.4586,
+      "step": 19576
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.97491668257629e-05,
+      "loss": 0.4429,
+      "step": 19577
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974913025889983e-05,
+      "loss": 0.5444,
+      "step": 19578
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9749093689405426e-05,
+      "loss": 0.4814,
+      "step": 19579
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9749057117279693e-05,
+      "loss": 0.4891,
+      "step": 19580
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9749020542522646e-05,
+      "loss": 0.497,
+      "step": 19581
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974898396513429e-05,
+      "loss": 0.4976,
+      "step": 19582
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974894738511464e-05,
+      "loss": 0.4668,
+      "step": 19583
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748910802463704e-05,
+      "loss": 0.5037,
+      "step": 19584
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974887421718149e-05,
+      "loss": 0.4304,
+      "step": 19585
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974883762926801e-05,
+      "loss": 0.4339,
+      "step": 19586
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748801038723272e-05,
+      "loss": 0.4114,
+      "step": 19587
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748764445547287e-05,
+      "loss": 0.462,
+      "step": 19588
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748727849740068e-05,
+      "loss": 0.4255,
+      "step": 19589
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974869125130162e-05,
+      "loss": 0.5041,
+      "step": 19590
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974865465023195e-05,
+      "loss": 0.4492,
+      "step": 19591
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974861804653108e-05,
+      "loss": 0.4849,
+      "step": 19592
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974858144019901e-05,
+      "loss": 0.4965,
+      "step": 19593
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974854483123575e-05,
+      "loss": 0.5438,
+      "step": 19594
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748508219641313e-05,
+      "loss": 0.5373,
+      "step": 19595
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748471605415708e-05,
+      "loss": 0.4908,
+      "step": 19596
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748434988558944e-05,
+      "loss": 0.3871,
+      "step": 19597
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748398369071036e-05,
+      "loss": 0.4887,
+      "step": 19598
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974836174695199e-05,
+      "loss": 0.4995,
+      "step": 19599
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974832512220181e-05,
+      "loss": 0.4644,
+      "step": 19600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748288494820513e-05,
+      "loss": 0.5256,
+      "step": 19601
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748251864808108e-05,
+      "loss": 0.5653,
+      "step": 19602
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748215232164605e-05,
+      "loss": 0.5936,
+      "step": 19603
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748178596890015e-05,
+      "loss": 0.4475,
+      "step": 19604
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974814195898434e-05,
+      "loss": 0.5198,
+      "step": 19605
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748105318447602e-05,
+      "loss": 0.516,
+      "step": 19606
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748068675279803e-05,
+      "loss": 0.482,
+      "step": 19607
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9748032029480953e-05,
+      "loss": 0.4777,
+      "step": 19608
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747995381051067e-05,
+      "loss": 0.5114,
+      "step": 19609
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747958729990148e-05,
+      "loss": 0.4946,
+      "step": 19610
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974792207629821e-05,
+      "loss": 0.471,
+      "step": 19611
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747885419975263e-05,
+      "loss": 0.5971,
+      "step": 19612
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747848761021314e-05,
+      "loss": 0.4094,
+      "step": 19613
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974781209943638e-05,
+      "loss": 0.4179,
+      "step": 19614
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747775435220462e-05,
+      "loss": 0.4383,
+      "step": 19615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747738768373576e-05,
+      "loss": 0.4551,
+      "step": 19616
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747702098895728e-05,
+      "loss": 0.466,
+      "step": 19617
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974766542678693e-05,
+      "loss": 0.4657,
+      "step": 19618
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747628752047195e-05,
+      "loss": 0.4836,
+      "step": 19619
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747592074676524e-05,
+      "loss": 0.4149,
+      "step": 19620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747555394674935e-05,
+      "loss": 0.4587,
+      "step": 19621
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747518712042434e-05,
+      "loss": 0.4889,
+      "step": 19622
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747482026779036e-05,
+      "loss": 0.4302,
+      "step": 19623
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747445338884743e-05,
+      "loss": 0.4519,
+      "step": 19624
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974740864835957e-05,
+      "loss": 0.4781,
+      "step": 19625
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747371955203528e-05,
+      "loss": 0.3721,
+      "step": 19626
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974733525941662e-05,
+      "loss": 0.4589,
+      "step": 19627
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747298560998863e-05,
+      "loss": 0.492,
+      "step": 19628
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747261859950267e-05,
+      "loss": 0.4707,
+      "step": 19629
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747225156270838e-05,
+      "loss": 0.4648,
+      "step": 19630
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747188449960585e-05,
+      "loss": 0.3895,
+      "step": 19631
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747151741019522e-05,
+      "loss": 0.4549,
+      "step": 19632
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747115029447656e-05,
+      "loss": 0.5351,
+      "step": 19633
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9747078315245e-05,
+      "loss": 0.4875,
+      "step": 19634
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974704159841156e-05,
+      "loss": 0.4961,
+      "step": 19635
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974700487894735e-05,
+      "loss": 0.4372,
+      "step": 19636
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746968156852377e-05,
+      "loss": 0.5481,
+      "step": 19637
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746931432126652e-05,
+      "loss": 0.4329,
+      "step": 19638
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746894704770185e-05,
+      "loss": 0.4944,
+      "step": 19639
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746857974782986e-05,
+      "loss": 0.4938,
+      "step": 19640
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746821242165062e-05,
+      "loss": 0.5057,
+      "step": 19641
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746784506916427e-05,
+      "loss": 0.4634,
+      "step": 19642
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974674776903709e-05,
+      "loss": 0.4991,
+      "step": 19643
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974671102852706e-05,
+      "loss": 0.5105,
+      "step": 19644
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746674285386346e-05,
+      "loss": 0.4378,
+      "step": 19645
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746637539614956e-05,
+      "loss": 0.5109,
+      "step": 19646
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974660079121291e-05,
+      "loss": 0.4518,
+      "step": 19647
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746564040180204e-05,
+      "loss": 0.4825,
+      "step": 19648
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746527286516862e-05,
+      "loss": 0.4364,
+      "step": 19649
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746490530222883e-05,
+      "loss": 0.4038,
+      "step": 19650
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746453771298277e-05,
+      "loss": 0.4216,
+      "step": 19651
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746417009743065e-05,
+      "loss": 0.4827,
+      "step": 19652
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746380245557243e-05,
+      "loss": 0.4043,
+      "step": 19653
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974634347874083e-05,
+      "loss": 0.4674,
+      "step": 19654
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974630670929383e-05,
+      "loss": 0.4165,
+      "step": 19655
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746269937216263e-05,
+      "loss": 0.4751,
+      "step": 19656
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974623316250813e-05,
+      "loss": 0.444,
+      "step": 19657
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746196385169442e-05,
+      "loss": 0.4803,
+      "step": 19658
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974615960520021e-05,
+      "loss": 0.4386,
+      "step": 19659
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746122822600442e-05,
+      "loss": 0.5098,
+      "step": 19660
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.974608603737015e-05,
+      "loss": 0.5275,
+      "step": 19661
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746049249509348e-05,
+      "loss": 0.5042,
+      "step": 19662
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9746012459018037e-05,
+      "loss": 0.4835,
+      "step": 19663
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9745975665896234e-05,
+      "loss": 0.4383,
+      "step": 19664
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974593887014395e-05,
+      "loss": 0.4188,
+      "step": 19665
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745902071761186e-05,
+      "loss": 0.4824,
+      "step": 19666
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974586527074796e-05,
+      "loss": 0.4784,
+      "step": 19667
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974582846710428e-05,
+      "loss": 0.4763,
+      "step": 19668
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745791660830154e-05,
+      "loss": 0.447,
+      "step": 19669
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745754851925593e-05,
+      "loss": 0.4768,
+      "step": 19670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974571804039061e-05,
+      "loss": 0.391,
+      "step": 19671
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745681226225208e-05,
+      "loss": 0.4663,
+      "step": 19672
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745644409429404e-05,
+      "loss": 0.506,
+      "step": 19673
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745607590003204e-05,
+      "loss": 0.5161,
+      "step": 19674
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745570767946617e-05,
+      "loss": 0.456,
+      "step": 19675
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745533943259657e-05,
+      "loss": 0.4797,
+      "step": 19676
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745497115942335e-05,
+      "loss": 0.5286,
+      "step": 19677
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745460285994654e-05,
+      "loss": 0.4815,
+      "step": 19678
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745423453416627e-05,
+      "loss": 0.554,
+      "step": 19679
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745386618208268e-05,
+      "loss": 0.4789,
+      "step": 19680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974534978036958e-05,
+      "loss": 0.4875,
+      "step": 19681
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745312939900577e-05,
+      "loss": 0.6047,
+      "step": 19682
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745276096801273e-05,
+      "loss": 0.5251,
+      "step": 19683
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745239251071667e-05,
+      "loss": 0.5235,
+      "step": 19684
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974520240271178e-05,
+      "loss": 0.4877,
+      "step": 19685
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745165551721615e-05,
+      "loss": 0.5275,
+      "step": 19686
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745128698101185e-05,
+      "loss": 0.4139,
+      "step": 19687
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.97450918418505e-05,
+      "loss": 0.473,
+      "step": 19688
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9745054982969568e-05,
+      "loss": 0.419,
+      "step": 19689
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.97450181214584e-05,
+      "loss": 0.5729,
+      "step": 19690
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744981257317005e-05,
+      "loss": 0.5612,
+      "step": 19691
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744944390545396e-05,
+      "loss": 0.4584,
+      "step": 19692
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974490752114358e-05,
+      "loss": 0.3952,
+      "step": 19693
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974487064911157e-05,
+      "loss": 0.5226,
+      "step": 19694
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974483377444937e-05,
+      "loss": 0.4671,
+      "step": 19695
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744796897156996e-05,
+      "loss": 0.4388,
+      "step": 19696
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744760017234456e-05,
+      "loss": 0.4456,
+      "step": 19697
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974472313468176e-05,
+      "loss": 0.5271,
+      "step": 19698
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744686249498917e-05,
+      "loss": 0.4514,
+      "step": 19699
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744649361685936e-05,
+      "loss": 0.4968,
+      "step": 19700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744612471242833e-05,
+      "loss": 0.509,
+      "step": 19701
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744575578169608e-05,
+      "loss": 0.4624,
+      "step": 19702
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974453868246628e-05,
+      "loss": 0.4888,
+      "step": 19703
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744501784132856e-05,
+      "loss": 0.5195,
+      "step": 19704
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744464883169347e-05,
+      "loss": 0.4839,
+      "step": 19705
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744427979575756e-05,
+      "loss": 0.581,
+      "step": 19706
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.97443910733521e-05,
+      "loss": 0.5034,
+      "step": 19707
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974435416449839e-05,
+      "loss": 0.4559,
+      "step": 19708
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974431725301463e-05,
+      "loss": 0.4189,
+      "step": 19709
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744280338900835e-05,
+      "loss": 0.5263,
+      "step": 19710
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744243422157014e-05,
+      "loss": 0.5305,
+      "step": 19711
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744206502783175e-05,
+      "loss": 0.47,
+      "step": 19712
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974416958077933e-05,
+      "loss": 0.5531,
+      "step": 19713
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744132656145488e-05,
+      "loss": 0.4247,
+      "step": 19714
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9744095728881656e-05,
+      "loss": 0.3703,
+      "step": 19715
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974405879898785e-05,
+      "loss": 0.3077,
+      "step": 19716
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974402186646408e-05,
+      "loss": 0.376,
+      "step": 19717
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743984931310346e-05,
+      "loss": 0.2655,
+      "step": 19718
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974394799352667e-05,
+      "loss": 0.282,
+      "step": 19719
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743911053113054e-05,
+      "loss": 0.36,
+      "step": 19720
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974387411006951e-05,
+      "loss": 0.3069,
+      "step": 19721
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974383716439605e-05,
+      "loss": 0.3115,
+      "step": 19722
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743800216092686e-05,
+      "loss": 0.3245,
+      "step": 19723
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743763265159423e-05,
+      "loss": 0.3324,
+      "step": 19724
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743726311596272e-05,
+      "loss": 0.266,
+      "step": 19725
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743689355403243e-05,
+      "loss": 0.3343,
+      "step": 19726
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974365239658035e-05,
+      "loss": 0.2818,
+      "step": 19727
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743615435127594e-05,
+      "loss": 0.3564,
+      "step": 19728
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743578471044995e-05,
+      "loss": 0.3285,
+      "step": 19729
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974354150433256e-05,
+      "loss": 0.3202,
+      "step": 19730
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743504534990295e-05,
+      "loss": 0.292,
+      "step": 19731
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974346756301821e-05,
+      "loss": 0.3072,
+      "step": 19732
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743430588416323e-05,
+      "loss": 0.2831,
+      "step": 19733
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743393611184635e-05,
+      "loss": 0.2522,
+      "step": 19734
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743356631323157e-05,
+      "loss": 0.2806,
+      "step": 19735
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743319648831906e-05,
+      "loss": 0.3246,
+      "step": 19736
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974328266371089e-05,
+      "loss": 0.3425,
+      "step": 19737
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974324567596011e-05,
+      "loss": 0.3386,
+      "step": 19738
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743208685579587e-05,
+      "loss": 0.2882,
+      "step": 19739
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743171692569322e-05,
+      "loss": 0.2695,
+      "step": 19740
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974313469692933e-05,
+      "loss": 0.2902,
+      "step": 19741
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743097698659625e-05,
+      "loss": 0.3071,
+      "step": 19742
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974306069776021e-05,
+      "loss": 0.3261,
+      "step": 19743
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9743023694231097e-05,
+      "loss": 0.2896,
+      "step": 19744
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.97429866880723e-05,
+      "loss": 0.3226,
+      "step": 19745
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974294967928382e-05,
+      "loss": 0.2909,
+      "step": 19746
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9742912667865672e-05,
+      "loss": 0.2763,
+      "step": 19747
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974287565381787e-05,
+      "loss": 0.2974,
+      "step": 19748
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974283863714042e-05,
+      "loss": 0.3276,
+      "step": 19749
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974280161783333e-05,
+      "loss": 0.3263,
+      "step": 19750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9742764595896615e-05,
+      "loss": 0.3567,
+      "step": 19751
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974272757133028e-05,
+      "loss": 0.2966,
+      "step": 19752
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9742690544134337e-05,
+      "loss": 0.3101,
+      "step": 19753
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9742653514308798e-05,
+      "loss": 0.3089,
+      "step": 19754
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9742616481853672e-05,
+      "loss": 0.3415,
+      "step": 19755
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9742579446768967e-05,
+      "loss": 0.3602,
+      "step": 19756
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974254240905469e-05,
+      "loss": 0.2897,
+      "step": 19757
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974250536871086e-05,
+      "loss": 0.352,
+      "step": 19758
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9742468325737482e-05,
+      "loss": 0.2772,
+      "step": 19759
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974243128013457e-05,
+      "loss": 0.364,
+      "step": 19760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9742394231902123e-05,
+      "loss": 0.2668,
+      "step": 19761
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9742357181040162e-05,
+      "loss": 0.2908,
+      "step": 19762
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.974232012754869e-05,
+      "loss": 0.2862,
+      "step": 19763
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9742283071427726e-05,
+      "loss": 0.271,
+      "step": 19764
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9742246012677272e-05,
+      "loss": 0.2975,
+      "step": 19765
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974220895129734e-05,
+      "loss": 0.2868,
+      "step": 19766
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974217188728794e-05,
+      "loss": 0.2756,
+      "step": 19767
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974213482064908e-05,
+      "loss": 0.2509,
+      "step": 19768
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9742097751380776e-05,
+      "loss": 0.3357,
+      "step": 19769
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9742060679483033e-05,
+      "loss": 0.3097,
+      "step": 19770
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974202360495586e-05,
+      "loss": 0.2739,
+      "step": 19771
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974198652779927e-05,
+      "loss": 0.2955,
+      "step": 19772
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741949448013272e-05,
+      "loss": 0.2638,
+      "step": 19773
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741912365597883e-05,
+      "loss": 0.323,
+      "step": 19774
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.97418752805531e-05,
+      "loss": 0.321,
+      "step": 19775
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741838192878938e-05,
+      "loss": 0.2762,
+      "step": 19776
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974180110257541e-05,
+      "loss": 0.2913,
+      "step": 19777
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741764009642524e-05,
+      "loss": 0.3184,
+      "step": 19778
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741726914080296e-05,
+      "loss": 0.3079,
+      "step": 19779
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741689815888722e-05,
+      "loss": 0.3204,
+      "step": 19780
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741652715067826e-05,
+      "loss": 0.2497,
+      "step": 19781
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974161561161761e-05,
+      "loss": 0.3415,
+      "step": 19782
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974157850553809e-05,
+      "loss": 0.3078,
+      "step": 19783
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741541396829267e-05,
+      "loss": 0.3315,
+      "step": 19784
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974150428549116e-05,
+      "loss": 0.302,
+      "step": 19785
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741467171523772e-05,
+      "loss": 0.2897,
+      "step": 19786
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974143005492712e-05,
+      "loss": 0.3232,
+      "step": 19787
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974139293570121e-05,
+      "loss": 0.2341,
+      "step": 19788
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974135581384605e-05,
+      "loss": 0.3052,
+      "step": 19789
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741318689361657e-05,
+      "loss": 0.3285,
+      "step": 19790
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741281562248034e-05,
+      "loss": 0.293,
+      "step": 19791
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741244432505194e-05,
+      "loss": 0.2531,
+      "step": 19792
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741207300133144e-05,
+      "loss": 0.2686,
+      "step": 19793
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741170165131897e-05,
+      "loss": 0.331,
+      "step": 19794
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741133027501466e-05,
+      "loss": 0.2785,
+      "step": 19795
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741095887241855e-05,
+      "loss": 0.2488,
+      "step": 19796
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741058744353077e-05,
+      "loss": 0.3047,
+      "step": 19797
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9741021598835143e-05,
+      "loss": 0.301,
+      "step": 19798
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974098445068806e-05,
+      "loss": 0.2625,
+      "step": 19799
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740947299911844e-05,
+      "loss": 0.3162,
+      "step": 19800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740910146506498e-05,
+      "loss": 0.2834,
+      "step": 19801
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740872990472033e-05,
+      "loss": 0.2449,
+      "step": 19802
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740835831808463e-05,
+      "loss": 0.3339,
+      "step": 19803
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740798670515794e-05,
+      "loss": 0.2929,
+      "step": 19804
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974076150659404e-05,
+      "loss": 0.318,
+      "step": 19805
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740724340043208e-05,
+      "loss": 0.2466,
+      "step": 19806
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974068717086331e-05,
+      "loss": 0.3064,
+      "step": 19807
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740649999054354e-05,
+      "loss": 0.2816,
+      "step": 19808
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974061282461635e-05,
+      "loss": 0.3007,
+      "step": 19809
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974057564754931e-05,
+      "loss": 0.3181,
+      "step": 19810
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740538467853243e-05,
+      "loss": 0.2784,
+      "step": 19811
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740501285528157e-05,
+      "loss": 0.2467,
+      "step": 19812
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974046410057407e-05,
+      "loss": 0.2835,
+      "step": 19813
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.974042691299098e-05,
+      "loss": 0.2858,
+      "step": 19814
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740389722778907e-05,
+      "loss": 0.2586,
+      "step": 19815
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740352529937856e-05,
+      "loss": 0.3155,
+      "step": 19816
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740315334467838e-05,
+      "loss": 0.3434,
+      "step": 19817
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740278136368863e-05,
+      "loss": 0.2912,
+      "step": 19818
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740240935640942e-05,
+      "loss": 0.2861,
+      "step": 19819
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740203732284085e-05,
+      "loss": 0.2734,
+      "step": 19820
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740166526298302e-05,
+      "loss": 0.2898,
+      "step": 19821
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740129317683603e-05,
+      "loss": 0.2445,
+      "step": 19822
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740092106439996e-05,
+      "loss": 0.2818,
+      "step": 19823
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740054892567492e-05,
+      "loss": 0.33,
+      "step": 19824
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9740017676066103e-05,
+      "loss": 0.2833,
+      "step": 19825
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739980456935836e-05,
+      "loss": 0.2587,
+      "step": 19826
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739943235176707e-05,
+      "loss": 0.2963,
+      "step": 19827
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.973990601078872e-05,
+      "loss": 0.3292,
+      "step": 19828
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739868783771885e-05,
+      "loss": 0.3148,
+      "step": 19829
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.973983155412621e-05,
+      "loss": 0.2771,
+      "step": 19830
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739794321851717e-05,
+      "loss": 0.279,
+      "step": 19831
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739757086948402e-05,
+      "loss": 0.2755,
+      "step": 19832
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739719849416287e-05,
+      "loss": 0.2725,
+      "step": 19833
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.973968260925537e-05,
+      "loss": 0.2756,
+      "step": 19834
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.973964536646567e-05,
+      "loss": 0.356,
+      "step": 19835
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739608121047197e-05,
+      "loss": 0.363,
+      "step": 19836
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739570872999953e-05,
+      "loss": 0.2681,
+      "step": 19837
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739533622323956e-05,
+      "loss": 0.3617,
+      "step": 19838
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739496369019213e-05,
+      "loss": 0.3088,
+      "step": 19839
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739459113085733e-05,
+      "loss": 0.3262,
+      "step": 19840
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739421854523527e-05,
+      "loss": 0.3395,
+      "step": 19841
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.973938459333261e-05,
+      "loss": 0.3631,
+      "step": 19842
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739347329512982e-05,
+      "loss": 0.2777,
+      "step": 19843
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739310063064662e-05,
+      "loss": 0.2827,
+      "step": 19844
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739272793987657e-05,
+      "loss": 0.2952,
+      "step": 19845
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739235522281977e-05,
+      "loss": 0.2358,
+      "step": 19846
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739198247947628e-05,
+      "loss": 0.2598,
+      "step": 19847
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739160970984628e-05,
+      "loss": 0.2716,
+      "step": 19848
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739123691392983e-05,
+      "loss": 0.2917,
+      "step": 19849
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.97390864091727e-05,
+      "loss": 0.3289,
+      "step": 19850
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9739049124323792e-05,
+      "loss": 0.2836,
+      "step": 19851
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.973901183684627e-05,
+      "loss": 0.2907,
+      "step": 19852
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9738974546740148e-05,
+      "loss": 0.2821,
+      "step": 19853
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9738937254005424e-05,
+      "loss": 0.278,
+      "step": 19854
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.973889995864212e-05,
+      "loss": 0.2777,
+      "step": 19855
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9738862660650236e-05,
+      "loss": 0.2944,
+      "step": 19856
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9738825360029796e-05,
+      "loss": 0.2832,
+      "step": 19857
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9738788056780795e-05,
+      "loss": 0.2793,
+      "step": 19858
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.973875075090325e-05,
+      "loss": 0.2927,
+      "step": 19859
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9738713442397174e-05,
+      "loss": 0.3129,
+      "step": 19860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9738676131262567e-05,
+      "loss": 0.3544,
+      "step": 19861
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738638817499455e-05,
+      "loss": 0.2811,
+      "step": 19862
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738601501107833e-05,
+      "loss": 0.3615,
+      "step": 19863
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973856418208772e-05,
+      "loss": 0.2805,
+      "step": 19864
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973852686043912e-05,
+      "loss": 0.296,
+      "step": 19865
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738489536162048e-05,
+      "loss": 0.3225,
+      "step": 19866
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738452209256508e-05,
+      "loss": 0.248,
+      "step": 19867
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973841487972252e-05,
+      "loss": 0.3388,
+      "step": 19868
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738377547560086e-05,
+      "loss": 0.3246,
+      "step": 19869
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738340212769218e-05,
+      "loss": 0.2834,
+      "step": 19870
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738302875349925e-05,
+      "loss": 0.2963,
+      "step": 19871
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738265535302222e-05,
+      "loss": 0.2802,
+      "step": 19872
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738228192626114e-05,
+      "loss": 0.2752,
+      "step": 19873
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738190847321612e-05,
+      "loss": 0.3644,
+      "step": 19874
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738153499388726e-05,
+      "loss": 0.2663,
+      "step": 19875
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738116148827467e-05,
+      "loss": 0.2952,
+      "step": 19876
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973807879563785e-05,
+      "loss": 0.2764,
+      "step": 19877
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738041439819874e-05,
+      "loss": 0.3298,
+      "step": 19878
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9738004081373558e-05,
+      "loss": 0.3258,
+      "step": 19879
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737966720298905e-05,
+      "loss": 0.308,
+      "step": 19880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737929356595937e-05,
+      "loss": 0.306,
+      "step": 19881
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973789199026465e-05,
+      "loss": 0.269,
+      "step": 19882
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737854621305064e-05,
+      "loss": 0.2783,
+      "step": 19883
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973781724971718e-05,
+      "loss": 0.2932,
+      "step": 19884
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973777987550102e-05,
+      "loss": 0.3331,
+      "step": 19885
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737742498656587e-05,
+      "loss": 0.307,
+      "step": 19886
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737705119183888e-05,
+      "loss": 0.305,
+      "step": 19887
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973766773708294e-05,
+      "loss": 0.284,
+      "step": 19888
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737630352353747e-05,
+      "loss": 0.3067,
+      "step": 19889
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737592964996326e-05,
+      "loss": 0.3007,
+      "step": 19890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973755557501068e-05,
+      "loss": 0.2625,
+      "step": 19891
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737518182396824e-05,
+      "loss": 0.2413,
+      "step": 19892
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737480787154766e-05,
+      "loss": 0.2723,
+      "step": 19893
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737443389284517e-05,
+      "loss": 0.3202,
+      "step": 19894
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737405988786085e-05,
+      "loss": 0.2784,
+      "step": 19895
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973736858565948e-05,
+      "loss": 0.2784,
+      "step": 19896
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973733117990472e-05,
+      "loss": 0.3498,
+      "step": 19897
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737293771521802e-05,
+      "loss": 0.2846,
+      "step": 19898
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737256360510746e-05,
+      "loss": 0.2926,
+      "step": 19899
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973721894687156e-05,
+      "loss": 0.3076,
+      "step": 19900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973718153060425e-05,
+      "loss": 0.2755,
+      "step": 19901
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973714411170883e-05,
+      "loss": 0.3191,
+      "step": 19902
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973710669018531e-05,
+      "loss": 0.326,
+      "step": 19903
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9737069266033702e-05,
+      "loss": 0.26,
+      "step": 19904
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973703183925401e-05,
+      "loss": 0.3077,
+      "step": 19905
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736994409846247e-05,
+      "loss": 0.2857,
+      "step": 19906
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736956977810427e-05,
+      "loss": 0.2678,
+      "step": 19907
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736919543146555e-05,
+      "loss": 0.3494,
+      "step": 19908
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973688210585464e-05,
+      "loss": 0.353,
+      "step": 19909
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.97368446659347e-05,
+      "loss": 0.2914,
+      "step": 19910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973680722338674e-05,
+      "loss": 0.2512,
+      "step": 19911
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736769778210764e-05,
+      "loss": 0.2419,
+      "step": 19912
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736732330406795e-05,
+      "loss": 0.3502,
+      "step": 19913
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736694879974835e-05,
+      "loss": 0.3338,
+      "step": 19914
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736657426914894e-05,
+      "loss": 0.295,
+      "step": 19915
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736619971226983e-05,
+      "loss": 0.2658,
+      "step": 19916
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736582512911115e-05,
+      "loss": 0.2779,
+      "step": 19917
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736545051967297e-05,
+      "loss": 0.3222,
+      "step": 19918
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973650758839554e-05,
+      "loss": 0.309,
+      "step": 19919
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973647012219585e-05,
+      "loss": 0.2858,
+      "step": 19920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736432653368246e-05,
+      "loss": 0.2651,
+      "step": 19921
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736395181912735e-05,
+      "loss": 0.3202,
+      "step": 19922
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736357707829322e-05,
+      "loss": 0.2657,
+      "step": 19923
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736320231118023e-05,
+      "loss": 0.2616,
+      "step": 19924
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736282751778844e-05,
+      "loss": 0.2726,
+      "step": 19925
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736245269811797e-05,
+      "loss": 0.3463,
+      "step": 19926
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973620778521689e-05,
+      "loss": 0.2756,
+      "step": 19927
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736170297994136e-05,
+      "loss": 0.3547,
+      "step": 19928
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736132808143547e-05,
+      "loss": 0.2715,
+      "step": 19929
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973609531566513e-05,
+      "loss": 0.3646,
+      "step": 19930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736057820558893e-05,
+      "loss": 0.2598,
+      "step": 19931
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9736020322824846e-05,
+      "loss": 0.2687,
+      "step": 19932
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735982822463008e-05,
+      "loss": 0.321,
+      "step": 19933
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973594531947338e-05,
+      "loss": 0.3761,
+      "step": 19934
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973590781385597e-05,
+      "loss": 0.2447,
+      "step": 19935
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.97358703056108e-05,
+      "loss": 0.3195,
+      "step": 19936
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735832794737868e-05,
+      "loss": 0.288,
+      "step": 19937
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973579528123719e-05,
+      "loss": 0.2873,
+      "step": 19938
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973575776510878e-05,
+      "loss": 0.3452,
+      "step": 19939
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735720246352637e-05,
+      "loss": 0.3094,
+      "step": 19940
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735682724968783e-05,
+      "loss": 0.3178,
+      "step": 19941
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973564520095722e-05,
+      "loss": 0.2839,
+      "step": 19942
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973560767431796e-05,
+      "loss": 0.3032,
+      "step": 19943
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735570145051017e-05,
+      "loss": 0.3324,
+      "step": 19944
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735532613156395e-05,
+      "loss": 0.3651,
+      "step": 19945
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973549507863411e-05,
+      "loss": 0.2451,
+      "step": 19946
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973545754148417e-05,
+      "loss": 0.3194,
+      "step": 19947
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973542000170658e-05,
+      "loss": 0.2987,
+      "step": 19948
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735382459301357e-05,
+      "loss": 0.2577,
+      "step": 19949
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973534491426851e-05,
+      "loss": 0.307,
+      "step": 19950
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735307366608046e-05,
+      "loss": 0.3415,
+      "step": 19951
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973526981631998e-05,
+      "loss": 0.3243,
+      "step": 19952
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735232263404317e-05,
+      "loss": 0.3564,
+      "step": 19953
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735194707861067e-05,
+      "loss": 0.2974,
+      "step": 19954
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735157149690244e-05,
+      "loss": 0.3401,
+      "step": 19955
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.973511958889186e-05,
+      "loss": 0.2748,
+      "step": 19956
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735082025465918e-05,
+      "loss": 0.3299,
+      "step": 19957
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735044459412433e-05,
+      "loss": 0.3498,
+      "step": 19958
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9735006890731414e-05,
+      "loss": 0.2982,
+      "step": 19959
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9734969319422872e-05,
+      "loss": 0.3036,
+      "step": 19960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734931745486814e-05,
+      "loss": 0.2639,
+      "step": 19961
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734894168923255e-05,
+      "loss": 0.3081,
+      "step": 19962
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.97348565897322e-05,
+      "loss": 0.3227,
+      "step": 19963
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734819007913664e-05,
+      "loss": 0.2756,
+      "step": 19964
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973478142346765e-05,
+      "loss": 0.2651,
+      "step": 19965
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973474383639418e-05,
+      "loss": 0.2885,
+      "step": 19966
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734706246693256e-05,
+      "loss": 0.2725,
+      "step": 19967
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734668654364887e-05,
+      "loss": 0.3379,
+      "step": 19968
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734631059409086e-05,
+      "loss": 0.2912,
+      "step": 19969
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734593461825863e-05,
+      "loss": 0.3202,
+      "step": 19970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734555861615225e-05,
+      "loss": 0.3089,
+      "step": 19971
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734518258777186e-05,
+      "loss": 0.2963,
+      "step": 19972
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734480653311756e-05,
+      "loss": 0.3088,
+      "step": 19973
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734443045218945e-05,
+      "loss": 0.3575,
+      "step": 19974
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734405434498763e-05,
+      "loss": 0.3402,
+      "step": 19975
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734367821151217e-05,
+      "loss": 0.2659,
+      "step": 19976
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973433020517632e-05,
+      "loss": 0.324,
+      "step": 19977
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734292586574083e-05,
+      "loss": 0.2772,
+      "step": 19978
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734254965344513e-05,
+      "loss": 0.2613,
+      "step": 19979
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734217341487626e-05,
+      "loss": 0.2704,
+      "step": 19980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734179715003422e-05,
+      "loss": 0.3121,
+      "step": 19981
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734142085891922e-05,
+      "loss": 0.2804,
+      "step": 19982
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734104454153132e-05,
+      "loss": 0.2531,
+      "step": 19983
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973406681978706e-05,
+      "loss": 0.3162,
+      "step": 19984
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9734029182793715e-05,
+      "loss": 0.3241,
+      "step": 19985
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733991543173118e-05,
+      "loss": 0.3044,
+      "step": 19986
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733953900925265e-05,
+      "loss": 0.3163,
+      "step": 19987
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973391625605017e-05,
+      "loss": 0.2616,
+      "step": 19988
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973387860854785e-05,
+      "loss": 0.3276,
+      "step": 19989
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733840958418312e-05,
+      "loss": 0.2782,
+      "step": 19990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733803305661562e-05,
+      "loss": 0.3279,
+      "step": 19991
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973376565027761e-05,
+      "loss": 0.2811,
+      "step": 19992
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733727992266474e-05,
+      "loss": 0.2926,
+      "step": 19993
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973369033162816e-05,
+      "loss": 0.2685,
+      "step": 19994
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733652668362676e-05,
+      "loss": 0.2544,
+      "step": 19995
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973361500247003e-05,
+      "loss": 0.383,
+      "step": 19996
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973357733395024e-05,
+      "loss": 0.2985,
+      "step": 19997
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733539662803312e-05,
+      "loss": 0.3222,
+      "step": 19998
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733501989029255e-05,
+      "loss": 0.3156,
+      "step": 19999
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733464312628082e-05,
+      "loss": 0.2916,
+      "step": 20000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.97334266335998e-05,
+      "loss": 0.2987,
+      "step": 20001
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973338895194442e-05,
+      "loss": 0.2561,
+      "step": 20002
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973335126766195e-05,
+      "loss": 0.3816,
+      "step": 20003
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733313580752408e-05,
+      "loss": 0.3903,
+      "step": 20004
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.97332758912158e-05,
+      "loss": 0.2973,
+      "step": 20005
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733238199052133e-05,
+      "loss": 0.2876,
+      "step": 20006
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733200504261417e-05,
+      "loss": 0.279,
+      "step": 20007
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733162806843665e-05,
+      "loss": 0.3342,
+      "step": 20008
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733125106798892e-05,
+      "loss": 0.3089,
+      "step": 20009
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9733087404127097e-05,
+      "loss": 0.3033,
+      "step": 20010
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.97330496988283e-05,
+      "loss": 0.2845,
+      "step": 20011
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973301199090251e-05,
+      "loss": 0.2846,
+      "step": 20012
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973297428034973e-05,
+      "loss": 0.2881,
+      "step": 20013
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732936567169972e-05,
+      "loss": 0.3056,
+      "step": 20014
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732898851363257e-05,
+      "loss": 0.2617,
+      "step": 20015
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732861132929582e-05,
+      "loss": 0.2896,
+      "step": 20016
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732823411868966e-05,
+      "loss": 0.2733,
+      "step": 20017
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973278568818141e-05,
+      "loss": 0.2898,
+      "step": 20018
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732747961866932e-05,
+      "loss": 0.3413,
+      "step": 20019
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973271023292554e-05,
+      "loss": 0.335,
+      "step": 20020
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732672501357245e-05,
+      "loss": 0.3122,
+      "step": 20021
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732634767162055e-05,
+      "loss": 0.2741,
+      "step": 20022
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732597030339978e-05,
+      "loss": 0.2538,
+      "step": 20023
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732559290891036e-05,
+      "loss": 0.3007,
+      "step": 20024
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732521548815224e-05,
+      "loss": 0.2961,
+      "step": 20025
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732483804112563e-05,
+      "loss": 0.3119,
+      "step": 20026
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732446056783056e-05,
+      "loss": 0.2918,
+      "step": 20027
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732408306826717e-05,
+      "loss": 0.34,
+      "step": 20028
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732370554243557e-05,
+      "loss": 0.3446,
+      "step": 20029
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973233279903358e-05,
+      "loss": 0.2694,
+      "step": 20030
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732295041196806e-05,
+      "loss": 0.3047,
+      "step": 20031
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732257280733237e-05,
+      "loss": 0.3071,
+      "step": 20032
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973221951764289e-05,
+      "loss": 0.3376,
+      "step": 20033
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973218175192577e-05,
+      "loss": 0.3115,
+      "step": 20034
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973214398358189e-05,
+      "loss": 0.3033,
+      "step": 20035
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732106212611253e-05,
+      "loss": 0.3125,
+      "step": 20036
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973206843901388e-05,
+      "loss": 0.2872,
+      "step": 20037
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9732030662789775e-05,
+      "loss": 0.285,
+      "step": 20038
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731992883938947e-05,
+      "loss": 0.3439,
+      "step": 20039
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731955102461412e-05,
+      "loss": 0.2791,
+      "step": 20040
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973191731835718e-05,
+      "loss": 0.2817,
+      "step": 20041
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731879531626252e-05,
+      "loss": 0.3186,
+      "step": 20042
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731841742268645e-05,
+      "loss": 0.2672,
+      "step": 20043
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973180395028437e-05,
+      "loss": 0.2876,
+      "step": 20044
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731766155673434e-05,
+      "loss": 0.2992,
+      "step": 20045
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731728358435852e-05,
+      "loss": 0.2894,
+      "step": 20046
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973169055857163e-05,
+      "loss": 0.3073,
+      "step": 20047
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731652756080777e-05,
+      "loss": 0.2855,
+      "step": 20048
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731614950963308e-05,
+      "loss": 0.2306,
+      "step": 20049
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731577143219227e-05,
+      "loss": 0.3322,
+      "step": 20050
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973153933284855e-05,
+      "loss": 0.3659,
+      "step": 20051
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731501519851288e-05,
+      "loss": 0.2937,
+      "step": 20052
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731463704227446e-05,
+      "loss": 0.3187,
+      "step": 20053
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731425885977032e-05,
+      "loss": 0.3372,
+      "step": 20054
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731388065100065e-05,
+      "loss": 0.3092,
+      "step": 20055
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731350241596553e-05,
+      "loss": 0.2827,
+      "step": 20056
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731312415466498e-05,
+      "loss": 0.2932,
+      "step": 20057
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9731274586709925e-05,
+      "loss": 0.3223,
+      "step": 20058
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9731236755326828e-05,
+      "loss": 0.3686,
+      "step": 20059
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9731198921317228e-05,
+      "loss": 0.2725,
+      "step": 20060
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.973116108468113e-05,
+      "loss": 0.3672,
+      "step": 20061
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9731123245418548e-05,
+      "loss": 0.2768,
+      "step": 20062
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9731085403529488e-05,
+      "loss": 0.3414,
+      "step": 20063
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9731047559013966e-05,
+      "loss": 0.3426,
+      "step": 20064
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9731009711871984e-05,
+      "loss": 0.3635,
+      "step": 20065
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730971862103563e-05,
+      "loss": 0.294,
+      "step": 20066
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730934009708703e-05,
+      "loss": 0.2714,
+      "step": 20067
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730896154687417e-05,
+      "loss": 0.2745,
+      "step": 20068
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730858297039723e-05,
+      "loss": 0.3055,
+      "step": 20069
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.973082043676562e-05,
+      "loss": 0.3312,
+      "step": 20070
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730782573865125e-05,
+      "loss": 0.2631,
+      "step": 20071
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730744708338242e-05,
+      "loss": 0.3313,
+      "step": 20072
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.973070684018499e-05,
+      "loss": 0.2923,
+      "step": 20073
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730668969405372e-05,
+      "loss": 0.3307,
+      "step": 20074
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730631095999403e-05,
+      "loss": 0.2748,
+      "step": 20075
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730593219967092e-05,
+      "loss": 0.3616,
+      "step": 20076
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730555341308447e-05,
+      "loss": 0.297,
+      "step": 20077
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.973051746002348e-05,
+      "loss": 0.3112,
+      "step": 20078
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.97304795761122e-05,
+      "loss": 0.3489,
+      "step": 20079
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730441689574617e-05,
+      "loss": 0.316,
+      "step": 20080
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730403800410744e-05,
+      "loss": 0.2982,
+      "step": 20081
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.973036590862059e-05,
+      "loss": 0.3356,
+      "step": 20082
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730328014204166e-05,
+      "loss": 0.3213,
+      "step": 20083
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730290117161478e-05,
+      "loss": 0.2621,
+      "step": 20084
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.973025221749254e-05,
+      "loss": 0.3066,
+      "step": 20085
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730214315197363e-05,
+      "loss": 0.2742,
+      "step": 20086
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730176410275957e-05,
+      "loss": 0.2638,
+      "step": 20087
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730138502728325e-05,
+      "loss": 0.2715,
+      "step": 20088
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730100592554488e-05,
+      "loss": 0.303,
+      "step": 20089
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.973006267975445e-05,
+      "loss": 0.235,
+      "step": 20090
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9730024764328225e-05,
+      "loss": 0.3557,
+      "step": 20091
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729986846275816e-05,
+      "loss": 0.3302,
+      "step": 20092
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972994892559724e-05,
+      "loss": 0.2794,
+      "step": 20093
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729911002292507e-05,
+      "loss": 0.296,
+      "step": 20094
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729873076361625e-05,
+      "loss": 0.3051,
+      "step": 20095
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729835147804604e-05,
+      "loss": 0.3128,
+      "step": 20096
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729797216621453e-05,
+      "loss": 0.2661,
+      "step": 20097
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729759282812185e-05,
+      "loss": 0.2709,
+      "step": 20098
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729721346376813e-05,
+      "loss": 0.3537,
+      "step": 20099
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972968340731534e-05,
+      "loss": 0.2976,
+      "step": 20100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729645465627785e-05,
+      "loss": 0.2839,
+      "step": 20101
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729607521314146e-05,
+      "loss": 0.3241,
+      "step": 20102
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729569574374444e-05,
+      "loss": 0.2879,
+      "step": 20103
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972953162480869e-05,
+      "loss": 0.284,
+      "step": 20104
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729493672616885e-05,
+      "loss": 0.2874,
+      "step": 20105
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729455717799045e-05,
+      "loss": 0.3056,
+      "step": 20106
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972941776035518e-05,
+      "loss": 0.2902,
+      "step": 20107
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729379800285298e-05,
+      "loss": 0.2833,
+      "step": 20108
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972934183758941e-05,
+      "loss": 0.3068,
+      "step": 20109
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972930387226753e-05,
+      "loss": 0.3313,
+      "step": 20110
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729265904319664e-05,
+      "loss": 0.2882,
+      "step": 20111
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729227933745826e-05,
+      "loss": 0.2764,
+      "step": 20112
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972918996054602e-05,
+      "loss": 0.2829,
+      "step": 20113
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729151984720264e-05,
+      "loss": 0.2584,
+      "step": 20114
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972911400626856e-05,
+      "loss": 0.3473,
+      "step": 20115
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729076025190927e-05,
+      "loss": 0.2642,
+      "step": 20116
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972903804148737e-05,
+      "loss": 0.3235,
+      "step": 20117
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9729000055157898e-05,
+      "loss": 0.3013,
+      "step": 20118
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728962066202526e-05,
+      "loss": 0.2614,
+      "step": 20119
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972892407462126e-05,
+      "loss": 0.2803,
+      "step": 20120
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728886080414115e-05,
+      "loss": 0.3381,
+      "step": 20121
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728848083581092e-05,
+      "loss": 0.2899,
+      "step": 20122
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972881008412221e-05,
+      "loss": 0.2254,
+      "step": 20123
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972877208203748e-05,
+      "loss": 0.2502,
+      "step": 20124
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728734077326908e-05,
+      "loss": 0.3069,
+      "step": 20125
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728696069990504e-05,
+      "loss": 0.3226,
+      "step": 20126
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728658060028278e-05,
+      "loss": 0.2726,
+      "step": 20127
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728620047440243e-05,
+      "loss": 0.3265,
+      "step": 20128
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728582032226405e-05,
+      "loss": 0.2999,
+      "step": 20129
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728544014386783e-05,
+      "loss": 0.2476,
+      "step": 20130
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972850599392138e-05,
+      "loss": 0.304,
+      "step": 20131
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728467970830203e-05,
+      "loss": 0.2678,
+      "step": 20132
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728429945113272e-05,
+      "loss": 0.3009,
+      "step": 20133
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972839191677059e-05,
+      "loss": 0.2847,
+      "step": 20134
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972835388580217e-05,
+      "loss": 0.323,
+      "step": 20135
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728315852208024e-05,
+      "loss": 0.3487,
+      "step": 20136
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728277815988156e-05,
+      "loss": 0.2631,
+      "step": 20137
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728239777142584e-05,
+      "loss": 0.2697,
+      "step": 20138
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728201735671315e-05,
+      "loss": 0.3103,
+      "step": 20139
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728163691574352e-05,
+      "loss": 0.3059,
+      "step": 20140
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972812564485172e-05,
+      "loss": 0.2665,
+      "step": 20141
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9728087595503416e-05,
+      "loss": 0.2714,
+      "step": 20142
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972804954352946e-05,
+      "loss": 0.2767,
+      "step": 20143
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972801148892986e-05,
+      "loss": 0.2793,
+      "step": 20144
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9727973431704617e-05,
+      "loss": 0.3184,
+      "step": 20145
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9727935371853753e-05,
+      "loss": 0.2797,
+      "step": 20146
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972789730937727e-05,
+      "loss": 0.2688,
+      "step": 20147
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9727859244275187e-05,
+      "loss": 0.2724,
+      "step": 20148
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9727821176547506e-05,
+      "loss": 0.2978,
+      "step": 20149
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972778310619424e-05,
+      "loss": 0.3528,
+      "step": 20150
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9727745033215405e-05,
+      "loss": 0.3127,
+      "step": 20151
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9727706957611002e-05,
+      "loss": 0.3163,
+      "step": 20152
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9727668879381044e-05,
+      "loss": 0.2802,
+      "step": 20153
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9727630798525545e-05,
+      "loss": 0.2859,
+      "step": 20154
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9727592715044515e-05,
+      "loss": 0.3138,
+      "step": 20155
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9727554628937957e-05,
+      "loss": 0.2756,
+      "step": 20156
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972751654020589e-05,
+      "loss": 0.3085,
+      "step": 20157
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9727478448848322e-05,
+      "loss": 0.3259,
+      "step": 20158
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972744035486526e-05,
+      "loss": 0.2975,
+      "step": 20159
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9727402258256716e-05,
+      "loss": 0.2779,
+      "step": 20160
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9727364159022703e-05,
+      "loss": 0.2651,
+      "step": 20161
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972732605716323e-05,
+      "loss": 0.2762,
+      "step": 20162
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.97272879526783e-05,
+      "loss": 0.3362,
+      "step": 20163
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9727249845567933e-05,
+      "loss": 0.2682,
+      "step": 20164
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9727211735832138e-05,
+      "loss": 0.2627,
+      "step": 20165
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972717362347092e-05,
+      "loss": 0.254,
+      "step": 20166
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9727135508484292e-05,
+      "loss": 0.3219,
+      "step": 20167
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972709739087227e-05,
+      "loss": 0.2674,
+      "step": 20168
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972705927063485e-05,
+      "loss": 0.257,
+      "step": 20169
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9727021147772057e-05,
+      "loss": 0.3191,
+      "step": 20170
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726983022283895e-05,
+      "loss": 0.2577,
+      "step": 20171
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726944894170374e-05,
+      "loss": 0.2519,
+      "step": 20172
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726906763431506e-05,
+      "loss": 0.3484,
+      "step": 20173
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726868630067303e-05,
+      "loss": 0.3136,
+      "step": 20174
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726830494077768e-05,
+      "loss": 0.2816,
+      "step": 20175
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972679235546292e-05,
+      "loss": 0.2892,
+      "step": 20176
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972675421422276e-05,
+      "loss": 0.2556,
+      "step": 20177
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726716070357307e-05,
+      "loss": 0.2713,
+      "step": 20178
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972667792386657e-05,
+      "loss": 0.2891,
+      "step": 20179
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726639774750552e-05,
+      "loss": 0.2774,
+      "step": 20180
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726601623009268e-05,
+      "loss": 0.2974,
+      "step": 20181
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726563468642734e-05,
+      "loss": 0.3167,
+      "step": 20182
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726525311650953e-05,
+      "loss": 0.2692,
+      "step": 20183
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726487152033936e-05,
+      "loss": 0.3293,
+      "step": 20184
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726448989791697e-05,
+      "loss": 0.3198,
+      "step": 20185
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972641082492424e-05,
+      "loss": 0.3198,
+      "step": 20186
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726372657431586e-05,
+      "loss": 0.3194,
+      "step": 20187
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726334487313736e-05,
+      "loss": 0.2869,
+      "step": 20188
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.97262963145707e-05,
+      "loss": 0.2782,
+      "step": 20189
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726258139202492e-05,
+      "loss": 0.2637,
+      "step": 20190
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726219961209122e-05,
+      "loss": 0.2628,
+      "step": 20191
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726181780590598e-05,
+      "loss": 0.3257,
+      "step": 20192
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726143597346935e-05,
+      "loss": 0.236,
+      "step": 20193
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972610541147814e-05,
+      "loss": 0.3022,
+      "step": 20194
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726067222984226e-05,
+      "loss": 0.3246,
+      "step": 20195
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9726029031865196e-05,
+      "loss": 0.294,
+      "step": 20196
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725990838121066e-05,
+      "loss": 0.3466,
+      "step": 20197
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725952641751845e-05,
+      "loss": 0.3051,
+      "step": 20198
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725914442757547e-05,
+      "loss": 0.2618,
+      "step": 20199
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972587624113818e-05,
+      "loss": 0.2585,
+      "step": 20200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972583803689375e-05,
+      "loss": 0.2612,
+      "step": 20201
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725799830024272e-05,
+      "loss": 0.2792,
+      "step": 20202
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725761620529757e-05,
+      "loss": 0.2943,
+      "step": 20203
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972572340841021e-05,
+      "loss": 0.2863,
+      "step": 20204
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972568519366565e-05,
+      "loss": 0.2824,
+      "step": 20205
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972564697629608e-05,
+      "loss": 0.2934,
+      "step": 20206
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725608756301507e-05,
+      "loss": 0.2736,
+      "step": 20207
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725570533681952e-05,
+      "loss": 0.3125,
+      "step": 20208
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972553230843742e-05,
+      "loss": 0.3003,
+      "step": 20209
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972549408056792e-05,
+      "loss": 0.276,
+      "step": 20210
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725455850073467e-05,
+      "loss": 0.2763,
+      "step": 20211
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725417616954064e-05,
+      "loss": 0.2907,
+      "step": 20212
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725379381209728e-05,
+      "loss": 0.3251,
+      "step": 20213
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725341142840463e-05,
+      "loss": 0.3512,
+      "step": 20214
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725302901846283e-05,
+      "loss": 0.3071,
+      "step": 20215
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725264658227203e-05,
+      "loss": 0.3105,
+      "step": 20216
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725226411983226e-05,
+      "loss": 0.2862,
+      "step": 20217
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725188163114364e-05,
+      "loss": 0.273,
+      "step": 20218
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725149911620632e-05,
+      "loss": 0.2784,
+      "step": 20219
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725111657502033e-05,
+      "loss": 0.2833,
+      "step": 20220
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972507340075858e-05,
+      "loss": 0.2973,
+      "step": 20221
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9725035141390284e-05,
+      "loss": 0.291,
+      "step": 20222
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972499687939716e-05,
+      "loss": 0.2814,
+      "step": 20223
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972495861477921e-05,
+      "loss": 0.2723,
+      "step": 20224
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972492034753645e-05,
+      "loss": 0.2879,
+      "step": 20225
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724882077668888e-05,
+      "loss": 0.344,
+      "step": 20226
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724843805176535e-05,
+      "loss": 0.2762,
+      "step": 20227
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724805530059403e-05,
+      "loss": 0.2667,
+      "step": 20228
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.97247672523175e-05,
+      "loss": 0.2659,
+      "step": 20229
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724728971950836e-05,
+      "loss": 0.3222,
+      "step": 20230
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972469068895942e-05,
+      "loss": 0.2538,
+      "step": 20231
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724652403343266e-05,
+      "loss": 0.291,
+      "step": 20232
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972461411510238e-05,
+      "loss": 0.2462,
+      "step": 20233
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724575824236783e-05,
+      "loss": 0.2753,
+      "step": 20234
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972453753074647e-05,
+      "loss": 0.2556,
+      "step": 20235
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972449923463146e-05,
+      "loss": 0.2873,
+      "step": 20236
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724460935891767e-05,
+      "loss": 0.3545,
+      "step": 20237
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724422634527394e-05,
+      "loss": 0.2725,
+      "step": 20238
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972438433053835e-05,
+      "loss": 0.2811,
+      "step": 20239
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724346023924655e-05,
+      "loss": 0.3289,
+      "step": 20240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724307714686312e-05,
+      "loss": 0.2781,
+      "step": 20241
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724269402823332e-05,
+      "loss": 0.3208,
+      "step": 20242
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724231088335726e-05,
+      "loss": 0.2748,
+      "step": 20243
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724192771223503e-05,
+      "loss": 0.2615,
+      "step": 20244
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724154451486678e-05,
+      "loss": 0.2645,
+      "step": 20245
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724116129125257e-05,
+      "loss": 0.2762,
+      "step": 20246
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724077804139253e-05,
+      "loss": 0.3307,
+      "step": 20247
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9724039476528674e-05,
+      "loss": 0.2839,
+      "step": 20248
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972400114629353e-05,
+      "loss": 0.3431,
+      "step": 20249
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9723962813433833e-05,
+      "loss": 0.3116,
+      "step": 20250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972392447794959e-05,
+      "loss": 0.2975,
+      "step": 20251
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972388613984082e-05,
+      "loss": 0.2895,
+      "step": 20252
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9723847799107526e-05,
+      "loss": 0.2898,
+      "step": 20253
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972380945574972e-05,
+      "loss": 0.3126,
+      "step": 20254
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9723771109767413e-05,
+      "loss": 0.2973,
+      "step": 20255
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.972373276116061e-05,
+      "loss": 0.2836,
+      "step": 20256
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972369440992933e-05,
+      "loss": 0.3117,
+      "step": 20257
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972365605607358e-05,
+      "loss": 0.2882,
+      "step": 20258
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972361769959337e-05,
+      "loss": 0.3227,
+      "step": 20259
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972357934048871e-05,
+      "loss": 0.3035,
+      "step": 20260
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972354097875961e-05,
+      "loss": 0.2918,
+      "step": 20261
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972350261440608e-05,
+      "loss": 0.2938,
+      "step": 20262
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972346424742813e-05,
+      "loss": 0.282,
+      "step": 20263
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9723425877825773e-05,
+      "loss": 0.2805,
+      "step": 20264
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9723387505599017e-05,
+      "loss": 0.2723,
+      "step": 20265
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9723349130747876e-05,
+      "loss": 0.2896,
+      "step": 20266
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9723310753272355e-05,
+      "loss": 0.2957,
+      "step": 20267
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972327237317247e-05,
+      "loss": 0.2902,
+      "step": 20268
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9723233990448224e-05,
+      "loss": 0.3452,
+      "step": 20269
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9723195605099633e-05,
+      "loss": 0.3098,
+      "step": 20270
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9723157217126708e-05,
+      "loss": 0.2777,
+      "step": 20271
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9723118826529454e-05,
+      "loss": 0.3002,
+      "step": 20272
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972308043330789e-05,
+      "loss": 0.2901,
+      "step": 20273
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9723042037462016e-05,
+      "loss": 0.2994,
+      "step": 20274
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9723003638991852e-05,
+      "loss": 0.2788,
+      "step": 20275
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.97229652378974e-05,
+      "loss": 0.313,
+      "step": 20276
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972292683417868e-05,
+      "loss": 0.3171,
+      "step": 20277
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972288842783569e-05,
+      "loss": 0.2907,
+      "step": 20278
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972285001886845e-05,
+      "loss": 0.3226,
+      "step": 20279
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722811607276965e-05,
+      "loss": 0.2861,
+      "step": 20280
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722773193061248e-05,
+      "loss": 0.4159,
+      "step": 20281
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722734776221314e-05,
+      "loss": 0.2621,
+      "step": 20282
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722696356757164e-05,
+      "loss": 0.2953,
+      "step": 20283
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722657934668815e-05,
+      "loss": 0.2556,
+      "step": 20284
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722619509956272e-05,
+      "loss": 0.3394,
+      "step": 20285
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972258108261955e-05,
+      "loss": 0.3254,
+      "step": 20286
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722542652658657e-05,
+      "loss": 0.2956,
+      "step": 20287
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722504220073607e-05,
+      "loss": 0.3264,
+      "step": 20288
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722465784864406e-05,
+      "loss": 0.3026,
+      "step": 20289
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722427347031065e-05,
+      "loss": 0.2971,
+      "step": 20290
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.97223889065736e-05,
+      "loss": 0.29,
+      "step": 20291
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972235046349201e-05,
+      "loss": 0.2854,
+      "step": 20292
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972231201778632e-05,
+      "loss": 0.3,
+      "step": 20293
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722273569456526e-05,
+      "loss": 0.2913,
+      "step": 20294
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722235118502645e-05,
+      "loss": 0.2631,
+      "step": 20295
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972219666492469e-05,
+      "loss": 0.2634,
+      "step": 20296
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972215820872267e-05,
+      "loss": 0.2483,
+      "step": 20297
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722119749896592e-05,
+      "loss": 0.2957,
+      "step": 20298
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972208128844647e-05,
+      "loss": 0.2708,
+      "step": 20299
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972204282437231e-05,
+      "loss": 0.2907,
+      "step": 20300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722004357674125e-05,
+      "loss": 0.3509,
+      "step": 20301
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972196588835193e-05,
+      "loss": 0.3333,
+      "step": 20302
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972192741640573e-05,
+      "loss": 0.2895,
+      "step": 20303
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721888941835533e-05,
+      "loss": 0.2608,
+      "step": 20304
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972185046464135e-05,
+      "loss": 0.258,
+      "step": 20305
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.97218119848232e-05,
+      "loss": 0.3183,
+      "step": 20306
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721773502381086e-05,
+      "loss": 0.2725,
+      "step": 20307
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972173501731502e-05,
+      "loss": 0.2343,
+      "step": 20308
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721696529625013e-05,
+      "loss": 0.2834,
+      "step": 20309
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721658039311073e-05,
+      "loss": 0.3323,
+      "step": 20310
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721619546373212e-05,
+      "loss": 0.28,
+      "step": 20311
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721581050811442e-05,
+      "loss": 0.3475,
+      "step": 20312
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972154255262577e-05,
+      "loss": 0.2999,
+      "step": 20313
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972150405181621e-05,
+      "loss": 0.3311,
+      "step": 20314
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972146554838277e-05,
+      "loss": 0.3322,
+      "step": 20315
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972142704232546e-05,
+      "loss": 0.2816,
+      "step": 20316
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721388533644293e-05,
+      "loss": 0.3058,
+      "step": 20317
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721350022339275e-05,
+      "loss": 0.2638,
+      "step": 20318
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721311508410422e-05,
+      "loss": 0.2888,
+      "step": 20319
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721272991857743e-05,
+      "loss": 0.2722,
+      "step": 20320
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721234472681244e-05,
+      "loss": 0.2997,
+      "step": 20321
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721195950880938e-05,
+      "loss": 0.2958,
+      "step": 20322
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972115742645684e-05,
+      "loss": 0.319,
+      "step": 20323
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9721118899408954e-05,
+      "loss": 0.2721,
+      "step": 20324
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972108036973729e-05,
+      "loss": 0.3208,
+      "step": 20325
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972104183744186e-05,
+      "loss": 0.3082,
+      "step": 20326
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972100330252268e-05,
+      "loss": 0.2555,
+      "step": 20327
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720964764979756e-05,
+      "loss": 0.3381,
+      "step": 20328
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720926224813092e-05,
+      "loss": 0.2919,
+      "step": 20329
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972088768202271e-05,
+      "loss": 0.2772,
+      "step": 20330
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720849136608617e-05,
+      "loss": 0.3435,
+      "step": 20331
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720810588570815e-05,
+      "loss": 0.306,
+      "step": 20332
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720772037909325e-05,
+      "loss": 0.298,
+      "step": 20333
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972073348462415e-05,
+      "loss": 0.324,
+      "step": 20334
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720694928715305e-05,
+      "loss": 0.2899,
+      "step": 20335
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.97206563701828e-05,
+      "loss": 0.3199,
+      "step": 20336
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720617809026642e-05,
+      "loss": 0.3185,
+      "step": 20337
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720579245246848e-05,
+      "loss": 0.3106,
+      "step": 20338
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720540678843424e-05,
+      "loss": 0.2851,
+      "step": 20339
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720502109816376e-05,
+      "loss": 0.2728,
+      "step": 20340
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972046353816572e-05,
+      "loss": 0.3109,
+      "step": 20341
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972042496389147e-05,
+      "loss": 0.2868,
+      "step": 20342
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972038638699363e-05,
+      "loss": 0.2882,
+      "step": 20343
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972034780747221e-05,
+      "loss": 0.2794,
+      "step": 20344
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972030922532722e-05,
+      "loss": 0.3147,
+      "step": 20345
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720270640558678e-05,
+      "loss": 0.2983,
+      "step": 20346
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972023205316659e-05,
+      "loss": 0.2727,
+      "step": 20347
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720193463150964e-05,
+      "loss": 0.3077,
+      "step": 20348
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972015487051181e-05,
+      "loss": 0.3173,
+      "step": 20349
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720116275249145e-05,
+      "loss": 0.2457,
+      "step": 20350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720077677362973e-05,
+      "loss": 0.2981,
+      "step": 20351
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.972003907685331e-05,
+      "loss": 0.3356,
+      "step": 20352
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9720000473720158e-05,
+      "loss": 0.3078,
+      "step": 20353
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9719961867963535e-05,
+      "loss": 0.3109,
+      "step": 20354
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719923259583452e-05,
+      "loss": 0.2693,
+      "step": 20355
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971988464857991e-05,
+      "loss": 0.3271,
+      "step": 20356
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971984603495293e-05,
+      "loss": 0.3557,
+      "step": 20357
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719807418702514e-05,
+      "loss": 0.2677,
+      "step": 20358
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971976879982868e-05,
+      "loss": 0.2716,
+      "step": 20359
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719730178331436e-05,
+      "loss": 0.2588,
+      "step": 20360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971969155421079e-05,
+      "loss": 0.3319,
+      "step": 20361
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971965292746675e-05,
+      "loss": 0.3154,
+      "step": 20362
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719614298099336e-05,
+      "loss": 0.2864,
+      "step": 20363
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971957566610855e-05,
+      "loss": 0.288,
+      "step": 20364
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719537031494407e-05,
+      "loss": 0.2647,
+      "step": 20365
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719498394256914e-05,
+      "loss": 0.2361,
+      "step": 20366
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719459754396082e-05,
+      "loss": 0.3311,
+      "step": 20367
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971942111191192e-05,
+      "loss": 0.2827,
+      "step": 20368
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719382466804445e-05,
+      "loss": 0.2773,
+      "step": 20369
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719343819073664e-05,
+      "loss": 0.336,
+      "step": 20370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719305168719584e-05,
+      "loss": 0.2806,
+      "step": 20371
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971926651574222e-05,
+      "loss": 0.2712,
+      "step": 20372
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971922786014158e-05,
+      "loss": 0.3063,
+      "step": 20373
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719189201917675e-05,
+      "loss": 0.2773,
+      "step": 20374
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719150541070514e-05,
+      "loss": 0.2971,
+      "step": 20375
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971911187760011e-05,
+      "loss": 0.2708,
+      "step": 20376
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9719073211506473e-05,
+      "loss": 0.3242,
+      "step": 20377
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971903454278961e-05,
+      "loss": 0.2895,
+      "step": 20378
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718995871449538e-05,
+      "loss": 0.2802,
+      "step": 20379
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971895719748626e-05,
+      "loss": 0.3096,
+      "step": 20380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718918520899792e-05,
+      "loss": 0.3031,
+      "step": 20381
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718879841690142e-05,
+      "loss": 0.3203,
+      "step": 20382
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971884115985732e-05,
+      "loss": 0.301,
+      "step": 20383
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971880247540134e-05,
+      "loss": 0.3321,
+      "step": 20384
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718763788322207e-05,
+      "loss": 0.2974,
+      "step": 20385
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718725098619934e-05,
+      "loss": 0.3318,
+      "step": 20386
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718686406294535e-05,
+      "loss": 0.2633,
+      "step": 20387
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718647711346014e-05,
+      "loss": 0.2962,
+      "step": 20388
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718609013774385e-05,
+      "loss": 0.3557,
+      "step": 20389
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971857031357966e-05,
+      "loss": 0.3305,
+      "step": 20390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718531610761844e-05,
+      "loss": 0.3246,
+      "step": 20391
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718492905320953e-05,
+      "loss": 0.3044,
+      "step": 20392
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718454197256997e-05,
+      "loss": 0.3035,
+      "step": 20393
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971841548656998e-05,
+      "loss": 0.2944,
+      "step": 20394
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718376773259923e-05,
+      "loss": 0.3469,
+      "step": 20395
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718338057326825e-05,
+      "loss": 0.3019,
+      "step": 20396
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718299338770706e-05,
+      "loss": 0.2899,
+      "step": 20397
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971826061759157e-05,
+      "loss": 0.2625,
+      "step": 20398
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718221893789435e-05,
+      "loss": 0.2582,
+      "step": 20399
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718183167364305e-05,
+      "loss": 0.2774,
+      "step": 20400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971814443831619e-05,
+      "loss": 0.2524,
+      "step": 20401
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.97181057066451e-05,
+      "loss": 0.2959,
+      "step": 20402
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9718066972351052e-05,
+      "loss": 0.2956,
+      "step": 20403
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971802823543405e-05,
+      "loss": 0.2504,
+      "step": 20404
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717989495894104e-05,
+      "loss": 0.3091,
+      "step": 20405
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971795075373123e-05,
+      "loss": 0.2706,
+      "step": 20406
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971791200894544e-05,
+      "loss": 0.2653,
+      "step": 20407
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717873261536736e-05,
+      "loss": 0.2944,
+      "step": 20408
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717834511505134e-05,
+      "loss": 0.2949,
+      "step": 20409
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971779575885064e-05,
+      "loss": 0.2592,
+      "step": 20410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971775700357327e-05,
+      "loss": 0.2606,
+      "step": 20411
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971771824567303e-05,
+      "loss": 0.308,
+      "step": 20412
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717679485149937e-05,
+      "loss": 0.3302,
+      "step": 20413
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971764072200399e-05,
+      "loss": 0.2724,
+      "step": 20414
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971760195623521e-05,
+      "loss": 0.2875,
+      "step": 20415
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717563187843604e-05,
+      "loss": 0.253,
+      "step": 20416
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717524416829184e-05,
+      "loss": 0.3064,
+      "step": 20417
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971748564319196e-05,
+      "loss": 0.3673,
+      "step": 20418
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717446866931934e-05,
+      "loss": 0.2661,
+      "step": 20419
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971740808804913e-05,
+      "loss": 0.3001,
+      "step": 20420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717369306543552e-05,
+      "loss": 0.3066,
+      "step": 20421
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717330522415204e-05,
+      "loss": 0.3611,
+      "step": 20422
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971729173566411e-05,
+      "loss": 0.2827,
+      "step": 20423
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971725294629027e-05,
+      "loss": 0.328,
+      "step": 20424
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.97172141542937e-05,
+      "loss": 0.2928,
+      "step": 20425
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717175359674408e-05,
+      "loss": 0.2844,
+      "step": 20426
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717136562432404e-05,
+      "loss": 0.2968,
+      "step": 20427
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.97170977625677e-05,
+      "loss": 0.3263,
+      "step": 20428
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9717058960080305e-05,
+      "loss": 0.3159,
+      "step": 20429
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971702015497023e-05,
+      "loss": 0.2905,
+      "step": 20430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716981347237487e-05,
+      "loss": 0.2761,
+      "step": 20431
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716942536882085e-05,
+      "loss": 0.2508,
+      "step": 20432
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716903723904036e-05,
+      "loss": 0.2912,
+      "step": 20433
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971686490830335e-05,
+      "loss": 0.2518,
+      "step": 20434
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716826090080035e-05,
+      "loss": 0.2969,
+      "step": 20435
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716787269234103e-05,
+      "loss": 0.3148,
+      "step": 20436
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716748445765564e-05,
+      "loss": 0.3212,
+      "step": 20437
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716709619674432e-05,
+      "loss": 0.3427,
+      "step": 20438
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716670790960713e-05,
+      "loss": 0.282,
+      "step": 20439
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971663195962442e-05,
+      "loss": 0.2326,
+      "step": 20440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971659312566556e-05,
+      "loss": 0.2894,
+      "step": 20441
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971655428908415e-05,
+      "loss": 0.2631,
+      "step": 20442
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716515449880192e-05,
+      "loss": 0.3142,
+      "step": 20443
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716476608053702e-05,
+      "loss": 0.3403,
+      "step": 20444
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716437763604692e-05,
+      "loss": 0.2594,
+      "step": 20445
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716398916533167e-05,
+      "loss": 0.3003,
+      "step": 20446
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716360066839143e-05,
+      "loss": 0.2846,
+      "step": 20447
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716321214522627e-05,
+      "loss": 0.2874,
+      "step": 20448
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716282359583633e-05,
+      "loss": 0.273,
+      "step": 20449
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716243502022164e-05,
+      "loss": 0.2755,
+      "step": 20450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.971620464183824e-05,
+      "loss": 0.2808,
+      "step": 20451
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716165779031865e-05,
+      "loss": 0.2951,
+      "step": 20452
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9716126913603052e-05,
+      "loss": 0.3105,
+      "step": 20453
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9716088045551812e-05,
+      "loss": 0.3382,
+      "step": 20454
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9716049174878154e-05,
+      "loss": 0.2905,
+      "step": 20455
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9716010301582086e-05,
+      "loss": 0.2543,
+      "step": 20456
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715971425663624e-05,
+      "loss": 0.2725,
+      "step": 20457
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715932547122775e-05,
+      "loss": 0.2717,
+      "step": 20458
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715893665959553e-05,
+      "loss": 0.3393,
+      "step": 20459
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715854782173964e-05,
+      "loss": 0.3155,
+      "step": 20460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971581589576602e-05,
+      "loss": 0.2911,
+      "step": 20461
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971577700673573e-05,
+      "loss": 0.272,
+      "step": 20462
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715738115083106e-05,
+      "loss": 0.2856,
+      "step": 20463
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715699220808163e-05,
+      "loss": 0.3156,
+      "step": 20464
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715660323910907e-05,
+      "loss": 0.3177,
+      "step": 20465
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715621424391346e-05,
+      "loss": 0.2836,
+      "step": 20466
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715582522249496e-05,
+      "loss": 0.3202,
+      "step": 20467
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971554361748536e-05,
+      "loss": 0.2306,
+      "step": 20468
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715504710098957e-05,
+      "loss": 0.2628,
+      "step": 20469
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715465800090295e-05,
+      "loss": 0.2987,
+      "step": 20470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715426887459382e-05,
+      "loss": 0.3029,
+      "step": 20471
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971538797220623e-05,
+      "loss": 0.2512,
+      "step": 20472
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971534905433085e-05,
+      "loss": 0.2817,
+      "step": 20473
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715310133833254e-05,
+      "loss": 0.286,
+      "step": 20474
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715271210713443e-05,
+      "loss": 0.3255,
+      "step": 20475
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715232284971443e-05,
+      "loss": 0.2634,
+      "step": 20476
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715193356607252e-05,
+      "loss": 0.2669,
+      "step": 20477
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715154425620884e-05,
+      "loss": 0.2465,
+      "step": 20478
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715115492012353e-05,
+      "loss": 0.2617,
+      "step": 20479
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715076555781666e-05,
+      "loss": 0.2767,
+      "step": 20480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9715037616928835e-05,
+      "loss": 0.247,
+      "step": 20481
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971499867545387e-05,
+      "loss": 0.2785,
+      "step": 20482
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971495973135678e-05,
+      "loss": 0.2879,
+      "step": 20483
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714920784637578e-05,
+      "loss": 0.2558,
+      "step": 20484
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714881835296274e-05,
+      "loss": 0.3095,
+      "step": 20485
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714842883332878e-05,
+      "loss": 0.2751,
+      "step": 20486
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.97148039287474e-05,
+      "loss": 0.3269,
+      "step": 20487
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971476497153985e-05,
+      "loss": 0.2972,
+      "step": 20488
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971472601171024e-05,
+      "loss": 0.3099,
+      "step": 20489
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971468704925858e-05,
+      "loss": 0.35,
+      "step": 20490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971464808418488e-05,
+      "loss": 0.321,
+      "step": 20491
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714609116489155e-05,
+      "loss": 0.2626,
+      "step": 20492
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714570146171406e-05,
+      "loss": 0.3012,
+      "step": 20493
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714531173231652e-05,
+      "loss": 0.3048,
+      "step": 20494
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.97144921976699e-05,
+      "loss": 0.3031,
+      "step": 20495
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714453219486163e-05,
+      "loss": 0.2634,
+      "step": 20496
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714414238680448e-05,
+      "loss": 0.3386,
+      "step": 20497
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714375255252766e-05,
+      "loss": 0.3971,
+      "step": 20498
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971433626920313e-05,
+      "loss": 0.3365,
+      "step": 20499
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971429728053155e-05,
+      "loss": 0.2706,
+      "step": 20500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714258289238035e-05,
+      "loss": 0.2688,
+      "step": 20501
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714219295322598e-05,
+      "loss": 0.3083,
+      "step": 20502
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714180298785245e-05,
+      "loss": 0.2736,
+      "step": 20503
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971414129962599e-05,
+      "loss": 0.2719,
+      "step": 20504
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714102297844845e-05,
+      "loss": 0.3279,
+      "step": 20505
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714063293441815e-05,
+      "loss": 0.3216,
+      "step": 20506
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9714024286416914e-05,
+      "loss": 0.3473,
+      "step": 20507
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713985276770157e-05,
+      "loss": 0.2656,
+      "step": 20508
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713946264501546e-05,
+      "loss": 0.2602,
+      "step": 20509
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713907249611096e-05,
+      "loss": 0.2843,
+      "step": 20510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713868232098815e-05,
+      "loss": 0.2902,
+      "step": 20511
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971382921196472e-05,
+      "loss": 0.2749,
+      "step": 20512
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713790189208817e-05,
+      "loss": 0.2743,
+      "step": 20513
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713751163831114e-05,
+      "loss": 0.3731,
+      "step": 20514
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713712135831623e-05,
+      "loss": 0.3016,
+      "step": 20515
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971367310521036e-05,
+      "loss": 0.3331,
+      "step": 20516
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713634071967328e-05,
+      "loss": 0.3323,
+      "step": 20517
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713595036102543e-05,
+      "loss": 0.2995,
+      "step": 20518
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971355599761601e-05,
+      "loss": 0.2813,
+      "step": 20519
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713516956507745e-05,
+      "loss": 0.3154,
+      "step": 20520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713477912777758e-05,
+      "loss": 0.2677,
+      "step": 20521
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713438866426052e-05,
+      "loss": 0.2728,
+      "step": 20522
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971339981745265e-05,
+      "loss": 0.2946,
+      "step": 20523
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713360765857552e-05,
+      "loss": 0.2952,
+      "step": 20524
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713321711640776e-05,
+      "loss": 0.2523,
+      "step": 20525
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713282654802327e-05,
+      "loss": 0.3162,
+      "step": 20526
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713243595342217e-05,
+      "loss": 0.242,
+      "step": 20527
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713204533260458e-05,
+      "loss": 0.2916,
+      "step": 20528
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971316546855706e-05,
+      "loss": 0.3194,
+      "step": 20529
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713126401232033e-05,
+      "loss": 0.3135,
+      "step": 20530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713087331285384e-05,
+      "loss": 0.2953,
+      "step": 20531
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713048258717134e-05,
+      "loss": 0.2947,
+      "step": 20532
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9713009183527283e-05,
+      "loss": 0.2843,
+      "step": 20533
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9712970105715845e-05,
+      "loss": 0.3069,
+      "step": 20534
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9712931025282833e-05,
+      "loss": 0.2923,
+      "step": 20535
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9712891942228254e-05,
+      "loss": 0.3083,
+      "step": 20536
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971285285655212e-05,
+      "loss": 0.2703,
+      "step": 20537
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9712813768254442e-05,
+      "loss": 0.3243,
+      "step": 20538
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971277467733523e-05,
+      "loss": 0.2931,
+      "step": 20539
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9712735583794495e-05,
+      "loss": 0.2893,
+      "step": 20540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971269648763225e-05,
+      "loss": 0.3031,
+      "step": 20541
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.97126573888485e-05,
+      "loss": 0.2657,
+      "step": 20542
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971261828744326e-05,
+      "loss": 0.283,
+      "step": 20543
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9712579183416536e-05,
+      "loss": 0.3068,
+      "step": 20544
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9712540076768342e-05,
+      "loss": 0.2608,
+      "step": 20545
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971250096749869e-05,
+      "loss": 0.2827,
+      "step": 20546
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971246185560759e-05,
+      "loss": 0.3581,
+      "step": 20547
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9712422741095046e-05,
+      "loss": 0.2999,
+      "step": 20548
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9712383623961076e-05,
+      "loss": 0.2803,
+      "step": 20549
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.971234450420569e-05,
+      "loss": 0.2382,
+      "step": 20550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9712305381828895e-05,
+      "loss": 0.2742,
+      "step": 20551
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9712266256830704e-05,
+      "loss": 0.3259,
+      "step": 20552
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9712227129211127e-05,
+      "loss": 0.2677,
+      "step": 20553
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9712187998970175e-05,
+      "loss": 0.3056,
+      "step": 20554
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9712148866107857e-05,
+      "loss": 0.2501,
+      "step": 20555
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9712109730624185e-05,
+      "loss": 0.283,
+      "step": 20556
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971207059251917e-05,
+      "loss": 0.257,
+      "step": 20557
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971203145179282e-05,
+      "loss": 0.2871,
+      "step": 20558
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711992308445145e-05,
+      "loss": 0.3062,
+      "step": 20559
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711953162476164e-05,
+      "loss": 0.2856,
+      "step": 20560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711914013885877e-05,
+      "loss": 0.278,
+      "step": 20561
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.97118748626743e-05,
+      "loss": 0.342,
+      "step": 20562
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711835708841445e-05,
+      "loss": 0.2904,
+      "step": 20563
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711796552387318e-05,
+      "loss": 0.3385,
+      "step": 20564
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971175739331193e-05,
+      "loss": 0.3228,
+      "step": 20565
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711718231615297e-05,
+      "loss": 0.3421,
+      "step": 20566
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711679067297424e-05,
+      "loss": 0.2859,
+      "step": 20567
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711639900358323e-05,
+      "loss": 0.2825,
+      "step": 20568
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711600730798007e-05,
+      "loss": 0.2719,
+      "step": 20569
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711561558616483e-05,
+      "loss": 0.323,
+      "step": 20570
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711522383813762e-05,
+      "loss": 0.2791,
+      "step": 20571
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711483206389857e-05,
+      "loss": 0.2881,
+      "step": 20572
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711444026344777e-05,
+      "loss": 0.3232,
+      "step": 20573
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711404843678534e-05,
+      "loss": 0.2709,
+      "step": 20574
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711365658391134e-05,
+      "loss": 0.2879,
+      "step": 20575
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711326470482594e-05,
+      "loss": 0.3036,
+      "step": 20576
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711287279952924e-05,
+      "loss": 0.3158,
+      "step": 20577
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711248086802127e-05,
+      "loss": 0.3397,
+      "step": 20578
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971120889103022e-05,
+      "loss": 0.2912,
+      "step": 20579
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971116969263722e-05,
+      "loss": 0.3754,
+      "step": 20580
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971113049162312e-05,
+      "loss": 0.2504,
+      "step": 20581
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711091287987943e-05,
+      "loss": 0.2975,
+      "step": 20582
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711052081731697e-05,
+      "loss": 0.3457,
+      "step": 20583
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9711012872854395e-05,
+      "loss": 0.3706,
+      "step": 20584
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710973661356042e-05,
+      "loss": 0.278,
+      "step": 20585
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710934447236653e-05,
+      "loss": 0.3402,
+      "step": 20586
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971089523049624e-05,
+      "loss": 0.264,
+      "step": 20587
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710856011134808e-05,
+      "loss": 0.2973,
+      "step": 20588
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710816789152372e-05,
+      "loss": 0.2855,
+      "step": 20589
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971077756454894e-05,
+      "loss": 0.3256,
+      "step": 20590
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710738337324527e-05,
+      "loss": 0.2848,
+      "step": 20591
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710699107479138e-05,
+      "loss": 0.2432,
+      "step": 20592
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710659875012782e-05,
+      "loss": 0.3375,
+      "step": 20593
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971062063992548e-05,
+      "loss": 0.2528,
+      "step": 20594
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971058140221723e-05,
+      "loss": 0.3431,
+      "step": 20595
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710542161888053e-05,
+      "loss": 0.2572,
+      "step": 20596
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710502918937956e-05,
+      "loss": 0.292,
+      "step": 20597
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710463673366947e-05,
+      "loss": 0.3492,
+      "step": 20598
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710424425175038e-05,
+      "loss": 0.2973,
+      "step": 20599
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971038517436224e-05,
+      "loss": 0.2858,
+      "step": 20600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710345920928565e-05,
+      "loss": 0.3052,
+      "step": 20601
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710306664874022e-05,
+      "loss": 0.2771,
+      "step": 20602
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971026740619862e-05,
+      "loss": 0.2443,
+      "step": 20603
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710228144902373e-05,
+      "loss": 0.2595,
+      "step": 20604
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971018888098529e-05,
+      "loss": 0.3198,
+      "step": 20605
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971014961444738e-05,
+      "loss": 0.2986,
+      "step": 20606
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710110345288658e-05,
+      "loss": 0.2947,
+      "step": 20607
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9710071073509128e-05,
+      "loss": 0.3563,
+      "step": 20608
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.971003179910881e-05,
+      "loss": 0.281,
+      "step": 20609
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709992522087704e-05,
+      "loss": 0.3239,
+      "step": 20610
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709953242445828e-05,
+      "loss": 0.3536,
+      "step": 20611
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970991396018319e-05,
+      "loss": 0.3173,
+      "step": 20612
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709874675299804e-05,
+      "loss": 0.3388,
+      "step": 20613
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709835387795673e-05,
+      "loss": 0.2887,
+      "step": 20614
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970979609767081e-05,
+      "loss": 0.3036,
+      "step": 20615
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709756804925234e-05,
+      "loss": 0.2778,
+      "step": 20616
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709717509558946e-05,
+      "loss": 0.3059,
+      "step": 20617
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970967821157196e-05,
+      "loss": 0.3242,
+      "step": 20618
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709638910964286e-05,
+      "loss": 0.3162,
+      "step": 20619
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709599607735934e-05,
+      "loss": 0.3123,
+      "step": 20620
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970956030188692e-05,
+      "loss": 0.2812,
+      "step": 20621
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709520993417245e-05,
+      "loss": 0.2575,
+      "step": 20622
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709481682326926e-05,
+      "loss": 0.2847,
+      "step": 20623
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709442368615975e-05,
+      "loss": 0.2889,
+      "step": 20624
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709403052284396e-05,
+      "loss": 0.3081,
+      "step": 20625
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709363733332208e-05,
+      "loss": 0.2634,
+      "step": 20626
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709324411759416e-05,
+      "loss": 0.3064,
+      "step": 20627
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709285087566036e-05,
+      "loss": 0.2768,
+      "step": 20628
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709245760752068e-05,
+      "loss": 0.2695,
+      "step": 20629
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709206431317533e-05,
+      "loss": 0.2817,
+      "step": 20630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709167099262437e-05,
+      "loss": 0.295,
+      "step": 20631
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970912776458679e-05,
+      "loss": 0.2834,
+      "step": 20632
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9709088427290603e-05,
+      "loss": 0.3547,
+      "step": 20633
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970904908737389e-05,
+      "loss": 0.2881,
+      "step": 20634
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970900974483666e-05,
+      "loss": 0.2753,
+      "step": 20635
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9708970399678922e-05,
+      "loss": 0.2985,
+      "step": 20636
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9708931051900685e-05,
+      "loss": 0.2827,
+      "step": 20637
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9708891701501966e-05,
+      "loss": 0.2742,
+      "step": 20638
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970885234848277e-05,
+      "loss": 0.2892,
+      "step": 20639
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970881299284311e-05,
+      "loss": 0.3066,
+      "step": 20640
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9708773634582992e-05,
+      "loss": 0.2907,
+      "step": 20641
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9708734273702437e-05,
+      "loss": 0.3484,
+      "step": 20642
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9708694910201446e-05,
+      "loss": 0.2893,
+      "step": 20643
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9708655544080033e-05,
+      "loss": 0.2749,
+      "step": 20644
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970861617533821e-05,
+      "loss": 0.283,
+      "step": 20645
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9708576803975986e-05,
+      "loss": 0.3187,
+      "step": 20646
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970853742999337e-05,
+      "loss": 0.2863,
+      "step": 20647
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970849805339037e-05,
+      "loss": 0.2728,
+      "step": 20648
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.970845867416701e-05,
+      "loss": 0.2592,
+      "step": 20649
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9708419292323286e-05,
+      "loss": 0.2926,
+      "step": 20650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9708379907859215e-05,
+      "loss": 0.3346,
+      "step": 20651
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9708340520774808e-05,
+      "loss": 0.3004,
+      "step": 20652
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9708301131070073e-05,
+      "loss": 0.315,
+      "step": 20653
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9708261738745025e-05,
+      "loss": 0.3024,
+      "step": 20654
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970822234379967e-05,
+      "loss": 0.2585,
+      "step": 20655
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970818294623402e-05,
+      "loss": 0.2904,
+      "step": 20656
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9708143546048087e-05,
+      "loss": 0.3126,
+      "step": 20657
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970810414324188e-05,
+      "loss": 0.303,
+      "step": 20658
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970806473781541e-05,
+      "loss": 0.3251,
+      "step": 20659
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970802532976869e-05,
+      "loss": 0.3045,
+      "step": 20660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707985919101726e-05,
+      "loss": 0.3588,
+      "step": 20661
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970794650581453e-05,
+      "loss": 0.293,
+      "step": 20662
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707907089907117e-05,
+      "loss": 0.2615,
+      "step": 20663
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707867671379494e-05,
+      "loss": 0.2806,
+      "step": 20664
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707828250231674e-05,
+      "loss": 0.2514,
+      "step": 20665
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707788826463662e-05,
+      "loss": 0.2491,
+      "step": 20666
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707749400075476e-05,
+      "loss": 0.3546,
+      "step": 20667
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707709971067118e-05,
+      "loss": 0.2727,
+      "step": 20668
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707670539438607e-05,
+      "loss": 0.2779,
+      "step": 20669
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970763110518995e-05,
+      "loss": 0.3791,
+      "step": 20670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970759166832116e-05,
+      "loss": 0.2815,
+      "step": 20671
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707552228832242e-05,
+      "loss": 0.2985,
+      "step": 20672
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707512786723212e-05,
+      "loss": 0.3194,
+      "step": 20673
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970747334199408e-05,
+      "loss": 0.315,
+      "step": 20674
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707433894644853e-05,
+      "loss": 0.2773,
+      "step": 20675
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707394444675544e-05,
+      "loss": 0.2619,
+      "step": 20676
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970735499208617e-05,
+      "loss": 0.3557,
+      "step": 20677
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970731553687673e-05,
+      "loss": 0.2637,
+      "step": 20678
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970727607904724e-05,
+      "loss": 0.289,
+      "step": 20679
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970723661859771e-05,
+      "loss": 0.3236,
+      "step": 20680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707197155528156e-05,
+      "loss": 0.2608,
+      "step": 20681
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970715768983858e-05,
+      "loss": 0.2719,
+      "step": 20682
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707118221528998e-05,
+      "loss": 0.3538,
+      "step": 20683
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970707875059942e-05,
+      "loss": 0.3369,
+      "step": 20684
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9707039277049857e-05,
+      "loss": 0.2908,
+      "step": 20685
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706999800880314e-05,
+      "loss": 0.2646,
+      "step": 20686
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970696032209081e-05,
+      "loss": 0.3375,
+      "step": 20687
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706920840681353e-05,
+      "loss": 0.2675,
+      "step": 20688
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970688135665195e-05,
+      "loss": 0.2474,
+      "step": 20689
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970684187000262e-05,
+      "loss": 0.294,
+      "step": 20690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706802380733362e-05,
+      "loss": 0.2885,
+      "step": 20691
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706762888844196e-05,
+      "loss": 0.3563,
+      "step": 20692
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706723394335127e-05,
+      "loss": 0.3474,
+      "step": 20693
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970668389720617e-05,
+      "loss": 0.3182,
+      "step": 20694
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970664439745733e-05,
+      "loss": 0.3179,
+      "step": 20695
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706604895088626e-05,
+      "loss": 0.3184,
+      "step": 20696
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970656539010006e-05,
+      "loss": 0.27,
+      "step": 20697
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970652588249165e-05,
+      "loss": 0.2781,
+      "step": 20698
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.97064863722634e-05,
+      "loss": 0.2706,
+      "step": 20699
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706446859415326e-05,
+      "loss": 0.2741,
+      "step": 20700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970640734394744e-05,
+      "loss": 0.2771,
+      "step": 20701
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706367825859744e-05,
+      "loss": 0.2667,
+      "step": 20702
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706328305152257e-05,
+      "loss": 0.2586,
+      "step": 20703
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706288781824985e-05,
+      "loss": 0.3102,
+      "step": 20704
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706249255877942e-05,
+      "loss": 0.3025,
+      "step": 20705
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970620972731113e-05,
+      "loss": 0.2658,
+      "step": 20706
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706170196124576e-05,
+      "loss": 0.2836,
+      "step": 20707
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706130662318273e-05,
+      "loss": 0.3483,
+      "step": 20708
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706091125892247e-05,
+      "loss": 0.2938,
+      "step": 20709
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.97060515868465e-05,
+      "loss": 0.2982,
+      "step": 20710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9706012045181043e-05,
+      "loss": 0.3002,
+      "step": 20711
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970597250089589e-05,
+      "loss": 0.2654,
+      "step": 20712
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705932953991045e-05,
+      "loss": 0.3625,
+      "step": 20713
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705893404466528e-05,
+      "loss": 0.281,
+      "step": 20714
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970585385232234e-05,
+      "loss": 0.3588,
+      "step": 20715
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705814297558502e-05,
+      "loss": 0.2594,
+      "step": 20716
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705774740175017e-05,
+      "loss": 0.3121,
+      "step": 20717
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705735180171896e-05,
+      "loss": 0.2624,
+      "step": 20718
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705695617549157e-05,
+      "loss": 0.3042,
+      "step": 20719
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.97056560523068e-05,
+      "loss": 0.2648,
+      "step": 20720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705616484444845e-05,
+      "loss": 0.2783,
+      "step": 20721
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705576913963296e-05,
+      "loss": 0.2745,
+      "step": 20722
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970553734086217e-05,
+      "loss": 0.2567,
+      "step": 20723
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970549776514147e-05,
+      "loss": 0.2598,
+      "step": 20724
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705458186801212e-05,
+      "loss": 0.3119,
+      "step": 20725
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705418605841405e-05,
+      "loss": 0.2643,
+      "step": 20726
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970537902226206e-05,
+      "loss": 0.3784,
+      "step": 20727
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705339436063192e-05,
+      "loss": 0.3316,
+      "step": 20728
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705299847244804e-05,
+      "loss": 0.2846,
+      "step": 20729
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970526025580691e-05,
+      "loss": 0.2819,
+      "step": 20730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705220661749522e-05,
+      "loss": 0.278,
+      "step": 20731
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970518106507265e-05,
+      "loss": 0.2864,
+      "step": 20732
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705141465776304e-05,
+      "loss": 0.3214,
+      "step": 20733
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705101863860495e-05,
+      "loss": 0.2772,
+      "step": 20734
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970506225932523e-05,
+      "loss": 0.2831,
+      "step": 20735
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9705022652170526e-05,
+      "loss": 0.2736,
+      "step": 20736
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9704983042396393e-05,
+      "loss": 0.2529,
+      "step": 20737
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9704943430002838e-05,
+      "loss": 0.3251,
+      "step": 20738
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9704903814989874e-05,
+      "loss": 0.2682,
+      "step": 20739
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970486419735751e-05,
+      "loss": 0.2725,
+      "step": 20740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970482457710576e-05,
+      "loss": 0.36,
+      "step": 20741
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9704784954234627e-05,
+      "loss": 0.2948,
+      "step": 20742
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9704745328744133e-05,
+      "loss": 0.3462,
+      "step": 20743
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9704705700634278e-05,
+      "loss": 0.3275,
+      "step": 20744
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9704666069905082e-05,
+      "loss": 0.3176,
+      "step": 20745
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.970462643655655e-05,
+      "loss": 0.3438,
+      "step": 20746
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9704586800588694e-05,
+      "loss": 0.2961,
+      "step": 20747
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9704547162001525e-05,
+      "loss": 0.3144,
+      "step": 20748
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970450752079505e-05,
+      "loss": 0.2801,
+      "step": 20749
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9704467876969288e-05,
+      "loss": 0.2656,
+      "step": 20750
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9704428230524242e-05,
+      "loss": 0.2978,
+      "step": 20751
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9704388581459927e-05,
+      "loss": 0.2965,
+      "step": 20752
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9704348929776352e-05,
+      "loss": 0.3488,
+      "step": 20753
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9704309275473524e-05,
+      "loss": 0.2883,
+      "step": 20754
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970426961855146e-05,
+      "loss": 0.3054,
+      "step": 20755
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970422995901017e-05,
+      "loss": 0.2997,
+      "step": 20756
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9704190296849662e-05,
+      "loss": 0.3249,
+      "step": 20757
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9704150632069948e-05,
+      "loss": 0.3927,
+      "step": 20758
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970411096467104e-05,
+      "loss": 0.3038,
+      "step": 20759
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970407129465294e-05,
+      "loss": 0.277,
+      "step": 20760
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970403162201567e-05,
+      "loss": 0.3392,
+      "step": 20761
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970399194675924e-05,
+      "loss": 0.3027,
+      "step": 20762
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703952268883655e-05,
+      "loss": 0.3113,
+      "step": 20763
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703912588388925e-05,
+      "loss": 0.269,
+      "step": 20764
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703872905275068e-05,
+      "loss": 0.2897,
+      "step": 20765
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703833219542086e-05,
+      "loss": 0.2544,
+      "step": 20766
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703793531189998e-05,
+      "loss": 0.2322,
+      "step": 20767
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970375384021881e-05,
+      "loss": 0.2943,
+      "step": 20768
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703714146628533e-05,
+      "loss": 0.3255,
+      "step": 20769
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703674450419176e-05,
+      "loss": 0.3095,
+      "step": 20770
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703634751590754e-05,
+      "loss": 0.2981,
+      "step": 20771
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703595050143275e-05,
+      "loss": 0.261,
+      "step": 20772
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970355534607675e-05,
+      "loss": 0.2789,
+      "step": 20773
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703515639391192e-05,
+      "loss": 0.313,
+      "step": 20774
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970347593008661e-05,
+      "loss": 0.3141,
+      "step": 20775
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970343621816301e-05,
+      "loss": 0.2973,
+      "step": 20776
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970339650362041e-05,
+      "loss": 0.3207,
+      "step": 20777
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703356786458817e-05,
+      "loss": 0.2901,
+      "step": 20778
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703317066678245e-05,
+      "loss": 0.3204,
+      "step": 20779
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.97032773442787e-05,
+      "loss": 0.3089,
+      "step": 20780
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703237619260197e-05,
+      "loss": 0.3016,
+      "step": 20781
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703197891622743e-05,
+      "loss": 0.2754,
+      "step": 20782
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703158161366353e-05,
+      "loss": 0.297,
+      "step": 20783
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703118428491034e-05,
+      "loss": 0.3157,
+      "step": 20784
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703078692996798e-05,
+      "loss": 0.2972,
+      "step": 20785
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9703038954883655e-05,
+      "loss": 0.3025,
+      "step": 20786
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702999214151618e-05,
+      "loss": 0.3225,
+      "step": 20787
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702959470800694e-05,
+      "loss": 0.3108,
+      "step": 20788
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702919724830897e-05,
+      "loss": 0.2752,
+      "step": 20789
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702879976242236e-05,
+      "loss": 0.3529,
+      "step": 20790
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702840225034726e-05,
+      "loss": 0.3118,
+      "step": 20791
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970280047120837e-05,
+      "loss": 0.3627,
+      "step": 20792
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702760714763184e-05,
+      "loss": 0.2821,
+      "step": 20793
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702720955699175e-05,
+      "loss": 0.2939,
+      "step": 20794
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970268119401636e-05,
+      "loss": 0.2931,
+      "step": 20795
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702641429714744e-05,
+      "loss": 0.3234,
+      "step": 20796
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970260166279434e-05,
+      "loss": 0.2546,
+      "step": 20797
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970256189325516e-05,
+      "loss": 0.2957,
+      "step": 20798
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970252212109721e-05,
+      "loss": 0.3044,
+      "step": 20799
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702482346320507e-05,
+      "loss": 0.2578,
+      "step": 20800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970244256892506e-05,
+      "loss": 0.2601,
+      "step": 20801
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702402788910873e-05,
+      "loss": 0.2872,
+      "step": 20802
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702363006277966e-05,
+      "loss": 0.3363,
+      "step": 20803
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970232322102634e-05,
+      "loss": 0.2811,
+      "step": 20804
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970228343315602e-05,
+      "loss": 0.2971,
+      "step": 20805
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702243642667004e-05,
+      "loss": 0.3159,
+      "step": 20806
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970220384955931e-05,
+      "loss": 0.2997,
+      "step": 20807
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702164053832943e-05,
+      "loss": 0.2902,
+      "step": 20808
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702124255487917e-05,
+      "loss": 0.2467,
+      "step": 20809
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9702084454524244e-05,
+      "loss": 0.3093,
+      "step": 20810
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970204465094193e-05,
+      "loss": 0.3205,
+      "step": 20811
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970200484474099e-05,
+      "loss": 0.2727,
+      "step": 20812
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701965035921434e-05,
+      "loss": 0.3108,
+      "step": 20813
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701925224483274e-05,
+      "loss": 0.2994,
+      "step": 20814
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701885410426515e-05,
+      "loss": 0.2972,
+      "step": 20815
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701845593751178e-05,
+      "loss": 0.2998,
+      "step": 20816
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701805774457262e-05,
+      "loss": 0.271,
+      "step": 20817
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970176595254478e-05,
+      "loss": 0.3405,
+      "step": 20818
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701726128013755e-05,
+      "loss": 0.3332,
+      "step": 20819
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701686300864185e-05,
+      "loss": 0.2794,
+      "step": 20820
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701646471096083e-05,
+      "loss": 0.2764,
+      "step": 20821
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701606638709464e-05,
+      "loss": 0.254,
+      "step": 20822
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701566803704334e-05,
+      "loss": 0.2793,
+      "step": 20823
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701526966080706e-05,
+      "loss": 0.3099,
+      "step": 20824
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970148712583859e-05,
+      "loss": 0.2749,
+      "step": 20825
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701447282978e-05,
+      "loss": 0.3246,
+      "step": 20826
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970140743749894e-05,
+      "loss": 0.2795,
+      "step": 20827
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970136758940143e-05,
+      "loss": 0.2411,
+      "step": 20828
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970132773868547e-05,
+      "loss": 0.2992,
+      "step": 20829
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970128788535108e-05,
+      "loss": 0.3107,
+      "step": 20830
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701248029398266e-05,
+      "loss": 0.3261,
+      "step": 20831
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970120817082704e-05,
+      "loss": 0.3199,
+      "step": 20832
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970116830963741e-05,
+      "loss": 0.281,
+      "step": 20833
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970112844582939e-05,
+      "loss": 0.2488,
+      "step": 20834
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970108857940299e-05,
+      "loss": 0.3173,
+      "step": 20835
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9701048710358226e-05,
+      "loss": 0.3247,
+      "step": 20836
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.97010088386951e-05,
+      "loss": 0.301,
+      "step": 20837
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9700968964413624e-05,
+      "loss": 0.2834,
+      "step": 20838
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9700929087513817e-05,
+      "loss": 0.2689,
+      "step": 20839
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970088920799568e-05,
+      "loss": 0.2722,
+      "step": 20840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970084932585923e-05,
+      "loss": 0.2788,
+      "step": 20841
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970080944110447e-05,
+      "loss": 0.2762,
+      "step": 20842
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970076955373142e-05,
+      "loss": 0.309,
+      "step": 20843
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9700729663740085e-05,
+      "loss": 0.3095,
+      "step": 20844
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.970068977113048e-05,
+      "loss": 0.2996,
+      "step": 20845
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9700649875902615e-05,
+      "loss": 0.3717,
+      "step": 20846
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9700609978056497e-05,
+      "loss": 0.256,
+      "step": 20847
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700570077592135e-05,
+      "loss": 0.2564,
+      "step": 20848
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.970053017450955e-05,
+      "loss": 0.3187,
+      "step": 20849
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700490268808747e-05,
+      "loss": 0.2697,
+      "step": 20850
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.970045036048973e-05,
+      "loss": 0.3498,
+      "step": 20851
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700410449552524e-05,
+      "loss": 0.3238,
+      "step": 20852
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700370535997123e-05,
+      "loss": 0.312,
+      "step": 20853
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700330619823553e-05,
+      "loss": 0.3316,
+      "step": 20854
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.970029070103182e-05,
+      "loss": 0.3255,
+      "step": 20855
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700250779621927e-05,
+      "loss": 0.3458,
+      "step": 20856
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700210855593895e-05,
+      "loss": 0.2789,
+      "step": 20857
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700170928947728e-05,
+      "loss": 0.2987,
+      "step": 20858
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700130999683443e-05,
+      "loss": 0.3167,
+      "step": 20859
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700091067801047e-05,
+      "loss": 0.329,
+      "step": 20860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9700051133300547e-05,
+      "loss": 0.2567,
+      "step": 20861
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.970001119618196e-05,
+      "loss": 0.2904,
+      "step": 20862
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699971256445295e-05,
+      "loss": 0.3032,
+      "step": 20863
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699931314090563e-05,
+      "loss": 0.2746,
+      "step": 20864
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699891369117774e-05,
+      "loss": 0.299,
+      "step": 20865
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969985142152694e-05,
+      "loss": 0.259,
+      "step": 20866
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699811471318067e-05,
+      "loss": 0.2362,
+      "step": 20867
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699771518491172e-05,
+      "loss": 0.2676,
+      "step": 20868
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699731563046265e-05,
+      "loss": 0.3215,
+      "step": 20869
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699691604983355e-05,
+      "loss": 0.3148,
+      "step": 20870
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969965164430245e-05,
+      "loss": 0.2968,
+      "step": 20871
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699611681003567e-05,
+      "loss": 0.3066,
+      "step": 20872
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969957171508671e-05,
+      "loss": 0.3542,
+      "step": 20873
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699531746551893e-05,
+      "loss": 0.3313,
+      "step": 20874
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969949177539913e-05,
+      "loss": 0.2602,
+      "step": 20875
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699451801628426e-05,
+      "loss": 0.3228,
+      "step": 20876
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699411825239796e-05,
+      "loss": 0.2721,
+      "step": 20877
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969937184623325e-05,
+      "loss": 0.3007,
+      "step": 20878
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.96993318646088e-05,
+      "loss": 0.2764,
+      "step": 20879
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969929188036645e-05,
+      "loss": 0.2873,
+      "step": 20880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969925189350622e-05,
+      "loss": 0.319,
+      "step": 20881
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699211904028113e-05,
+      "loss": 0.3621,
+      "step": 20882
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699171911932148e-05,
+      "loss": 0.3513,
+      "step": 20883
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699131917218328e-05,
+      "loss": 0.3065,
+      "step": 20884
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969909191988667e-05,
+      "loss": 0.3109,
+      "step": 20885
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699051919937176e-05,
+      "loss": 0.2834,
+      "step": 20886
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9699011917369864e-05,
+      "loss": 0.3021,
+      "step": 20887
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969897191218475e-05,
+      "loss": 0.2762,
+      "step": 20888
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969893190438183e-05,
+      "loss": 0.2936,
+      "step": 20889
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698891893961128e-05,
+      "loss": 0.293,
+      "step": 20890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698851880922645e-05,
+      "loss": 0.2935,
+      "step": 20891
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698811865266404e-05,
+      "loss": 0.279,
+      "step": 20892
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698771846992404e-05,
+      "loss": 0.2615,
+      "step": 20893
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969873182610066e-05,
+      "loss": 0.2941,
+      "step": 20894
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698691802591182e-05,
+      "loss": 0.3022,
+      "step": 20895
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969865177646398e-05,
+      "loss": 0.3131,
+      "step": 20896
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969861174771907e-05,
+      "loss": 0.2475,
+      "step": 20897
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969857171635646e-05,
+      "loss": 0.2905,
+      "step": 20898
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698531682376162e-05,
+      "loss": 0.2841,
+      "step": 20899
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698491645778183e-05,
+      "loss": 0.2933,
+      "step": 20900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698451606562534e-05,
+      "loss": 0.288,
+      "step": 20901
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698411564729224e-05,
+      "loss": 0.2988,
+      "step": 20902
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698371520278273e-05,
+      "loss": 0.2727,
+      "step": 20903
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698331473209683e-05,
+      "loss": 0.315,
+      "step": 20904
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698291423523473e-05,
+      "loss": 0.3077,
+      "step": 20905
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698251371219643e-05,
+      "loss": 0.3149,
+      "step": 20906
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698211316298214e-05,
+      "loss": 0.2719,
+      "step": 20907
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969817125875919e-05,
+      "loss": 0.3208,
+      "step": 20908
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698131198602587e-05,
+      "loss": 0.3042,
+      "step": 20909
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698091135828407e-05,
+      "loss": 0.3217,
+      "step": 20910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698051070436674e-05,
+      "loss": 0.2599,
+      "step": 20911
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9698011002427387e-05,
+      "loss": 0.2726,
+      "step": 20912
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697970931800564e-05,
+      "loss": 0.3631,
+      "step": 20913
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697930858556212e-05,
+      "loss": 0.2705,
+      "step": 20914
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697890782694343e-05,
+      "loss": 0.3003,
+      "step": 20915
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969785070421497e-05,
+      "loss": 0.2926,
+      "step": 20916
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697810623118097e-05,
+      "loss": 0.3009,
+      "step": 20917
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697770539403745e-05,
+      "loss": 0.3095,
+      "step": 20918
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697730453071918e-05,
+      "loss": 0.3032,
+      "step": 20919
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697690364122627e-05,
+      "loss": 0.3379,
+      "step": 20920
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697650272555883e-05,
+      "loss": 0.2964,
+      "step": 20921
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.96976101783717e-05,
+      "loss": 0.3378,
+      "step": 20922
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697570081570087e-05,
+      "loss": 0.2799,
+      "step": 20923
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697529982151052e-05,
+      "loss": 0.346,
+      "step": 20924
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697489880114607e-05,
+      "loss": 0.3089,
+      "step": 20925
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969744977546077e-05,
+      "loss": 0.2278,
+      "step": 20926
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697409668189544e-05,
+      "loss": 0.2943,
+      "step": 20927
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969736955830094e-05,
+      "loss": 0.3008,
+      "step": 20928
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969732944579497e-05,
+      "loss": 0.2658,
+      "step": 20929
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697289330671646e-05,
+      "loss": 0.3094,
+      "step": 20930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969724921293098e-05,
+      "loss": 0.2754,
+      "step": 20931
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969720909257298e-05,
+      "loss": 0.2676,
+      "step": 20932
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697168969597657e-05,
+      "loss": 0.3056,
+      "step": 20933
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697128844005026e-05,
+      "loss": 0.2821,
+      "step": 20934
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969708871579509e-05,
+      "loss": 0.3022,
+      "step": 20935
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697048584967868e-05,
+      "loss": 0.2758,
+      "step": 20936
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9697008451523367e-05,
+      "loss": 0.245,
+      "step": 20937
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9696968315461596e-05,
+      "loss": 0.3336,
+      "step": 20938
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969692817678257e-05,
+      "loss": 0.2614,
+      "step": 20939
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9696888035486298e-05,
+      "loss": 0.2976,
+      "step": 20940
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969684789157279e-05,
+      "loss": 0.3085,
+      "step": 20941
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9696807745042056e-05,
+      "loss": 0.2963,
+      "step": 20942
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969676759589411e-05,
+      "loss": 0.2707,
+      "step": 20943
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.969672744412896e-05,
+      "loss": 0.2896,
+      "step": 20944
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9696687289746615e-05,
+      "loss": 0.3631,
+      "step": 20945
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9696647132747094e-05,
+      "loss": 0.3238,
+      "step": 20946
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.96966069731304e-05,
+      "loss": 0.3459,
+      "step": 20947
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9696566810896547e-05,
+      "loss": 0.2938,
+      "step": 20948
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9696526646045545e-05,
+      "loss": 0.3012,
+      "step": 20949
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9696486478577405e-05,
+      "loss": 0.2891,
+      "step": 20950
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969644630849214e-05,
+      "loss": 0.2877,
+      "step": 20951
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969640613578976e-05,
+      "loss": 0.2769,
+      "step": 20952
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969636596047027e-05,
+      "loss": 0.2765,
+      "step": 20953
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9696325782533687e-05,
+      "loss": 0.2709,
+      "step": 20954
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969628560198002e-05,
+      "loss": 0.2914,
+      "step": 20955
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969624541880928e-05,
+      "loss": 0.2578,
+      "step": 20956
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9696205233021478e-05,
+      "loss": 0.2706,
+      "step": 20957
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9696165044616628e-05,
+      "loss": 0.3583,
+      "step": 20958
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9696124853594737e-05,
+      "loss": 0.2729,
+      "step": 20959
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9696084659955813e-05,
+      "loss": 0.3065,
+      "step": 20960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9696044463699872e-05,
+      "loss": 0.2725,
+      "step": 20961
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9696004264826925e-05,
+      "loss": 0.273,
+      "step": 20962
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969596406333698e-05,
+      "loss": 0.2887,
+      "step": 20963
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695923859230047e-05,
+      "loss": 0.2714,
+      "step": 20964
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969588365250614e-05,
+      "loss": 0.3254,
+      "step": 20965
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695843443165273e-05,
+      "loss": 0.2889,
+      "step": 20966
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969580323120745e-05,
+      "loss": 0.2894,
+      "step": 20967
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695763016632682e-05,
+      "loss": 0.2868,
+      "step": 20968
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695722799440984e-05,
+      "loss": 0.2966,
+      "step": 20969
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695682579632363e-05,
+      "loss": 0.2961,
+      "step": 20970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695642357206835e-05,
+      "loss": 0.2884,
+      "step": 20971
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695602132164408e-05,
+      "loss": 0.3048,
+      "step": 20972
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969556190450509e-05,
+      "loss": 0.2821,
+      "step": 20973
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695521674228894e-05,
+      "loss": 0.3171,
+      "step": 20974
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695481441335834e-05,
+      "loss": 0.2826,
+      "step": 20975
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695441205825916e-05,
+      "loss": 0.3234,
+      "step": 20976
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695400967699158e-05,
+      "loss": 0.2857,
+      "step": 20977
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695360726955563e-05,
+      "loss": 0.3235,
+      "step": 20978
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969532048359514e-05,
+      "loss": 0.2832,
+      "step": 20979
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969528023761791e-05,
+      "loss": 0.2777,
+      "step": 20980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695239989023876e-05,
+      "loss": 0.2767,
+      "step": 20981
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695199737813054e-05,
+      "loss": 0.2643,
+      "step": 20982
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969515948398545e-05,
+      "loss": 0.2988,
+      "step": 20983
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969511922754108e-05,
+      "loss": 0.2726,
+      "step": 20984
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969507896847995e-05,
+      "loss": 0.2843,
+      "step": 20985
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9695038706802073e-05,
+      "loss": 0.2911,
+      "step": 20986
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969499844250746e-05,
+      "loss": 0.2929,
+      "step": 20987
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969495817559612e-05,
+      "loss": 0.2849,
+      "step": 20988
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969491790606807e-05,
+      "loss": 0.2773,
+      "step": 20989
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694877633923314e-05,
+      "loss": 0.2817,
+      "step": 20990
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694837359161862e-05,
+      "loss": 0.2567,
+      "step": 20991
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969479708178373e-05,
+      "loss": 0.2975,
+      "step": 20992
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969475680178893e-05,
+      "loss": 0.2932,
+      "step": 20993
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694716519177464e-05,
+      "loss": 0.3094,
+      "step": 20994
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694676233949352e-05,
+      "loss": 0.3424,
+      "step": 20995
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694635946104603e-05,
+      "loss": 0.3302,
+      "step": 20996
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694595655643223e-05,
+      "loss": 0.256,
+      "step": 20997
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969455536256523e-05,
+      "loss": 0.2676,
+      "step": 20998
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969451506687063e-05,
+      "loss": 0.3173,
+      "step": 20999
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694474768559436e-05,
+      "loss": 0.2809,
+      "step": 21000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694434467631652e-05,
+      "loss": 0.3566,
+      "step": 21001
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694394164087303e-05,
+      "loss": 0.2908,
+      "step": 21002
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694353857926388e-05,
+      "loss": 0.3349,
+      "step": 21003
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969431354914892e-05,
+      "loss": 0.2839,
+      "step": 21004
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694273237754913e-05,
+      "loss": 0.2911,
+      "step": 21005
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694232923744377e-05,
+      "loss": 0.2867,
+      "step": 21006
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694192607117323e-05,
+      "loss": 0.3568,
+      "step": 21007
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694152287873756e-05,
+      "loss": 0.3123,
+      "step": 21008
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694111966013695e-05,
+      "loss": 0.2612,
+      "step": 21009
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9694071641537153e-05,
+      "loss": 0.3307,
+      "step": 21010
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969403131444413e-05,
+      "loss": 0.3215,
+      "step": 21011
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693990984734644e-05,
+      "loss": 0.2988,
+      "step": 21012
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.96939506524087e-05,
+      "loss": 0.2524,
+      "step": 21013
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969391031746632e-05,
+      "loss": 0.269,
+      "step": 21014
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693869979907507e-05,
+      "loss": 0.3298,
+      "step": 21015
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693829639732273e-05,
+      "loss": 0.2454,
+      "step": 21016
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693789296940628e-05,
+      "loss": 0.2698,
+      "step": 21017
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693748951532587e-05,
+      "loss": 0.3153,
+      "step": 21018
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693708603508152e-05,
+      "loss": 0.2605,
+      "step": 21019
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969366825286734e-05,
+      "loss": 0.265,
+      "step": 21020
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693627899610167e-05,
+      "loss": 0.3612,
+      "step": 21021
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693587543736637e-05,
+      "loss": 0.2788,
+      "step": 21022
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693547185246762e-05,
+      "loss": 0.2699,
+      "step": 21023
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969350682414055e-05,
+      "loss": 0.2889,
+      "step": 21024
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969346646041802e-05,
+      "loss": 0.2419,
+      "step": 21025
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693426094079178e-05,
+      "loss": 0.3093,
+      "step": 21026
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969338572512403e-05,
+      "loss": 0.3167,
+      "step": 21027
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.96933453535526e-05,
+      "loss": 0.2973,
+      "step": 21028
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693304979364882e-05,
+      "loss": 0.2771,
+      "step": 21029
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.96932646025609e-05,
+      "loss": 0.3256,
+      "step": 21030
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693224223140658e-05,
+      "loss": 0.317,
+      "step": 21031
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693183841104174e-05,
+      "loss": 0.3229,
+      "step": 21032
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969314345645145e-05,
+      "loss": 0.2913,
+      "step": 21033
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.96931030691825e-05,
+      "loss": 0.3138,
+      "step": 21034
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9693062679297343e-05,
+      "loss": 0.3173,
+      "step": 21035
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969302228679598e-05,
+      "loss": 0.294,
+      "step": 21036
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969298189167842e-05,
+      "loss": 0.2843,
+      "step": 21037
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9692941493944686e-05,
+      "loss": 0.3162,
+      "step": 21038
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.969290109359478e-05,
+      "loss": 0.3053,
+      "step": 21039
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9692860690628713e-05,
+      "loss": 0.2875,
+      "step": 21040
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9692820285046496e-05,
+      "loss": 0.2718,
+      "step": 21041
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9692779876848146e-05,
+      "loss": 0.3309,
+      "step": 21042
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9692739466033667e-05,
+      "loss": 0.2775,
+      "step": 21043
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9692699052603073e-05,
+      "loss": 0.3174,
+      "step": 21044
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692658636556373e-05,
+      "loss": 0.2541,
+      "step": 21045
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969261821789358e-05,
+      "loss": 0.2946,
+      "step": 21046
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692577796614706e-05,
+      "loss": 0.2498,
+      "step": 21047
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692537372719755e-05,
+      "loss": 0.3145,
+      "step": 21048
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692496946208746e-05,
+      "loss": 0.2897,
+      "step": 21049
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692456517081686e-05,
+      "loss": 0.3213,
+      "step": 21050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692416085338588e-05,
+      "loss": 0.3311,
+      "step": 21051
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969237565097946e-05,
+      "loss": 0.2802,
+      "step": 21052
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692335214004314e-05,
+      "loss": 0.3349,
+      "step": 21053
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692294774413162e-05,
+      "loss": 0.2715,
+      "step": 21054
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692254332206016e-05,
+      "loss": 0.2739,
+      "step": 21055
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692213887382886e-05,
+      "loss": 0.2676,
+      "step": 21056
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969217343994378e-05,
+      "loss": 0.2636,
+      "step": 21057
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969213298988871e-05,
+      "loss": 0.2745,
+      "step": 21058
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969209253721769e-05,
+      "loss": 0.2804,
+      "step": 21059
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692052081930728e-05,
+      "loss": 0.2396,
+      "step": 21060
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9692011624027835e-05,
+      "loss": 0.2946,
+      "step": 21061
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691971163509026e-05,
+      "loss": 0.269,
+      "step": 21062
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691930700374304e-05,
+      "loss": 0.2752,
+      "step": 21063
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969189023462369e-05,
+      "loss": 0.236,
+      "step": 21064
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691849766257186e-05,
+      "loss": 0.3003,
+      "step": 21065
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691809295274806e-05,
+      "loss": 0.3156,
+      "step": 21066
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691768821676565e-05,
+      "loss": 0.3102,
+      "step": 21067
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691728345462465e-05,
+      "loss": 0.2779,
+      "step": 21068
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969168786663253e-05,
+      "loss": 0.3254,
+      "step": 21069
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691647385186754e-05,
+      "loss": 0.2878,
+      "step": 21070
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691606901125162e-05,
+      "loss": 0.365,
+      "step": 21071
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969156641444776e-05,
+      "loss": 0.3078,
+      "step": 21072
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691525925154557e-05,
+      "loss": 0.3081,
+      "step": 21073
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691485433245567e-05,
+      "loss": 0.2438,
+      "step": 21074
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691444938720797e-05,
+      "loss": 0.2503,
+      "step": 21075
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691404441580268e-05,
+      "loss": 0.3057,
+      "step": 21076
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969136394182398e-05,
+      "loss": 0.3218,
+      "step": 21077
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691323439451944e-05,
+      "loss": 0.2676,
+      "step": 21078
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969128293446418e-05,
+      "loss": 0.3658,
+      "step": 21079
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969124242686069e-05,
+      "loss": 0.3408,
+      "step": 21080
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691201916641486e-05,
+      "loss": 0.274,
+      "step": 21081
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691161403806585e-05,
+      "loss": 0.2919,
+      "step": 21082
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691120888355995e-05,
+      "loss": 0.2967,
+      "step": 21083
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691080370289723e-05,
+      "loss": 0.3218,
+      "step": 21084
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9691039849607783e-05,
+      "loss": 0.2607,
+      "step": 21085
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969099932631019e-05,
+      "loss": 0.2592,
+      "step": 21086
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969095880039695e-05,
+      "loss": 0.2926,
+      "step": 21087
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690918271868073e-05,
+      "loss": 0.2929,
+      "step": 21088
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690877740723572e-05,
+      "loss": 0.2731,
+      "step": 21089
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690837206963457e-05,
+      "loss": 0.3014,
+      "step": 21090
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690796670587745e-05,
+      "loss": 0.2982,
+      "step": 21091
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690756131596436e-05,
+      "loss": 0.2539,
+      "step": 21092
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969071558998955e-05,
+      "loss": 0.2784,
+      "step": 21093
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969067504576709e-05,
+      "loss": 0.3018,
+      "step": 21094
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690634498929074e-05,
+      "loss": 0.3086,
+      "step": 21095
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690593949475512e-05,
+      "loss": 0.2743,
+      "step": 21096
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969055339740641e-05,
+      "loss": 0.2787,
+      "step": 21097
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690512842721787e-05,
+      "loss": 0.2851,
+      "step": 21098
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690472285421647e-05,
+      "loss": 0.2988,
+      "step": 21099
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690431725506e-05,
+      "loss": 0.2646,
+      "step": 21100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690391162974867e-05,
+      "loss": 0.3375,
+      "step": 21101
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690350597828248e-05,
+      "loss": 0.2668,
+      "step": 21102
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690310030066157e-05,
+      "loss": 0.3871,
+      "step": 21103
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969026945968861e-05,
+      "loss": 0.3059,
+      "step": 21104
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.969022888669561e-05,
+      "loss": 0.3018,
+      "step": 21105
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690188311087175e-05,
+      "loss": 0.2693,
+      "step": 21106
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690147732863313e-05,
+      "loss": 0.2819,
+      "step": 21107
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690107152024032e-05,
+      "loss": 0.3655,
+      "step": 21108
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690066568569345e-05,
+      "loss": 0.3097,
+      "step": 21109
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9690025982499267e-05,
+      "loss": 0.2963,
+      "step": 21110
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689985393813805e-05,
+      "loss": 0.2668,
+      "step": 21111
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.968994480251297e-05,
+      "loss": 0.2888,
+      "step": 21112
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.968990420859677e-05,
+      "loss": 0.2487,
+      "step": 21113
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689863612065226e-05,
+      "loss": 0.2833,
+      "step": 21114
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689823012918342e-05,
+      "loss": 0.319,
+      "step": 21115
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689782411156125e-05,
+      "loss": 0.2974,
+      "step": 21116
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689741806778592e-05,
+      "loss": 0.3742,
+      "step": 21117
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689701199785753e-05,
+      "loss": 0.2688,
+      "step": 21118
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.968966059017762e-05,
+      "loss": 0.2721,
+      "step": 21119
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.96896199779542e-05,
+      "loss": 0.3278,
+      "step": 21120
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689579363115507e-05,
+      "loss": 0.2887,
+      "step": 21121
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.968953874566155e-05,
+      "loss": 0.3057,
+      "step": 21122
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689498125592346e-05,
+      "loss": 0.3163,
+      "step": 21123
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689457502907897e-05,
+      "loss": 0.2679,
+      "step": 21124
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.968941687760822e-05,
+      "loss": 0.2818,
+      "step": 21125
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689376249693318e-05,
+      "loss": 0.3086,
+      "step": 21126
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689335619163216e-05,
+      "loss": 0.2824,
+      "step": 21127
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689294986017913e-05,
+      "loss": 0.3028,
+      "step": 21128
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689254350257426e-05,
+      "loss": 0.3428,
+      "step": 21129
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689213711881763e-05,
+      "loss": 0.3177,
+      "step": 21130
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689173070890936e-05,
+      "loss": 0.267,
+      "step": 21131
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689132427284952e-05,
+      "loss": 0.3114,
+      "step": 21132
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689091781063832e-05,
+      "loss": 0.2386,
+      "step": 21133
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.968905113222758e-05,
+      "loss": 0.3312,
+      "step": 21134
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9689010480776204e-05,
+      "loss": 0.305,
+      "step": 21135
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9688969826709722e-05,
+      "loss": 0.3028,
+      "step": 21136
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9688929170028142e-05,
+      "loss": 0.3228,
+      "step": 21137
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9688888510731473e-05,
+      "loss": 0.3089,
+      "step": 21138
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9688847848819725e-05,
+      "loss": 0.2563,
+      "step": 21139
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.968880718429292e-05,
+      "loss": 0.2637,
+      "step": 21140
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9688766517151054e-05,
+      "loss": 0.32,
+      "step": 21141
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9688725847394144e-05,
+      "loss": 0.3362,
+      "step": 21142
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9688685175022203e-05,
+      "loss": 0.3056,
+      "step": 21143
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9688644500035245e-05,
+      "loss": 0.2879,
+      "step": 21144
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968860382243327e-05,
+      "loss": 0.28,
+      "step": 21145
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.96885631422163e-05,
+      "loss": 0.2965,
+      "step": 21146
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9688522459384338e-05,
+      "loss": 0.2963,
+      "step": 21147
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9688481773937403e-05,
+      "loss": 0.324,
+      "step": 21148
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9688441085875495e-05,
+      "loss": 0.3164,
+      "step": 21149
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9688400395198637e-05,
+      "loss": 0.2618,
+      "step": 21150
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9688359701906836e-05,
+      "loss": 0.2584,
+      "step": 21151
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9688319006000095e-05,
+      "loss": 0.2931,
+      "step": 21152
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9688278307478434e-05,
+      "loss": 0.267,
+      "step": 21153
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968823760634186e-05,
+      "loss": 0.2779,
+      "step": 21154
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968819690259039e-05,
+      "loss": 0.3063,
+      "step": 21155
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9688156196224027e-05,
+      "loss": 0.3285,
+      "step": 21156
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9688115487242787e-05,
+      "loss": 0.2594,
+      "step": 21157
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968807477564668e-05,
+      "loss": 0.3489,
+      "step": 21158
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968803406143571e-05,
+      "loss": 0.3001,
+      "step": 21159
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687993344609902e-05,
+      "loss": 0.2599,
+      "step": 21160
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968795262516926e-05,
+      "loss": 0.3088,
+      "step": 21161
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687911903113788e-05,
+      "loss": 0.2407,
+      "step": 21162
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968787117844351e-05,
+      "loss": 0.2921,
+      "step": 21163
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968783045115842e-05,
+      "loss": 0.264,
+      "step": 21164
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687789721258548e-05,
+      "loss": 0.2807,
+      "step": 21165
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687748988743893e-05,
+      "loss": 0.2963,
+      "step": 21166
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968770825361447e-05,
+      "loss": 0.2999,
+      "step": 21167
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968766751587029e-05,
+      "loss": 0.3412,
+      "step": 21168
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968762677551136e-05,
+      "loss": 0.2831,
+      "step": 21169
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.96875860325377e-05,
+      "loss": 0.2932,
+      "step": 21170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968754528694931e-05,
+      "loss": 0.2494,
+      "step": 21171
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968750453874621e-05,
+      "loss": 0.3041,
+      "step": 21172
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687463787928407e-05,
+      "loss": 0.2457,
+      "step": 21173
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687423034495908e-05,
+      "loss": 0.3162,
+      "step": 21174
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687382278448734e-05,
+      "loss": 0.3175,
+      "step": 21175
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687341519786883e-05,
+      "loss": 0.2757,
+      "step": 21176
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687300758510378e-05,
+      "loss": 0.2657,
+      "step": 21177
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687259994619228e-05,
+      "loss": 0.2765,
+      "step": 21178
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687219228113436e-05,
+      "loss": 0.2796,
+      "step": 21179
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687178458993022e-05,
+      "loss": 0.2709,
+      "step": 21180
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968713768725799e-05,
+      "loss": 0.3205,
+      "step": 21181
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687096912908355e-05,
+      "loss": 0.3788,
+      "step": 21182
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9687056135944132e-05,
+      "loss": 0.254,
+      "step": 21183
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968701535636532e-05,
+      "loss": 0.3109,
+      "step": 21184
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968697457417194e-05,
+      "loss": 0.3002,
+      "step": 21185
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9686933789364003e-05,
+      "loss": 0.2926,
+      "step": 21186
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9686893001941515e-05,
+      "loss": 0.2693,
+      "step": 21187
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968685221190449e-05,
+      "loss": 0.2623,
+      "step": 21188
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9686811419252936e-05,
+      "loss": 0.2901,
+      "step": 21189
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968677062398687e-05,
+      "loss": 0.3158,
+      "step": 21190
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9686729826106295e-05,
+      "loss": 0.3413,
+      "step": 21191
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968668902561123e-05,
+      "loss": 0.2927,
+      "step": 21192
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968664822250168e-05,
+      "loss": 0.3086,
+      "step": 21193
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968660741677766e-05,
+      "loss": 0.267,
+      "step": 21194
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968656660843918e-05,
+      "loss": 0.3069,
+      "step": 21195
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968652579748625e-05,
+      "loss": 0.257,
+      "step": 21196
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9686484983918877e-05,
+      "loss": 0.3054,
+      "step": 21197
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9686444167737083e-05,
+      "loss": 0.2917,
+      "step": 21198
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968640334894087e-05,
+      "loss": 0.282,
+      "step": 21199
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968636252753025e-05,
+      "loss": 0.3592,
+      "step": 21200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9686321703505236e-05,
+      "loss": 0.2903,
+      "step": 21201
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968628087686584e-05,
+      "loss": 0.3064,
+      "step": 21202
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9686240047612072e-05,
+      "loss": 0.3167,
+      "step": 21203
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968619921574394e-05,
+      "loss": 0.3106,
+      "step": 21204
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968615838126146e-05,
+      "loss": 0.3019,
+      "step": 21205
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968611754416464e-05,
+      "loss": 0.2746,
+      "step": 21206
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9686076704453486e-05,
+      "loss": 0.2936,
+      "step": 21207
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968603586212802e-05,
+      "loss": 0.2945,
+      "step": 21208
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685995017188247e-05,
+      "loss": 0.3644,
+      "step": 21209
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968595416963418e-05,
+      "loss": 0.2809,
+      "step": 21210
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685913319465828e-05,
+      "loss": 0.2954,
+      "step": 21211
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.96858724666832e-05,
+      "loss": 0.3308,
+      "step": 21212
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.3127,
+      "step": 21213
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685790753275175e-05,
+      "loss": 0.3451,
+      "step": 21214
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685749892649792e-05,
+      "loss": 0.2632,
+      "step": 21215
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685709029410185e-05,
+      "loss": 0.297,
+      "step": 21216
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685668163556356e-05,
+      "loss": 0.3051,
+      "step": 21217
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685627295088326e-05,
+      "loss": 0.272,
+      "step": 21218
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685586424006094e-05,
+      "loss": 0.3037,
+      "step": 21219
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968554555030968e-05,
+      "loss": 0.2881,
+      "step": 21220
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968550467399909e-05,
+      "loss": 0.2901,
+      "step": 21221
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968546379507434e-05,
+      "loss": 0.2547,
+      "step": 21222
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685422913535435e-05,
+      "loss": 0.3238,
+      "step": 21223
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968538202938239e-05,
+      "loss": 0.2914,
+      "step": 21224
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685341142615213e-05,
+      "loss": 0.2606,
+      "step": 21225
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968530025323392e-05,
+      "loss": 0.2562,
+      "step": 21226
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968525936123852e-05,
+      "loss": 0.284,
+      "step": 21227
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968521846662902e-05,
+      "loss": 0.3045,
+      "step": 21228
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685177569405437e-05,
+      "loss": 0.266,
+      "step": 21229
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968513666956778e-05,
+      "loss": 0.2909,
+      "step": 21230
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9685095767116055e-05,
+      "loss": 0.3152,
+      "step": 21231
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968505486205028e-05,
+      "loss": 0.3046,
+      "step": 21232
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968501395437046e-05,
+      "loss": 0.3146,
+      "step": 21233
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9684973044076615e-05,
+      "loss": 0.29,
+      "step": 21234
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9684932131168747e-05,
+      "loss": 0.2722,
+      "step": 21235
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9684891215646874e-05,
+      "loss": 0.2755,
+      "step": 21236
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9684850297511e-05,
+      "loss": 0.3348,
+      "step": 21237
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9684809376761142e-05,
+      "loss": 0.2815,
+      "step": 21238
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9684768453397307e-05,
+      "loss": 0.2781,
+      "step": 21239
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.968472752741951e-05,
+      "loss": 0.3118,
+      "step": 21240
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9684686598827757e-05,
+      "loss": 0.2576,
+      "step": 21241
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684645667622063e-05,
+      "loss": 0.286,
+      "step": 21242
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968460473380244e-05,
+      "loss": 0.2453,
+      "step": 21243
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684563797368895e-05,
+      "loss": 0.2584,
+      "step": 21244
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968452285832144e-05,
+      "loss": 0.2767,
+      "step": 21245
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684481916660085e-05,
+      "loss": 0.3433,
+      "step": 21246
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968444097238485e-05,
+      "loss": 0.348,
+      "step": 21247
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684400025495733e-05,
+      "loss": 0.2667,
+      "step": 21248
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684359075992755e-05,
+      "loss": 0.3198,
+      "step": 21249
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968431812387592e-05,
+      "loss": 0.2944,
+      "step": 21250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968427716914524e-05,
+      "loss": 0.2675,
+      "step": 21251
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684236211800733e-05,
+      "loss": 0.3136,
+      "step": 21252
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684195251842407e-05,
+      "loss": 0.2976,
+      "step": 21253
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684154289270267e-05,
+      "loss": 0.2848,
+      "step": 21254
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684113324084332e-05,
+      "loss": 0.314,
+      "step": 21255
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684072356284606e-05,
+      "loss": 0.2942,
+      "step": 21256
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9684031385871105e-05,
+      "loss": 0.316,
+      "step": 21257
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683990412843837e-05,
+      "loss": 0.2826,
+      "step": 21258
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683949437202822e-05,
+      "loss": 0.2694,
+      "step": 21259
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683908458948057e-05,
+      "loss": 0.3255,
+      "step": 21260
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968386747807956e-05,
+      "loss": 0.2803,
+      "step": 21261
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968382649459734e-05,
+      "loss": 0.2713,
+      "step": 21262
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683785508501415e-05,
+      "loss": 0.2707,
+      "step": 21263
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968374451979179e-05,
+      "loss": 0.36,
+      "step": 21264
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683703528468475e-05,
+      "loss": 0.2929,
+      "step": 21265
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683662534531487e-05,
+      "loss": 0.2766,
+      "step": 21266
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968362153798083e-05,
+      "loss": 0.3507,
+      "step": 21267
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968358053881652e-05,
+      "loss": 0.2492,
+      "step": 21268
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683539537038566e-05,
+      "loss": 0.34,
+      "step": 21269
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683498532646975e-05,
+      "loss": 0.3287,
+      "step": 21270
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968345752564177e-05,
+      "loss": 0.2489,
+      "step": 21271
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968341651602295e-05,
+      "loss": 0.27,
+      "step": 21272
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968337550379053e-05,
+      "loss": 0.3323,
+      "step": 21273
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683334488944522e-05,
+      "loss": 0.2906,
+      "step": 21274
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683293471484938e-05,
+      "loss": 0.3028,
+      "step": 21275
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968325245141179e-05,
+      "loss": 0.3095,
+      "step": 21276
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683211428725084e-05,
+      "loss": 0.3401,
+      "step": 21277
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683170403424835e-05,
+      "loss": 0.2448,
+      "step": 21278
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968312937551105e-05,
+      "loss": 0.3216,
+      "step": 21279
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683088344983748e-05,
+      "loss": 0.2721,
+      "step": 21280
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683047311842935e-05,
+      "loss": 0.2658,
+      "step": 21281
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9683006276088618e-05,
+      "loss": 0.256,
+      "step": 21282
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682965237720815e-05,
+      "loss": 0.2615,
+      "step": 21283
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682924196739536e-05,
+      "loss": 0.2486,
+      "step": 21284
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682883153144784e-05,
+      "loss": 0.285,
+      "step": 21285
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682842106936584e-05,
+      "loss": 0.287,
+      "step": 21286
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682801058114934e-05,
+      "loss": 0.3296,
+      "step": 21287
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682760006679856e-05,
+      "loss": 0.2601,
+      "step": 21288
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682718952631352e-05,
+      "loss": 0.265,
+      "step": 21289
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968267789596944e-05,
+      "loss": 0.3158,
+      "step": 21290
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682636836694126e-05,
+      "loss": 0.29,
+      "step": 21291
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968259577480542e-05,
+      "loss": 0.2794,
+      "step": 21292
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968255471030334e-05,
+      "loss": 0.3009,
+      "step": 21293
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682513643187894e-05,
+      "loss": 0.3006,
+      "step": 21294
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968247257345909e-05,
+      "loss": 0.3423,
+      "step": 21295
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968243150111694e-05,
+      "loss": 0.3329,
+      "step": 21296
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968239042616146e-05,
+      "loss": 0.2971,
+      "step": 21297
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968234934859266e-05,
+      "loss": 0.2623,
+      "step": 21298
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682308268410543e-05,
+      "loss": 0.2899,
+      "step": 21299
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682267185615127e-05,
+      "loss": 0.3418,
+      "step": 21300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968222610020642e-05,
+      "loss": 0.2843,
+      "step": 21301
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682185012184442e-05,
+      "loss": 0.2624,
+      "step": 21302
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968214392154919e-05,
+      "loss": 0.3005,
+      "step": 21303
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682102828300683e-05,
+      "loss": 0.2891,
+      "step": 21304
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9682061732438936e-05,
+      "loss": 0.2684,
+      "step": 21305
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968202063396395e-05,
+      "loss": 0.2575,
+      "step": 21306
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681979532875744e-05,
+      "loss": 0.2957,
+      "step": 21307
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681938429174324e-05,
+      "loss": 0.2961,
+      "step": 21308
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681897322859707e-05,
+      "loss": 0.2913,
+      "step": 21309
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681856213931898e-05,
+      "loss": 0.295,
+      "step": 21310
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968181510239091e-05,
+      "loss": 0.2556,
+      "step": 21311
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968177398823676e-05,
+      "loss": 0.2926,
+      "step": 21312
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681732871469447e-05,
+      "loss": 0.2615,
+      "step": 21313
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681691752088997e-05,
+      "loss": 0.346,
+      "step": 21314
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681650630095407e-05,
+      "loss": 0.2694,
+      "step": 21315
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681609505488696e-05,
+      "loss": 0.3273,
+      "step": 21316
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681568378268872e-05,
+      "loss": 0.261,
+      "step": 21317
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968152724843595e-05,
+      "loss": 0.2995,
+      "step": 21318
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681486115989937e-05,
+      "loss": 0.3026,
+      "step": 21319
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681444980930843e-05,
+      "loss": 0.2824,
+      "step": 21320
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681403843258686e-05,
+      "loss": 0.3647,
+      "step": 21321
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681362702973474e-05,
+      "loss": 0.307,
+      "step": 21322
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968132156007521e-05,
+      "loss": 0.3131,
+      "step": 21323
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681280414563918e-05,
+      "loss": 0.2686,
+      "step": 21324
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.96812392664396e-05,
+      "loss": 0.296,
+      "step": 21325
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681198115702273e-05,
+      "loss": 0.2657,
+      "step": 21326
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681156962351947e-05,
+      "loss": 0.3181,
+      "step": 21327
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968111580638863e-05,
+      "loss": 0.2924,
+      "step": 21328
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681074647812332e-05,
+      "loss": 0.274,
+      "step": 21329
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9681033486623068e-05,
+      "loss": 0.2729,
+      "step": 21330
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968099232282085e-05,
+      "loss": 0.2717,
+      "step": 21331
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9680951156405683e-05,
+      "loss": 0.3349,
+      "step": 21332
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9680909987377583e-05,
+      "loss": 0.2864,
+      "step": 21333
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968086881573656e-05,
+      "loss": 0.2621,
+      "step": 21334
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968082764148263e-05,
+      "loss": 0.3015,
+      "step": 21335
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9680786464615794e-05,
+      "loss": 0.2896,
+      "step": 21336
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.968074528513607e-05,
+      "loss": 0.2631,
+      "step": 21337
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9680704103043466e-05,
+      "loss": 0.2666,
+      "step": 21338
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9680662918338e-05,
+      "loss": 0.3015,
+      "step": 21339
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9680621731019673e-05,
+      "loss": 0.2871,
+      "step": 21340
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.96805805410885e-05,
+      "loss": 0.2486,
+      "step": 21341
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9680539348544496e-05,
+      "loss": 0.2882,
+      "step": 21342
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.968049815338767e-05,
+      "loss": 0.3263,
+      "step": 21343
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.968045695561803e-05,
+      "loss": 0.3071,
+      "step": 21344
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.968041575523559e-05,
+      "loss": 0.2898,
+      "step": 21345
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9680374552240356e-05,
+      "loss": 0.28,
+      "step": 21346
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9680333346632352e-05,
+      "loss": 0.3483,
+      "step": 21347
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9680292138411577e-05,
+      "loss": 0.32,
+      "step": 21348
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9680250927578047e-05,
+      "loss": 0.2801,
+      "step": 21349
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.968020971413177e-05,
+      "loss": 0.303,
+      "step": 21350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.968016849807276e-05,
+      "loss": 0.3251,
+      "step": 21351
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9680127279401026e-05,
+      "loss": 0.2771,
+      "step": 21352
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9680086058116584e-05,
+      "loss": 0.2853,
+      "step": 21353
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9680044834219438e-05,
+      "loss": 0.3119,
+      "step": 21354
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9680003607709605e-05,
+      "loss": 0.2722,
+      "step": 21355
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679962378587092e-05,
+      "loss": 0.2995,
+      "step": 21356
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679921146851913e-05,
+      "loss": 0.2755,
+      "step": 21357
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679879912504075e-05,
+      "loss": 0.3244,
+      "step": 21358
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679838675543597e-05,
+      "loss": 0.2711,
+      "step": 21359
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679797435970483e-05,
+      "loss": 0.3015,
+      "step": 21360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679756193784747e-05,
+      "loss": 0.3314,
+      "step": 21361
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.96797149489864e-05,
+      "loss": 0.2724,
+      "step": 21362
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967967370157545e-05,
+      "loss": 0.3167,
+      "step": 21363
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679632451551914e-05,
+      "loss": 0.2877,
+      "step": 21364
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.96795911989158e-05,
+      "loss": 0.2684,
+      "step": 21365
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679549943667115e-05,
+      "loss": 0.3077,
+      "step": 21366
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679508685805876e-05,
+      "loss": 0.3059,
+      "step": 21367
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679467425332095e-05,
+      "loss": 0.2599,
+      "step": 21368
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679426162245777e-05,
+      "loss": 0.2857,
+      "step": 21369
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967938489654694e-05,
+      "loss": 0.3003,
+      "step": 21370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967934362823559e-05,
+      "loss": 0.2792,
+      "step": 21371
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967930235731174e-05,
+      "loss": 0.2975,
+      "step": 21372
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679261083775404e-05,
+      "loss": 0.3084,
+      "step": 21373
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679219807626584e-05,
+      "loss": 0.2611,
+      "step": 21374
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.96791785288653e-05,
+      "loss": 0.2966,
+      "step": 21375
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679137247491562e-05,
+      "loss": 0.2745,
+      "step": 21376
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967909596350538e-05,
+      "loss": 0.2643,
+      "step": 21377
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679054676906763e-05,
+      "loss": 0.3374,
+      "step": 21378
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9679013387695726e-05,
+      "loss": 0.2839,
+      "step": 21379
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678972095872274e-05,
+      "loss": 0.2536,
+      "step": 21380
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678930801436426e-05,
+      "loss": 0.2807,
+      "step": 21381
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678889504388187e-05,
+      "loss": 0.2676,
+      "step": 21382
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678848204727572e-05,
+      "loss": 0.3017,
+      "step": 21383
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967880690245459e-05,
+      "loss": 0.3241,
+      "step": 21384
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678765597569252e-05,
+      "loss": 0.3088,
+      "step": 21385
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967872429007157e-05,
+      "loss": 0.319,
+      "step": 21386
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678682979961557e-05,
+      "loss": 0.2763,
+      "step": 21387
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967864166723922e-05,
+      "loss": 0.2957,
+      "step": 21388
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678600351904573e-05,
+      "loss": 0.3431,
+      "step": 21389
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678559033957628e-05,
+      "loss": 0.2617,
+      "step": 21390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678517713398393e-05,
+      "loss": 0.2518,
+      "step": 21391
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678476390226882e-05,
+      "loss": 0.3255,
+      "step": 21392
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678435064443104e-05,
+      "loss": 0.3108,
+      "step": 21393
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678393736047073e-05,
+      "loss": 0.2885,
+      "step": 21394
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678352405038795e-05,
+      "loss": 0.307,
+      "step": 21395
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967831107141829e-05,
+      "loss": 0.3148,
+      "step": 21396
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678269735185556e-05,
+      "loss": 0.2907,
+      "step": 21397
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678228396340615e-05,
+      "loss": 0.3315,
+      "step": 21398
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678187054883478e-05,
+      "loss": 0.2958,
+      "step": 21399
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967814571081415e-05,
+      "loss": 0.2772,
+      "step": 21400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678104364132645e-05,
+      "loss": 0.2903,
+      "step": 21401
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9678063014838976e-05,
+      "loss": 0.2548,
+      "step": 21402
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967802166293315e-05,
+      "loss": 0.3033,
+      "step": 21403
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677980308415185e-05,
+      "loss": 0.3265,
+      "step": 21404
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677938951285082e-05,
+      "loss": 0.2608,
+      "step": 21405
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677897591542865e-05,
+      "loss": 0.243,
+      "step": 21406
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967785622918853e-05,
+      "loss": 0.2886,
+      "step": 21407
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677814864222103e-05,
+      "loss": 0.272,
+      "step": 21408
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967777349664359e-05,
+      "loss": 0.3323,
+      "step": 21409
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677732126452994e-05,
+      "loss": 0.3012,
+      "step": 21410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677690753650335e-05,
+      "loss": 0.2848,
+      "step": 21411
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677649378235624e-05,
+      "loss": 0.3209,
+      "step": 21412
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967760800020887e-05,
+      "loss": 0.3016,
+      "step": 21413
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677566619570086e-05,
+      "loss": 0.3158,
+      "step": 21414
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677525236319276e-05,
+      "loss": 0.3614,
+      "step": 21415
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967748385045646e-05,
+      "loss": 0.2878,
+      "step": 21416
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677442461981646e-05,
+      "loss": 0.3193,
+      "step": 21417
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677401070894847e-05,
+      "loss": 0.2806,
+      "step": 21418
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967735967719607e-05,
+      "loss": 0.2553,
+      "step": 21419
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677318280885325e-05,
+      "loss": 0.3072,
+      "step": 21420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967727688196263e-05,
+      "loss": 0.2929,
+      "step": 21421
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677235480427995e-05,
+      "loss": 0.3023,
+      "step": 21422
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677194076281428e-05,
+      "loss": 0.3618,
+      "step": 21423
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967715266952294e-05,
+      "loss": 0.2997,
+      "step": 21424
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677111260152538e-05,
+      "loss": 0.296,
+      "step": 21425
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967706984817025e-05,
+      "loss": 0.3,
+      "step": 21426
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9677028433576065e-05,
+      "loss": 0.2674,
+      "step": 21427
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9676987016370012e-05,
+      "loss": 0.2711,
+      "step": 21428
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967694559655209e-05,
+      "loss": 0.2879,
+      "step": 21429
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9676904174122315e-05,
+      "loss": 0.2538,
+      "step": 21430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.96768627490807e-05,
+      "loss": 0.2807,
+      "step": 21431
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9676821321427254e-05,
+      "loss": 0.3423,
+      "step": 21432
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967677989116199e-05,
+      "loss": 0.2981,
+      "step": 21433
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9676738458284918e-05,
+      "loss": 0.2555,
+      "step": 21434
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9676697022796048e-05,
+      "loss": 0.3275,
+      "step": 21435
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9676655584695393e-05,
+      "loss": 0.3084,
+      "step": 21436
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9676614143982962e-05,
+      "loss": 0.3105,
+      "step": 21437
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.967657270065877e-05,
+      "loss": 0.3218,
+      "step": 21438
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9676531254722823e-05,
+      "loss": 0.2942,
+      "step": 21439
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9676489806175138e-05,
+      "loss": 0.3028,
+      "step": 21440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967644835501572e-05,
+      "loss": 0.2726,
+      "step": 21441
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9676406901244584e-05,
+      "loss": 0.342,
+      "step": 21442
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9676365444861742e-05,
+      "loss": 0.3254,
+      "step": 21443
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9676323985867203e-05,
+      "loss": 0.2579,
+      "step": 21444
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967628252426098e-05,
+      "loss": 0.271,
+      "step": 21445
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9676241060043084e-05,
+      "loss": 0.309,
+      "step": 21446
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9676199593213525e-05,
+      "loss": 0.3506,
+      "step": 21447
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967615812377231e-05,
+      "loss": 0.3292,
+      "step": 21448
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967611665171946e-05,
+      "loss": 0.2896,
+      "step": 21449
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967607517705498e-05,
+      "loss": 0.2756,
+      "step": 21450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9676033699778878e-05,
+      "loss": 0.2742,
+      "step": 21451
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967599221989117e-05,
+      "loss": 0.2639,
+      "step": 21452
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967595073739187e-05,
+      "loss": 0.3051,
+      "step": 21453
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675909252280985e-05,
+      "loss": 0.2953,
+      "step": 21454
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675867764558524e-05,
+      "loss": 0.3264,
+      "step": 21455
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675826274224504e-05,
+      "loss": 0.2675,
+      "step": 21456
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675784781278933e-05,
+      "loss": 0.2762,
+      "step": 21457
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967574328572182e-05,
+      "loss": 0.2558,
+      "step": 21458
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967570178755318e-05,
+      "loss": 0.3463,
+      "step": 21459
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675660286773026e-05,
+      "loss": 0.2738,
+      "step": 21460
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675618783381363e-05,
+      "loss": 0.2937,
+      "step": 21461
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675577277378207e-05,
+      "loss": 0.2957,
+      "step": 21462
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675535768763563e-05,
+      "loss": 0.3207,
+      "step": 21463
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675494257537452e-05,
+      "loss": 0.3234,
+      "step": 21464
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675452743699878e-05,
+      "loss": 0.2986,
+      "step": 21465
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967541122725085e-05,
+      "loss": 0.2988,
+      "step": 21466
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967536970819039e-05,
+      "loss": 0.271,
+      "step": 21467
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.96753281865185e-05,
+      "loss": 0.3318,
+      "step": 21468
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675286662235193e-05,
+      "loss": 0.2976,
+      "step": 21469
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967524513534048e-05,
+      "loss": 0.2514,
+      "step": 21470
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675203605834375e-05,
+      "loss": 0.291,
+      "step": 21471
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675162073716888e-05,
+      "loss": 0.2488,
+      "step": 21472
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675120538988025e-05,
+      "loss": 0.2838,
+      "step": 21473
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675079001647804e-05,
+      "loss": 0.2544,
+      "step": 21474
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9675037461696235e-05,
+      "loss": 0.3364,
+      "step": 21475
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674995919133327e-05,
+      "loss": 0.2683,
+      "step": 21476
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674954373959095e-05,
+      "loss": 0.3137,
+      "step": 21477
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674912826173544e-05,
+      "loss": 0.3018,
+      "step": 21478
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674871275776693e-05,
+      "loss": 0.3175,
+      "step": 21479
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674829722768548e-05,
+      "loss": 0.3067,
+      "step": 21480
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967478816714912e-05,
+      "loss": 0.338,
+      "step": 21481
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967474660891842e-05,
+      "loss": 0.3116,
+      "step": 21482
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674705048076462e-05,
+      "loss": 0.3127,
+      "step": 21483
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674663484623255e-05,
+      "loss": 0.2907,
+      "step": 21484
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674621918558812e-05,
+      "loss": 0.3051,
+      "step": 21485
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674580349883142e-05,
+      "loss": 0.2676,
+      "step": 21486
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967453877859626e-05,
+      "loss": 0.2959,
+      "step": 21487
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674497204698174e-05,
+      "loss": 0.345,
+      "step": 21488
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674455628188897e-05,
+      "loss": 0.3117,
+      "step": 21489
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674414049068437e-05,
+      "loss": 0.2636,
+      "step": 21490
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674372467336808e-05,
+      "loss": 0.2846,
+      "step": 21491
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967433088299402e-05,
+      "loss": 0.2849,
+      "step": 21492
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674289296040084e-05,
+      "loss": 0.2989,
+      "step": 21493
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674247706475017e-05,
+      "loss": 0.2787,
+      "step": 21494
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967420611429882e-05,
+      "loss": 0.2748,
+      "step": 21495
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674164519511514e-05,
+      "loss": 0.3081,
+      "step": 21496
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9674122922113103e-05,
+      "loss": 0.3208,
+      "step": 21497
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.96740813221036e-05,
+      "loss": 0.286,
+      "step": 21498
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967403971948302e-05,
+      "loss": 0.3036,
+      "step": 21499
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673998114251368e-05,
+      "loss": 0.2979,
+      "step": 21500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967395650640866e-05,
+      "loss": 0.3021,
+      "step": 21501
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673914895954907e-05,
+      "loss": 0.262,
+      "step": 21502
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967387328289012e-05,
+      "loss": 0.3356,
+      "step": 21503
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967383166721431e-05,
+      "loss": 0.3292,
+      "step": 21504
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673790048927483e-05,
+      "loss": 0.3334,
+      "step": 21505
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967374842802966e-05,
+      "loss": 0.2489,
+      "step": 21506
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673706804520846e-05,
+      "loss": 0.2554,
+      "step": 21507
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673665178401048e-05,
+      "loss": 0.3109,
+      "step": 21508
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673623549670288e-05,
+      "loss": 0.307,
+      "step": 21509
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673581918328573e-05,
+      "loss": 0.3849,
+      "step": 21510
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967354028437591e-05,
+      "loss": 0.2857,
+      "step": 21511
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967349864781231e-05,
+      "loss": 0.3285,
+      "step": 21512
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673457008637794e-05,
+      "loss": 0.2526,
+      "step": 21513
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673415366852363e-05,
+      "loss": 0.2989,
+      "step": 21514
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673373722456034e-05,
+      "loss": 0.3284,
+      "step": 21515
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673332075448814e-05,
+      "loss": 0.3312,
+      "step": 21516
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673290425830716e-05,
+      "loss": 0.3041,
+      "step": 21517
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673248773601755e-05,
+      "loss": 0.3084,
+      "step": 21518
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967320711876194e-05,
+      "loss": 0.3342,
+      "step": 21519
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673165461311278e-05,
+      "loss": 0.304,
+      "step": 21520
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673123801249782e-05,
+      "loss": 0.3221,
+      "step": 21521
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673082138577467e-05,
+      "loss": 0.2762,
+      "step": 21522
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9673040473294342e-05,
+      "loss": 0.3441,
+      "step": 21523
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9672998805400415e-05,
+      "loss": 0.2814,
+      "step": 21524
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9672957134895708e-05,
+      "loss": 0.2573,
+      "step": 21525
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967291546178022e-05,
+      "loss": 0.2617,
+      "step": 21526
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9672873786053963e-05,
+      "loss": 0.2772,
+      "step": 21527
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9672832107716956e-05,
+      "loss": 0.3061,
+      "step": 21528
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9672790426769207e-05,
+      "loss": 0.3201,
+      "step": 21529
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9672748743210727e-05,
+      "loss": 0.3124,
+      "step": 21530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9672707057041524e-05,
+      "loss": 0.329,
+      "step": 21531
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9672665368261614e-05,
+      "loss": 0.2992,
+      "step": 21532
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967262367687101e-05,
+      "loss": 0.2983,
+      "step": 21533
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967258198286971e-05,
+      "loss": 0.349,
+      "step": 21534
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9672540286257744e-05,
+      "loss": 0.2881,
+      "step": 21535
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.967249858703511e-05,
+      "loss": 0.3315,
+      "step": 21536
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9672456885201825e-05,
+      "loss": 0.3124,
+      "step": 21537
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9672415180757894e-05,
+      "loss": 0.2539,
+      "step": 21538
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967237347370334e-05,
+      "loss": 0.3336,
+      "step": 21539
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9672331764038165e-05,
+      "loss": 0.2439,
+      "step": 21540
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967229005176238e-05,
+      "loss": 0.307,
+      "step": 21541
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9672248336876e-05,
+      "loss": 0.3097,
+      "step": 21542
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9672206619379034e-05,
+      "loss": 0.26,
+      "step": 21543
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9672164899271498e-05,
+      "loss": 0.298,
+      "step": 21544
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9672123176553397e-05,
+      "loss": 0.3356,
+      "step": 21545
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9672081451224746e-05,
+      "loss": 0.2765,
+      "step": 21546
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967203972328555e-05,
+      "loss": 0.2608,
+      "step": 21547
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967199799273583e-05,
+      "loss": 0.3019,
+      "step": 21548
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967195625957559e-05,
+      "loss": 0.295,
+      "step": 21549
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671914523804848e-05,
+      "loss": 0.2657,
+      "step": 21550
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967187278542361e-05,
+      "loss": 0.3004,
+      "step": 21551
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671831044431884e-05,
+      "loss": 0.2684,
+      "step": 21552
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671789300829687e-05,
+      "loss": 0.2652,
+      "step": 21553
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967174755461703e-05,
+      "loss": 0.3298,
+      "step": 21554
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671705805793926e-05,
+      "loss": 0.3046,
+      "step": 21555
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967166405436038e-05,
+      "loss": 0.2704,
+      "step": 21556
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671622300316406e-05,
+      "loss": 0.3279,
+      "step": 21557
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671580543662017e-05,
+      "loss": 0.2838,
+      "step": 21558
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671538784397226e-05,
+      "loss": 0.2971,
+      "step": 21559
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671497022522038e-05,
+      "loss": 0.335,
+      "step": 21560
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671455258036468e-05,
+      "loss": 0.3326,
+      "step": 21561
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671413490940528e-05,
+      "loss": 0.2944,
+      "step": 21562
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967137172123423e-05,
+      "loss": 0.3147,
+      "step": 21563
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967132994891758e-05,
+      "loss": 0.2688,
+      "step": 21564
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671288173990596e-05,
+      "loss": 0.2548,
+      "step": 21565
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671246396453285e-05,
+      "loss": 0.3281,
+      "step": 21566
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671204616305658e-05,
+      "loss": 0.2526,
+      "step": 21567
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671162833547727e-05,
+      "loss": 0.29,
+      "step": 21568
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671121048179508e-05,
+      "loss": 0.3505,
+      "step": 21569
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9671079260201e-05,
+      "loss": 0.3053,
+      "step": 21570
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967103746961223e-05,
+      "loss": 0.2593,
+      "step": 21571
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.96709956764132e-05,
+      "loss": 0.3684,
+      "step": 21572
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670953880603926e-05,
+      "loss": 0.306,
+      "step": 21573
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670912082184413e-05,
+      "loss": 0.3039,
+      "step": 21574
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670870281154673e-05,
+      "loss": 0.3125,
+      "step": 21575
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670828477514724e-05,
+      "loss": 0.2807,
+      "step": 21576
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967078667126457e-05,
+      "loss": 0.2507,
+      "step": 21577
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967074486240423e-05,
+      "loss": 0.2855,
+      "step": 21578
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670703050933707e-05,
+      "loss": 0.3179,
+      "step": 21579
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967066123685302e-05,
+      "loss": 0.274,
+      "step": 21580
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670619420162172e-05,
+      "loss": 0.2989,
+      "step": 21581
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670577600861183e-05,
+      "loss": 0.3002,
+      "step": 21582
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670535778950055e-05,
+      "loss": 0.304,
+      "step": 21583
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670493954428805e-05,
+      "loss": 0.3001,
+      "step": 21584
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670452127297447e-05,
+      "loss": 0.3061,
+      "step": 21585
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670410297555987e-05,
+      "loss": 0.2692,
+      "step": 21586
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670368465204437e-05,
+      "loss": 0.3176,
+      "step": 21587
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967032663024281e-05,
+      "loss": 0.2734,
+      "step": 21588
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670284792671118e-05,
+      "loss": 0.2677,
+      "step": 21589
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670242952489367e-05,
+      "loss": 0.3075,
+      "step": 21590
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670201109697576e-05,
+      "loss": 0.2714,
+      "step": 21591
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967015926429575e-05,
+      "loss": 0.3015,
+      "step": 21592
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670117416283906e-05,
+      "loss": 0.3047,
+      "step": 21593
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.967007556566205e-05,
+      "loss": 0.287,
+      "step": 21594
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9670033712430196e-05,
+      "loss": 0.2407,
+      "step": 21595
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669991856588355e-05,
+      "loss": 0.3053,
+      "step": 21596
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669949998136534e-05,
+      "loss": 0.3318,
+      "step": 21597
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669908137074754e-05,
+      "loss": 0.3077,
+      "step": 21598
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966986627340302e-05,
+      "loss": 0.2863,
+      "step": 21599
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966982440712134e-05,
+      "loss": 0.2343,
+      "step": 21600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669782538229732e-05,
+      "loss": 0.2612,
+      "step": 21601
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669740666728203e-05,
+      "loss": 0.3097,
+      "step": 21602
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669698792616766e-05,
+      "loss": 0.2857,
+      "step": 21603
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669656915895434e-05,
+      "loss": 0.2767,
+      "step": 21604
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669615036564217e-05,
+      "loss": 0.2804,
+      "step": 21605
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669573154623122e-05,
+      "loss": 0.2824,
+      "step": 21606
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669531270072166e-05,
+      "loss": 0.2844,
+      "step": 21607
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669489382911356e-05,
+      "loss": 0.2751,
+      "step": 21608
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966944749314071e-05,
+      "loss": 0.2859,
+      "step": 21609
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966940560076023e-05,
+      "loss": 0.3075,
+      "step": 21610
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669363705769936e-05,
+      "loss": 0.3302,
+      "step": 21611
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669321808169838e-05,
+      "loss": 0.2894,
+      "step": 21612
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966927990795994e-05,
+      "loss": 0.318,
+      "step": 21613
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966923800514026e-05,
+      "loss": 0.3291,
+      "step": 21614
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669196099710805e-05,
+      "loss": 0.2863,
+      "step": 21615
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966915419167159e-05,
+      "loss": 0.3029,
+      "step": 21616
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669112281022625e-05,
+      "loss": 0.2779,
+      "step": 21617
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669070367763924e-05,
+      "loss": 0.2861,
+      "step": 21618
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9669028451895493e-05,
+      "loss": 0.364,
+      "step": 21619
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9668986533417348e-05,
+      "loss": 0.3179,
+      "step": 21620
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9668944612329497e-05,
+      "loss": 0.2405,
+      "step": 21621
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966890268863195e-05,
+      "loss": 0.2771,
+      "step": 21622
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9668860762324727e-05,
+      "loss": 0.3378,
+      "step": 21623
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966881883340783e-05,
+      "loss": 0.2839,
+      "step": 21624
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9668776901881273e-05,
+      "loss": 0.301,
+      "step": 21625
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966873496774507e-05,
+      "loss": 0.2753,
+      "step": 21626
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9668693030999225e-05,
+      "loss": 0.2809,
+      "step": 21627
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9668651091643764e-05,
+      "loss": 0.2885,
+      "step": 21628
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966860914967868e-05,
+      "loss": 0.2962,
+      "step": 21629
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9668567205103997e-05,
+      "loss": 0.288,
+      "step": 21630
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966852525791972e-05,
+      "loss": 0.308,
+      "step": 21631
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9668483308125866e-05,
+      "loss": 0.272,
+      "step": 21632
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9668441355722442e-05,
+      "loss": 0.2738,
+      "step": 21633
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966839940070946e-05,
+      "loss": 0.315,
+      "step": 21634
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9668357443086935e-05,
+      "loss": 0.2595,
+      "step": 21635
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.966831548285487e-05,
+      "loss": 0.2584,
+      "step": 21636
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9668273520013285e-05,
+      "loss": 0.3236,
+      "step": 21637
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9668231554562185e-05,
+      "loss": 0.3346,
+      "step": 21638
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9668189586501587e-05,
+      "loss": 0.2846,
+      "step": 21639
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.96681476158315e-05,
+      "loss": 0.2971,
+      "step": 21640
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9668105642551932e-05,
+      "loss": 0.2712,
+      "step": 21641
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9668063666662896e-05,
+      "loss": 0.3015,
+      "step": 21642
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966802168816441e-05,
+      "loss": 0.2654,
+      "step": 21643
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966797970705647e-05,
+      "loss": 0.3412,
+      "step": 21644
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667937723339106e-05,
+      "loss": 0.2715,
+      "step": 21645
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966789573701232e-05,
+      "loss": 0.314,
+      "step": 21646
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966785374807612e-05,
+      "loss": 0.3294,
+      "step": 21647
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667811756530525e-05,
+      "loss": 0.3543,
+      "step": 21648
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667769762375538e-05,
+      "loss": 0.2902,
+      "step": 21649
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966772776561118e-05,
+      "loss": 0.2926,
+      "step": 21650
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966768576623745e-05,
+      "loss": 0.3043,
+      "step": 21651
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966764376425437e-05,
+      "loss": 0.2806,
+      "step": 21652
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966760175966195e-05,
+      "loss": 0.2574,
+      "step": 21653
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667559752460196e-05,
+      "loss": 0.2962,
+      "step": 21654
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667517742649124e-05,
+      "loss": 0.2654,
+      "step": 21655
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667475730228743e-05,
+      "loss": 0.2819,
+      "step": 21656
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667433715199067e-05,
+      "loss": 0.3108,
+      "step": 21657
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667391697560102e-05,
+      "loss": 0.2736,
+      "step": 21658
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667349677311866e-05,
+      "loss": 0.3226,
+      "step": 21659
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667307654454365e-05,
+      "loss": 0.2983,
+      "step": 21660
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667265628987613e-05,
+      "loss": 0.2987,
+      "step": 21661
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667223600911624e-05,
+      "loss": 0.2702,
+      "step": 21662
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.96671815702264e-05,
+      "loss": 0.3078,
+      "step": 21663
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966713953693196e-05,
+      "loss": 0.2882,
+      "step": 21664
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667097501028318e-05,
+      "loss": 0.3094,
+      "step": 21665
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966705546251548e-05,
+      "loss": 0.317,
+      "step": 21666
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9667013421393455e-05,
+      "loss": 0.3287,
+      "step": 21667
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966697137766226e-05,
+      "loss": 0.2643,
+      "step": 21668
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666929331321903e-05,
+      "loss": 0.2799,
+      "step": 21669
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.96668872823724e-05,
+      "loss": 0.2808,
+      "step": 21670
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666845230813757e-05,
+      "loss": 0.2846,
+      "step": 21671
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666803176645988e-05,
+      "loss": 0.2914,
+      "step": 21672
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666761119869104e-05,
+      "loss": 0.3103,
+      "step": 21673
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666719060483114e-05,
+      "loss": 0.3509,
+      "step": 21674
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666676998488032e-05,
+      "loss": 0.2897,
+      "step": 21675
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966663493388387e-05,
+      "loss": 0.3549,
+      "step": 21676
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666592866670634e-05,
+      "loss": 0.3097,
+      "step": 21677
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966655079684834e-05,
+      "loss": 0.3367,
+      "step": 21678
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666508724417004e-05,
+      "loss": 0.3413,
+      "step": 21679
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966646664937663e-05,
+      "loss": 0.3114,
+      "step": 21680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666424571727233e-05,
+      "loss": 0.2903,
+      "step": 21681
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666382491468817e-05,
+      "loss": 0.3213,
+      "step": 21682
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666340408601404e-05,
+      "loss": 0.3357,
+      "step": 21683
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666298323125e-05,
+      "loss": 0.3629,
+      "step": 21684
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666256235039616e-05,
+      "loss": 0.3008,
+      "step": 21685
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666214144345263e-05,
+      "loss": 0.3109,
+      "step": 21686
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666172051041956e-05,
+      "loss": 0.3021,
+      "step": 21687
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666129955129704e-05,
+      "loss": 0.3356,
+      "step": 21688
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666087856608515e-05,
+      "loss": 0.2746,
+      "step": 21689
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666045755478407e-05,
+      "loss": 0.3208,
+      "step": 21690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9666003651739387e-05,
+      "loss": 0.2379,
+      "step": 21691
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665961545391468e-05,
+      "loss": 0.2989,
+      "step": 21692
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966591943643466e-05,
+      "loss": 0.3557,
+      "step": 21693
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665877324868975e-05,
+      "loss": 0.2795,
+      "step": 21694
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665835210694424e-05,
+      "loss": 0.2664,
+      "step": 21695
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665793093911022e-05,
+      "loss": 0.278,
+      "step": 21696
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665750974518776e-05,
+      "loss": 0.3067,
+      "step": 21697
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665708852517698e-05,
+      "loss": 0.2865,
+      "step": 21698
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.96656667279078e-05,
+      "loss": 0.2719,
+      "step": 21699
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966562460068909e-05,
+      "loss": 0.3274,
+      "step": 21700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665582470861588e-05,
+      "loss": 0.3409,
+      "step": 21701
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665540338425298e-05,
+      "loss": 0.2683,
+      "step": 21702
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665498203380232e-05,
+      "loss": 0.3211,
+      "step": 21703
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665456065726405e-05,
+      "loss": 0.2584,
+      "step": 21704
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665413925463826e-05,
+      "loss": 0.37,
+      "step": 21705
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665371782592507e-05,
+      "loss": 0.2758,
+      "step": 21706
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665329637112456e-05,
+      "loss": 0.237,
+      "step": 21707
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966528748902369e-05,
+      "loss": 0.2842,
+      "step": 21708
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966524533832622e-05,
+      "loss": 0.3029,
+      "step": 21709
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966520318502005e-05,
+      "loss": 0.2695,
+      "step": 21710
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665161029105197e-05,
+      "loss": 0.3062,
+      "step": 21711
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9665118870581678e-05,
+      "loss": 0.3219,
+      "step": 21712
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966507670944949e-05,
+      "loss": 0.2995,
+      "step": 21713
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966503454570866e-05,
+      "loss": 0.3142,
+      "step": 21714
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664992379359186e-05,
+      "loss": 0.2856,
+      "step": 21715
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966495021040109e-05,
+      "loss": 0.2528,
+      "step": 21716
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664908038834378e-05,
+      "loss": 0.338,
+      "step": 21717
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966486586465906e-05,
+      "loss": 0.3143,
+      "step": 21718
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664823687875146e-05,
+      "loss": 0.2824,
+      "step": 21719
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664781508482657e-05,
+      "loss": 0.3253,
+      "step": 21720
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664739326481597e-05,
+      "loss": 0.2681,
+      "step": 21721
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664697141871978e-05,
+      "loss": 0.2735,
+      "step": 21722
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664654954653812e-05,
+      "loss": 0.2666,
+      "step": 21723
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.966461276482711e-05,
+      "loss": 0.294,
+      "step": 21724
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664570572391885e-05,
+      "loss": 0.3586,
+      "step": 21725
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664528377348148e-05,
+      "loss": 0.2767,
+      "step": 21726
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664486179695908e-05,
+      "loss": 0.2957,
+      "step": 21727
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664443979435176e-05,
+      "loss": 0.2782,
+      "step": 21728
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664401776565973e-05,
+      "loss": 0.306,
+      "step": 21729
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664359571088294e-05,
+      "loss": 0.357,
+      "step": 21730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664317363002164e-05,
+      "loss": 0.3499,
+      "step": 21731
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664275152307585e-05,
+      "loss": 0.2852,
+      "step": 21732
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664232939004576e-05,
+      "loss": 0.3016,
+      "step": 21733
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9664190723093148e-05,
+      "loss": 0.2706,
+      "step": 21734
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9664148504573307e-05,
+      "loss": 0.3498,
+      "step": 21735
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9664106283445068e-05,
+      "loss": 0.334,
+      "step": 21736
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9664064059708442e-05,
+      "loss": 0.2995,
+      "step": 21737
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966402183336344e-05,
+      "loss": 0.2964,
+      "step": 21738
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966397960441007e-05,
+      "loss": 0.2898,
+      "step": 21739
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966393737284835e-05,
+      "loss": 0.2643,
+      "step": 21740
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663895138678287e-05,
+      "loss": 0.3167,
+      "step": 21741
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663852901899895e-05,
+      "loss": 0.3259,
+      "step": 21742
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966381066251318e-05,
+      "loss": 0.3252,
+      "step": 21743
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663768420518162e-05,
+      "loss": 0.3473,
+      "step": 21744
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663726175914846e-05,
+      "loss": 0.2977,
+      "step": 21745
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663683928703244e-05,
+      "loss": 0.3001,
+      "step": 21746
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663641678883366e-05,
+      "loss": 0.3172,
+      "step": 21747
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663599426455234e-05,
+      "loss": 0.3334,
+      "step": 21748
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663557171418843e-05,
+      "loss": 0.2859,
+      "step": 21749
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663514913774218e-05,
+      "loss": 0.2977,
+      "step": 21750
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966347265352136e-05,
+      "loss": 0.2892,
+      "step": 21751
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663430390660294e-05,
+      "loss": 0.2861,
+      "step": 21752
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663388125191015e-05,
+      "loss": 0.2645,
+      "step": 21753
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663345857113546e-05,
+      "loss": 0.3062,
+      "step": 21754
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663303586427896e-05,
+      "loss": 0.314,
+      "step": 21755
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663261313134076e-05,
+      "loss": 0.3281,
+      "step": 21756
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966321903723209e-05,
+      "loss": 0.2689,
+      "step": 21757
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663176758721962e-05,
+      "loss": 0.2681,
+      "step": 21758
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663134477603693e-05,
+      "loss": 0.2823,
+      "step": 21759
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663092193877303e-05,
+      "loss": 0.2585,
+      "step": 21760
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9663049907542798e-05,
+      "loss": 0.2731,
+      "step": 21761
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966300761860019e-05,
+      "loss": 0.317,
+      "step": 21762
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966296532704949e-05,
+      "loss": 0.2608,
+      "step": 21763
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662923032890714e-05,
+      "loss": 0.3259,
+      "step": 21764
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662880736123866e-05,
+      "loss": 0.3391,
+      "step": 21765
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662838436748963e-05,
+      "loss": 0.2804,
+      "step": 21766
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662796134766015e-05,
+      "loss": 0.3392,
+      "step": 21767
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662753830175032e-05,
+      "loss": 0.2556,
+      "step": 21768
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966271152297603e-05,
+      "loss": 0.268,
+      "step": 21769
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662669213169013e-05,
+      "loss": 0.2821,
+      "step": 21770
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662626900753998e-05,
+      "loss": 0.2944,
+      "step": 21771
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662584585730994e-05,
+      "loss": 0.2694,
+      "step": 21772
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662542268100014e-05,
+      "loss": 0.2928,
+      "step": 21773
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966249994786107e-05,
+      "loss": 0.3056,
+      "step": 21774
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966245762501417e-05,
+      "loss": 0.3062,
+      "step": 21775
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966241529955933e-05,
+      "loss": 0.2692,
+      "step": 21776
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662372971496555e-05,
+      "loss": 0.2648,
+      "step": 21777
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966233064082586e-05,
+      "loss": 0.2901,
+      "step": 21778
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662288307547262e-05,
+      "loss": 0.2738,
+      "step": 21779
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662245971660763e-05,
+      "loss": 0.2744,
+      "step": 21780
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662203633166382e-05,
+      "loss": 0.3036,
+      "step": 21781
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662161292064126e-05,
+      "loss": 0.3251,
+      "step": 21782
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662118948354004e-05,
+      "loss": 0.2549,
+      "step": 21783
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662076602036033e-05,
+      "loss": 0.2837,
+      "step": 21784
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9662034253110225e-05,
+      "loss": 0.3,
+      "step": 21785
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661991901576588e-05,
+      "loss": 0.2569,
+      "step": 21786
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966194954743513e-05,
+      "loss": 0.3249,
+      "step": 21787
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966190719068587e-05,
+      "loss": 0.2848,
+      "step": 21788
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661864831328818e-05,
+      "loss": 0.2974,
+      "step": 21789
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966182246936398e-05,
+      "loss": 0.3391,
+      "step": 21790
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661780104791373e-05,
+      "loss": 0.3549,
+      "step": 21791
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661737737611005e-05,
+      "loss": 0.2829,
+      "step": 21792
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966169536782289e-05,
+      "loss": 0.2825,
+      "step": 21793
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966165299542704e-05,
+      "loss": 0.327,
+      "step": 21794
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966161062042346e-05,
+      "loss": 0.3032,
+      "step": 21795
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966156824281217e-05,
+      "loss": 0.2873,
+      "step": 21796
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661525862593173e-05,
+      "loss": 0.3155,
+      "step": 21797
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966148347976649e-05,
+      "loss": 0.3183,
+      "step": 21798
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661441094332125e-05,
+      "loss": 0.2956,
+      "step": 21799
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661398706290094e-05,
+      "loss": 0.3035,
+      "step": 21800
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661356315640405e-05,
+      "loss": 0.2726,
+      "step": 21801
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966131392238307e-05,
+      "loss": 0.2753,
+      "step": 21802
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.96612715265181e-05,
+      "loss": 0.2807,
+      "step": 21803
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966122912804551e-05,
+      "loss": 0.3102,
+      "step": 21804
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661186726965308e-05,
+      "loss": 0.2782,
+      "step": 21805
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661144323277508e-05,
+      "loss": 0.284,
+      "step": 21806
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966110191698212e-05,
+      "loss": 0.2716,
+      "step": 21807
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661059508079156e-05,
+      "loss": 0.2924,
+      "step": 21808
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9661017096568624e-05,
+      "loss": 0.3066,
+      "step": 21809
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660974682450538e-05,
+      "loss": 0.3119,
+      "step": 21810
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660932265724913e-05,
+      "loss": 0.2833,
+      "step": 21811
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660889846391757e-05,
+      "loss": 0.3649,
+      "step": 21812
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660847424451078e-05,
+      "loss": 0.2771,
+      "step": 21813
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660804999902893e-05,
+      "loss": 0.2713,
+      "step": 21814
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660762572747215e-05,
+      "loss": 0.336,
+      "step": 21815
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660720142984048e-05,
+      "loss": 0.2617,
+      "step": 21816
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966067771061341e-05,
+      "loss": 0.302,
+      "step": 21817
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660635275635304e-05,
+      "loss": 0.4066,
+      "step": 21818
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660592838049754e-05,
+      "loss": 0.3386,
+      "step": 21819
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660550397856762e-05,
+      "loss": 0.295,
+      "step": 21820
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660507955056345e-05,
+      "loss": 0.3371,
+      "step": 21821
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660465509648506e-05,
+      "loss": 0.3145,
+      "step": 21822
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660423061633267e-05,
+      "loss": 0.2847,
+      "step": 21823
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660380611010633e-05,
+      "loss": 0.2461,
+      "step": 21824
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966033815778062e-05,
+      "loss": 0.2905,
+      "step": 21825
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660295701943232e-05,
+      "loss": 0.3266,
+      "step": 21826
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966025324349849e-05,
+      "loss": 0.2636,
+      "step": 21827
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.96602107824464e-05,
+      "loss": 0.2923,
+      "step": 21828
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966016831878697e-05,
+      "loss": 0.3107,
+      "step": 21829
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9660125852520217e-05,
+      "loss": 0.2487,
+      "step": 21830
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966008338364615e-05,
+      "loss": 0.3065,
+      "step": 21831
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.966004091216478e-05,
+      "loss": 0.2973,
+      "step": 21832
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9659998438076126e-05,
+      "loss": 0.2965,
+      "step": 21833
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659955961380188e-05,
+      "loss": 0.2768,
+      "step": 21834
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659913482076984e-05,
+      "loss": 0.3019,
+      "step": 21835
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659871000166525e-05,
+      "loss": 0.3532,
+      "step": 21836
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965982851564882e-05,
+      "loss": 0.2954,
+      "step": 21837
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659786028523885e-05,
+      "loss": 0.3189,
+      "step": 21838
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659743538791724e-05,
+      "loss": 0.2754,
+      "step": 21839
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659701046452358e-05,
+      "loss": 0.2787,
+      "step": 21840
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965965855150579e-05,
+      "loss": 0.321,
+      "step": 21841
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659616053952036e-05,
+      "loss": 0.3146,
+      "step": 21842
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659573553791107e-05,
+      "loss": 0.2637,
+      "step": 21843
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659531051023017e-05,
+      "loss": 0.313,
+      "step": 21844
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965948854564777e-05,
+      "loss": 0.3213,
+      "step": 21845
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659446037665383e-05,
+      "loss": 0.2822,
+      "step": 21846
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659403527075865e-05,
+      "loss": 0.3226,
+      "step": 21847
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965936101387923e-05,
+      "loss": 0.2655,
+      "step": 21848
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659318498075488e-05,
+      "loss": 0.3099,
+      "step": 21849
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965927597966465e-05,
+      "loss": 0.2932,
+      "step": 21850
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965923345864673e-05,
+      "loss": 0.2723,
+      "step": 21851
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965919093502174e-05,
+      "loss": 0.2903,
+      "step": 21852
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659148408789685e-05,
+      "loss": 0.2929,
+      "step": 21853
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659105879950582e-05,
+      "loss": 0.3081,
+      "step": 21854
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659063348504444e-05,
+      "loss": 0.2919,
+      "step": 21855
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9659020814451277e-05,
+      "loss": 0.2759,
+      "step": 21856
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658978277791095e-05,
+      "loss": 0.3015,
+      "step": 21857
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658935738523908e-05,
+      "loss": 0.2918,
+      "step": 21858
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965889319664973e-05,
+      "loss": 0.3264,
+      "step": 21859
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658850652168575e-05,
+      "loss": 0.3239,
+      "step": 21860
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965880810508045e-05,
+      "loss": 0.3353,
+      "step": 21861
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658765555385365e-05,
+      "loss": 0.3211,
+      "step": 21862
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658723003083334e-05,
+      "loss": 0.2875,
+      "step": 21863
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965868044817437e-05,
+      "loss": 0.2959,
+      "step": 21864
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965863789065848e-05,
+      "loss": 0.2676,
+      "step": 21865
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658595330535683e-05,
+      "loss": 0.2813,
+      "step": 21866
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658552767805985e-05,
+      "loss": 0.2919,
+      "step": 21867
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658510202469398e-05,
+      "loss": 0.3063,
+      "step": 21868
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965846763452593e-05,
+      "loss": 0.3085,
+      "step": 21869
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658425063975602e-05,
+      "loss": 0.2872,
+      "step": 21870
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658382490818417e-05,
+      "loss": 0.272,
+      "step": 21871
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658339915054393e-05,
+      "loss": 0.2818,
+      "step": 21872
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658297336683534e-05,
+      "loss": 0.3029,
+      "step": 21873
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658254755705857e-05,
+      "loss": 0.289,
+      "step": 21874
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965821217212137e-05,
+      "loss": 0.2757,
+      "step": 21875
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965816958593009e-05,
+      "loss": 0.2888,
+      "step": 21876
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9658126997132022e-05,
+      "loss": 0.2785,
+      "step": 21877
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965808440572718e-05,
+      "loss": 0.3403,
+      "step": 21878
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965804181171558e-05,
+      "loss": 0.2885,
+      "step": 21879
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657999215097225e-05,
+      "loss": 0.2989,
+      "step": 21880
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965795661587213e-05,
+      "loss": 0.3133,
+      "step": 21881
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965791401404031e-05,
+      "loss": 0.2769,
+      "step": 21882
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657871409601773e-05,
+      "loss": 0.2749,
+      "step": 21883
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657828802556535e-05,
+      "loss": 0.2898,
+      "step": 21884
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657786192904596e-05,
+      "loss": 0.2644,
+      "step": 21885
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657743580645982e-05,
+      "loss": 0.3258,
+      "step": 21886
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657700965780698e-05,
+      "loss": 0.2742,
+      "step": 21887
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657658348308752e-05,
+      "loss": 0.3274,
+      "step": 21888
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965761572823016e-05,
+      "loss": 0.2659,
+      "step": 21889
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657573105544933e-05,
+      "loss": 0.3079,
+      "step": 21890
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657530480253083e-05,
+      "loss": 0.3177,
+      "step": 21891
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965748785235462e-05,
+      "loss": 0.302,
+      "step": 21892
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657445221849553e-05,
+      "loss": 0.2828,
+      "step": 21893
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.96574025887379e-05,
+      "loss": 0.2672,
+      "step": 21894
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657359953019663e-05,
+      "loss": 0.323,
+      "step": 21895
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657317314694863e-05,
+      "loss": 0.3194,
+      "step": 21896
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965727467376351e-05,
+      "loss": 0.3423,
+      "step": 21897
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965723203022561e-05,
+      "loss": 0.3323,
+      "step": 21898
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965718938408118e-05,
+      "loss": 0.3522,
+      "step": 21899
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965714673533023e-05,
+      "loss": 0.2467,
+      "step": 21900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657104083972772e-05,
+      "loss": 0.2705,
+      "step": 21901
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9657061430008814e-05,
+      "loss": 0.3085,
+      "step": 21902
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965701877343837e-05,
+      "loss": 0.2729,
+      "step": 21903
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965697611426145e-05,
+      "loss": 0.3043,
+      "step": 21904
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965693345247807e-05,
+      "loss": 0.2823,
+      "step": 21905
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965689078808824e-05,
+      "loss": 0.2664,
+      "step": 21906
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656848121091965e-05,
+      "loss": 0.3135,
+      "step": 21907
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656805451489264e-05,
+      "loss": 0.299,
+      "step": 21908
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656762779280148e-05,
+      "loss": 0.2664,
+      "step": 21909
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656720104464623e-05,
+      "loss": 0.3043,
+      "step": 21910
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656677427042704e-05,
+      "loss": 0.2982,
+      "step": 21911
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656634747014406e-05,
+      "loss": 0.2848,
+      "step": 21912
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656592064379738e-05,
+      "loss": 0.2656,
+      "step": 21913
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656549379138704e-05,
+      "loss": 0.2798,
+      "step": 21914
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656506691291327e-05,
+      "loss": 0.2945,
+      "step": 21915
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656464000837613e-05,
+      "loss": 0.2507,
+      "step": 21916
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656421307777574e-05,
+      "loss": 0.3073,
+      "step": 21917
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656378612111226e-05,
+      "loss": 0.2864,
+      "step": 21918
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965633591383857e-05,
+      "loss": 0.364,
+      "step": 21919
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965629321295963e-05,
+      "loss": 0.2711,
+      "step": 21920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656250509474404e-05,
+      "loss": 0.2875,
+      "step": 21921
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656207803382913e-05,
+      "loss": 0.3548,
+      "step": 21922
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965616509468517e-05,
+      "loss": 0.3153,
+      "step": 21923
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965612238338118e-05,
+      "loss": 0.3211,
+      "step": 21924
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656079669470958e-05,
+      "loss": 0.2657,
+      "step": 21925
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9656036952954514e-05,
+      "loss": 0.2894,
+      "step": 21926
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.965599423383186e-05,
+      "loss": 0.3638,
+      "step": 21927
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9655951512103013e-05,
+      "loss": 0.3208,
+      "step": 21928
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9655908787767974e-05,
+      "loss": 0.2693,
+      "step": 21929
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9655866060826764e-05,
+      "loss": 0.3123,
+      "step": 21930
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9655823331279385e-05,
+      "loss": 0.3298,
+      "step": 21931
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965578059912586e-05,
+      "loss": 0.3115,
+      "step": 21932
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655737864366193e-05,
+      "loss": 0.2775,
+      "step": 21933
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.96556951270004e-05,
+      "loss": 0.3357,
+      "step": 21934
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655652387028484e-05,
+      "loss": 0.3557,
+      "step": 21935
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655609644450466e-05,
+      "loss": 0.2975,
+      "step": 21936
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965556689926635e-05,
+      "loss": 0.3172,
+      "step": 21937
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655524151476156e-05,
+      "loss": 0.3128,
+      "step": 21938
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965548140107989e-05,
+      "loss": 0.3079,
+      "step": 21939
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965543864807756e-05,
+      "loss": 0.3499,
+      "step": 21940
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655395892469188e-05,
+      "loss": 0.3235,
+      "step": 21941
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655353134254776e-05,
+      "loss": 0.3033,
+      "step": 21942
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965531037343434e-05,
+      "loss": 0.3137,
+      "step": 21943
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655267610007888e-05,
+      "loss": 0.3141,
+      "step": 21944
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965522484397544e-05,
+      "loss": 0.3155,
+      "step": 21945
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655182075336996e-05,
+      "loss": 0.2627,
+      "step": 21946
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655139304092578e-05,
+      "loss": 0.3016,
+      "step": 21947
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655096530242188e-05,
+      "loss": 0.302,
+      "step": 21948
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9655053753785847e-05,
+      "loss": 0.3265,
+      "step": 21949
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965501097472356e-05,
+      "loss": 0.303,
+      "step": 21950
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654968193055336e-05,
+      "loss": 0.2637,
+      "step": 21951
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654925408781196e-05,
+      "loss": 0.3088,
+      "step": 21952
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965488262190115e-05,
+      "loss": 0.3394,
+      "step": 21953
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.96548398324152e-05,
+      "loss": 0.2882,
+      "step": 21954
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654797040323365e-05,
+      "loss": 0.3406,
+      "step": 21955
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654754245625654e-05,
+      "loss": 0.2966,
+      "step": 21956
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965471144832208e-05,
+      "loss": 0.3085,
+      "step": 21957
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654668648412655e-05,
+      "loss": 0.333,
+      "step": 21958
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965462584589739e-05,
+      "loss": 0.2968,
+      "step": 21959
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654583040776295e-05,
+      "loss": 0.2889,
+      "step": 21960
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654540233049385e-05,
+      "loss": 0.2825,
+      "step": 21961
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965449742271667e-05,
+      "loss": 0.3212,
+      "step": 21962
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965445460977816e-05,
+      "loss": 0.273,
+      "step": 21963
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654411794233866e-05,
+      "loss": 0.3281,
+      "step": 21964
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654368976083802e-05,
+      "loss": 0.3207,
+      "step": 21965
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965432615532798e-05,
+      "loss": 0.3556,
+      "step": 21966
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654283331966408e-05,
+      "loss": 0.3204,
+      "step": 21967
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654240505999104e-05,
+      "loss": 0.2952,
+      "step": 21968
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965419767742607e-05,
+      "loss": 0.2736,
+      "step": 21969
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654154846247327e-05,
+      "loss": 0.2735,
+      "step": 21970
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965411201246288e-05,
+      "loss": 0.3047,
+      "step": 21971
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654069176072743e-05,
+      "loss": 0.2281,
+      "step": 21972
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9654026337076927e-05,
+      "loss": 0.3077,
+      "step": 21973
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653983495475448e-05,
+      "loss": 0.3534,
+      "step": 21974
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965394065126831e-05,
+      "loss": 0.3478,
+      "step": 21975
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653897804455532e-05,
+      "loss": 0.3042,
+      "step": 21976
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653854955037118e-05,
+      "loss": 0.3181,
+      "step": 21977
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653812103013085e-05,
+      "loss": 0.2622,
+      "step": 21978
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653769248383443e-05,
+      "loss": 0.2971,
+      "step": 21979
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653726391148203e-05,
+      "loss": 0.3132,
+      "step": 21980
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653683531307377e-05,
+      "loss": 0.3185,
+      "step": 21981
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653640668860977e-05,
+      "loss": 0.2774,
+      "step": 21982
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653597803809018e-05,
+      "loss": 0.3486,
+      "step": 21983
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.96535549361515e-05,
+      "loss": 0.2862,
+      "step": 21984
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965351206588845e-05,
+      "loss": 0.3628,
+      "step": 21985
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965346919301987e-05,
+      "loss": 0.2818,
+      "step": 21986
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965342631754577e-05,
+      "loss": 0.306,
+      "step": 21987
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653383439466166e-05,
+      "loss": 0.3418,
+      "step": 21988
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653340558781072e-05,
+      "loss": 0.3386,
+      "step": 21989
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653297675490495e-05,
+      "loss": 0.4196,
+      "step": 21990
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653254789594445e-05,
+      "loss": 0.2759,
+      "step": 21991
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653211901092938e-05,
+      "loss": 0.3034,
+      "step": 21992
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653169009985985e-05,
+      "loss": 0.2738,
+      "step": 21993
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653126116273596e-05,
+      "loss": 0.275,
+      "step": 21994
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9653083219955784e-05,
+      "loss": 0.2539,
+      "step": 21995
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965304032103256e-05,
+      "loss": 0.3489,
+      "step": 21996
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965299741950393e-05,
+      "loss": 0.3379,
+      "step": 21997
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652954515369916e-05,
+      "loss": 0.2668,
+      "step": 21998
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652911608630526e-05,
+      "loss": 0.2953,
+      "step": 21999
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652868699285768e-05,
+      "loss": 0.3022,
+      "step": 22000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652825787335655e-05,
+      "loss": 0.2945,
+      "step": 22001
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652782872780198e-05,
+      "loss": 0.3056,
+      "step": 22002
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652739955619412e-05,
+      "loss": 0.3182,
+      "step": 22003
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652697035853306e-05,
+      "loss": 0.2753,
+      "step": 22004
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965265411348189e-05,
+      "loss": 0.2929,
+      "step": 22005
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965261118850518e-05,
+      "loss": 0.2901,
+      "step": 22006
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652568260923185e-05,
+      "loss": 0.3441,
+      "step": 22007
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652525330735917e-05,
+      "loss": 0.3009,
+      "step": 22008
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652482397943383e-05,
+      "loss": 0.3342,
+      "step": 22009
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652439462545602e-05,
+      "loss": 0.3411,
+      "step": 22010
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652396524542582e-05,
+      "loss": 0.3334,
+      "step": 22011
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965235358393434e-05,
+      "loss": 0.299,
+      "step": 22012
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652310640720874e-05,
+      "loss": 0.3074,
+      "step": 22013
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965226769490221e-05,
+      "loss": 0.2635,
+      "step": 22014
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652224746478353e-05,
+      "loss": 0.3,
+      "step": 22015
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652181795449314e-05,
+      "loss": 0.2745,
+      "step": 22016
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652138841815107e-05,
+      "loss": 0.2687,
+      "step": 22017
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652095885575742e-05,
+      "loss": 0.35,
+      "step": 22018
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652052926731233e-05,
+      "loss": 0.3211,
+      "step": 22019
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9652009965281588e-05,
+      "loss": 0.2625,
+      "step": 22020
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965196700122682e-05,
+      "loss": 0.2826,
+      "step": 22021
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965192403456694e-05,
+      "loss": 0.323,
+      "step": 22022
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9651881065301963e-05,
+      "loss": 0.2462,
+      "step": 22023
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.96518380934319e-05,
+      "loss": 0.2885,
+      "step": 22024
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965179511895676e-05,
+      "loss": 0.2783,
+      "step": 22025
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965175214187655e-05,
+      "loss": 0.3805,
+      "step": 22026
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9651709162191293e-05,
+      "loss": 0.3062,
+      "step": 22027
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965166617990099e-05,
+      "loss": 0.2586,
+      "step": 22028
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965162319500566e-05,
+      "loss": 0.2971,
+      "step": 22029
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.965158020750531e-05,
+      "loss": 0.2918,
+      "step": 22030
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651537217399955e-05,
+      "loss": 0.3191,
+      "step": 22031
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651494224689604e-05,
+      "loss": 0.348,
+      "step": 22032
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.965145122937427e-05,
+      "loss": 0.2692,
+      "step": 22033
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651408231453967e-05,
+      "loss": 0.2976,
+      "step": 22034
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651365230928697e-05,
+      "loss": 0.2597,
+      "step": 22035
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651322227798486e-05,
+      "loss": 0.3215,
+      "step": 22036
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651279222063336e-05,
+      "loss": 0.2797,
+      "step": 22037
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651236213723256e-05,
+      "loss": 0.2678,
+      "step": 22038
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651193202778268e-05,
+      "loss": 0.2985,
+      "step": 22039
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651150189228372e-05,
+      "loss": 0.3069,
+      "step": 22040
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651107173073593e-05,
+      "loss": 0.2598,
+      "step": 22041
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651064154313926e-05,
+      "loss": 0.2865,
+      "step": 22042
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9651021132949396e-05,
+      "loss": 0.3026,
+      "step": 22043
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650978108980013e-05,
+      "loss": 0.2936,
+      "step": 22044
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650935082405783e-05,
+      "loss": 0.3002,
+      "step": 22045
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650892053226724e-05,
+      "loss": 0.3423,
+      "step": 22046
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650849021442838e-05,
+      "loss": 0.3005,
+      "step": 22047
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650805987054146e-05,
+      "loss": 0.2882,
+      "step": 22048
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.965076295006066e-05,
+      "loss": 0.2897,
+      "step": 22049
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650719910462383e-05,
+      "loss": 0.2933,
+      "step": 22050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.965067686825933e-05,
+      "loss": 0.3071,
+      "step": 22051
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650633823451518e-05,
+      "loss": 0.3232,
+      "step": 22052
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650590776038956e-05,
+      "loss": 0.2757,
+      "step": 22053
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.965054772602165e-05,
+      "loss": 0.2787,
+      "step": 22054
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.965050467339962e-05,
+      "loss": 0.3276,
+      "step": 22055
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650461618172872e-05,
+      "loss": 0.3307,
+      "step": 22056
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.965041856034142e-05,
+      "loss": 0.3358,
+      "step": 22057
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.965037549990527e-05,
+      "loss": 0.2869,
+      "step": 22058
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650332436864447e-05,
+      "loss": 0.3082,
+      "step": 22059
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.965028937121895e-05,
+      "loss": 0.2702,
+      "step": 22060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650246302968794e-05,
+      "loss": 0.3134,
+      "step": 22061
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650203232113992e-05,
+      "loss": 0.312,
+      "step": 22062
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650160158654557e-05,
+      "loss": 0.2592,
+      "step": 22063
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650117082590493e-05,
+      "loss": 0.2881,
+      "step": 22064
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650074003921823e-05,
+      "loss": 0.3092,
+      "step": 22065
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9650030922648553e-05,
+      "loss": 0.2602,
+      "step": 22066
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964998783877069e-05,
+      "loss": 0.3261,
+      "step": 22067
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649944752288253e-05,
+      "loss": 0.3089,
+      "step": 22068
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964990166320125e-05,
+      "loss": 0.2575,
+      "step": 22069
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649858571509696e-05,
+      "loss": 0.3578,
+      "step": 22070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.96498154772136e-05,
+      "loss": 0.3017,
+      "step": 22071
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649772380312968e-05,
+      "loss": 0.2794,
+      "step": 22072
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649729280807823e-05,
+      "loss": 0.3332,
+      "step": 22073
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964968617869817e-05,
+      "loss": 0.3002,
+      "step": 22074
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649643073984018e-05,
+      "loss": 0.3194,
+      "step": 22075
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649599966665387e-05,
+      "loss": 0.3219,
+      "step": 22076
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964955685674228e-05,
+      "loss": 0.2677,
+      "step": 22077
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649513744214716e-05,
+      "loss": 0.325,
+      "step": 22078
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.96494706290827e-05,
+      "loss": 0.3016,
+      "step": 22079
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964942751134625e-05,
+      "loss": 0.32,
+      "step": 22080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649384391005372e-05,
+      "loss": 0.2882,
+      "step": 22081
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964934126806008e-05,
+      "loss": 0.2778,
+      "step": 22082
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649298142510386e-05,
+      "loss": 0.3218,
+      "step": 22083
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649255014356302e-05,
+      "loss": 0.3434,
+      "step": 22084
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964921188359784e-05,
+      "loss": 0.2933,
+      "step": 22085
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964916875023501e-05,
+      "loss": 0.2939,
+      "step": 22086
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649125614267822e-05,
+      "loss": 0.3553,
+      "step": 22087
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649082475696294e-05,
+      "loss": 0.259,
+      "step": 22088
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9649039334520427e-05,
+      "loss": 0.3506,
+      "step": 22089
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648996190740243e-05,
+      "loss": 0.3612,
+      "step": 22090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964895304435575e-05,
+      "loss": 0.285,
+      "step": 22091
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964890989536696e-05,
+      "loss": 0.2895,
+      "step": 22092
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648866743773886e-05,
+      "loss": 0.3172,
+      "step": 22093
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648823589576534e-05,
+      "loss": 0.2686,
+      "step": 22094
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648780432774922e-05,
+      "loss": 0.2831,
+      "step": 22095
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648737273369054e-05,
+      "loss": 0.3212,
+      "step": 22096
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964869411135895e-05,
+      "loss": 0.2815,
+      "step": 22097
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964865094674462e-05,
+      "loss": 0.2614,
+      "step": 22098
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964860777952607e-05,
+      "loss": 0.2835,
+      "step": 22099
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648564609703323e-05,
+      "loss": 0.3124,
+      "step": 22100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964852143727638e-05,
+      "loss": 0.283,
+      "step": 22101
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648478262245253e-05,
+      "loss": 0.3311,
+      "step": 22102
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648435084609957e-05,
+      "loss": 0.2735,
+      "step": 22103
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648391904370507e-05,
+      "loss": 0.3332,
+      "step": 22104
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648348721526907e-05,
+      "loss": 0.3072,
+      "step": 22105
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648305536079177e-05,
+      "loss": 0.2788,
+      "step": 22106
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648262348027317e-05,
+      "loss": 0.2943,
+      "step": 22107
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648219157371354e-05,
+      "loss": 0.3305,
+      "step": 22108
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648175964111286e-05,
+      "loss": 0.2812,
+      "step": 22109
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964813276824713e-05,
+      "loss": 0.2648,
+      "step": 22110
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.96480895697789e-05,
+      "loss": 0.3354,
+      "step": 22111
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964804636870661e-05,
+      "loss": 0.2262,
+      "step": 22112
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9648003165030262e-05,
+      "loss": 0.3469,
+      "step": 22113
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964795995874987e-05,
+      "loss": 0.3268,
+      "step": 22114
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9647916749865453e-05,
+      "loss": 0.2641,
+      "step": 22115
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964787353837702e-05,
+      "loss": 0.2856,
+      "step": 22116
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9647830324284576e-05,
+      "loss": 0.3588,
+      "step": 22117
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9647787107588136e-05,
+      "loss": 0.313,
+      "step": 22118
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9647743888287717e-05,
+      "loss": 0.2755,
+      "step": 22119
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964770066638333e-05,
+      "loss": 0.3527,
+      "step": 22120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9647657441874978e-05,
+      "loss": 0.3188,
+      "step": 22121
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964761421476268e-05,
+      "loss": 0.3151,
+      "step": 22122
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964757098504645e-05,
+      "loss": 0.2627,
+      "step": 22123
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9647527752726286e-05,
+      "loss": 0.2865,
+      "step": 22124
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9647484517802216e-05,
+      "loss": 0.3096,
+      "step": 22125
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9647441280274244e-05,
+      "loss": 0.269,
+      "step": 22126
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964739804014238e-05,
+      "loss": 0.331,
+      "step": 22127
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.964735479740664e-05,
+      "loss": 0.3612,
+      "step": 22128
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9647311552067036e-05,
+      "loss": 0.2643,
+      "step": 22129
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9647268304123575e-05,
+      "loss": 0.2721,
+      "step": 22130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964722505357627e-05,
+      "loss": 0.2608,
+      "step": 22131
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9647181800425137e-05,
+      "loss": 0.3069,
+      "step": 22132
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9647138544670182e-05,
+      "loss": 0.3014,
+      "step": 22133
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964709528631142e-05,
+      "loss": 0.2667,
+      "step": 22134
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9647052025348863e-05,
+      "loss": 0.2606,
+      "step": 22135
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964700876178252e-05,
+      "loss": 0.2815,
+      "step": 22136
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646965495612406e-05,
+      "loss": 0.3901,
+      "step": 22137
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964692222683853e-05,
+      "loss": 0.2748,
+      "step": 22138
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646878955460903e-05,
+      "loss": 0.3041,
+      "step": 22139
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964683568147954e-05,
+      "loss": 0.325,
+      "step": 22140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646792404894452e-05,
+      "loss": 0.27,
+      "step": 22141
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646749125705646e-05,
+      "loss": 0.2815,
+      "step": 22142
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646705843913142e-05,
+      "loss": 0.2894,
+      "step": 22143
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646662559516943e-05,
+      "loss": 0.3322,
+      "step": 22144
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646619272517067e-05,
+      "loss": 0.3117,
+      "step": 22145
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646575982913523e-05,
+      "loss": 0.2861,
+      "step": 22146
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646532690706325e-05,
+      "loss": 0.2979,
+      "step": 22147
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646489395895477e-05,
+      "loss": 0.3086,
+      "step": 22148
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646446098481002e-05,
+      "loss": 0.3088,
+      "step": 22149
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646402798462903e-05,
+      "loss": 0.2598,
+      "step": 22150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646359495841195e-05,
+      "loss": 0.3347,
+      "step": 22151
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964631619061589e-05,
+      "loss": 0.2594,
+      "step": 22152
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646272882787e-05,
+      "loss": 0.2742,
+      "step": 22153
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646229572354536e-05,
+      "loss": 0.2953,
+      "step": 22154
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9646186259318507e-05,
+      "loss": 0.3644,
+      "step": 22155
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964614294367893e-05,
+      "loss": 0.3021,
+      "step": 22156
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964609962543581e-05,
+      "loss": 0.2736,
+      "step": 22157
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964605630458917e-05,
+      "loss": 0.3663,
+      "step": 22158
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964601298113901e-05,
+      "loss": 0.2491,
+      "step": 22159
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645969655085344e-05,
+      "loss": 0.3634,
+      "step": 22160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645926326428187e-05,
+      "loss": 0.2776,
+      "step": 22161
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964588299516755e-05,
+      "loss": 0.3091,
+      "step": 22162
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645839661303444e-05,
+      "loss": 0.2755,
+      "step": 22163
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964579632483588e-05,
+      "loss": 0.3371,
+      "step": 22164
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964575298576487e-05,
+      "loss": 0.2823,
+      "step": 22165
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964570964409043e-05,
+      "loss": 0.3286,
+      "step": 22166
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645666299812568e-05,
+      "loss": 0.3085,
+      "step": 22167
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964562295293129e-05,
+      "loss": 0.2903,
+      "step": 22168
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645579603446617e-05,
+      "loss": 0.3332,
+      "step": 22169
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645536251358557e-05,
+      "loss": 0.2637,
+      "step": 22170
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964549289666712e-05,
+      "loss": 0.2767,
+      "step": 22171
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964544953937232e-05,
+      "loss": 0.2829,
+      "step": 22172
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964540617947417e-05,
+      "loss": 0.3232,
+      "step": 22173
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645362816972676e-05,
+      "loss": 0.3443,
+      "step": 22174
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645319451867857e-05,
+      "loss": 0.2866,
+      "step": 22175
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964527608415972e-05,
+      "loss": 0.2761,
+      "step": 22176
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964523271384828e-05,
+      "loss": 0.2923,
+      "step": 22177
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645189340933542e-05,
+      "loss": 0.2676,
+      "step": 22178
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645145965415525e-05,
+      "loss": 0.3597,
+      "step": 22179
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964510258729424e-05,
+      "loss": 0.338,
+      "step": 22180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9645059206569693e-05,
+      "loss": 0.3011,
+      "step": 22181
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.96450158232419e-05,
+      "loss": 0.3034,
+      "step": 22182
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644972437310872e-05,
+      "loss": 0.2974,
+      "step": 22183
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964492904877662e-05,
+      "loss": 0.2785,
+      "step": 22184
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964488565763916e-05,
+      "loss": 0.355,
+      "step": 22185
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.96448422638985e-05,
+      "loss": 0.2994,
+      "step": 22186
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964479886755465e-05,
+      "loss": 0.3228,
+      "step": 22187
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644755468607624e-05,
+      "loss": 0.3253,
+      "step": 22188
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644712067057435e-05,
+      "loss": 0.2937,
+      "step": 22189
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644668662904093e-05,
+      "loss": 0.262,
+      "step": 22190
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644625256147608e-05,
+      "loss": 0.2613,
+      "step": 22191
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644581846787992e-05,
+      "loss": 0.2673,
+      "step": 22192
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644538434825263e-05,
+      "loss": 0.3474,
+      "step": 22193
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644495020259423e-05,
+      "loss": 0.3089,
+      "step": 22194
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964445160309049e-05,
+      "loss": 0.2846,
+      "step": 22195
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964440818331848e-05,
+      "loss": 0.26,
+      "step": 22196
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644364760943394e-05,
+      "loss": 0.3207,
+      "step": 22197
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964432133596525e-05,
+      "loss": 0.2955,
+      "step": 22198
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644277908384057e-05,
+      "loss": 0.2996,
+      "step": 22199
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964423447819983e-05,
+      "loss": 0.259,
+      "step": 22200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964419104541258e-05,
+      "loss": 0.2813,
+      "step": 22201
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644147610022316e-05,
+      "loss": 0.3095,
+      "step": 22202
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9644104172029052e-05,
+      "loss": 0.2706,
+      "step": 22203
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.96440607314328e-05,
+      "loss": 0.2725,
+      "step": 22204
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964401728823357e-05,
+      "loss": 0.2378,
+      "step": 22205
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643973842431373e-05,
+      "loss": 0.307,
+      "step": 22206
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643930394026224e-05,
+      "loss": 0.2519,
+      "step": 22207
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643886943018133e-05,
+      "loss": 0.3014,
+      "step": 22208
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643843489407113e-05,
+      "loss": 0.3253,
+      "step": 22209
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964380003319317e-05,
+      "loss": 0.2709,
+      "step": 22210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643756574376328e-05,
+      "loss": 0.2799,
+      "step": 22211
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643713112956585e-05,
+      "loss": 0.369,
+      "step": 22212
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643669648933962e-05,
+      "loss": 0.2874,
+      "step": 22213
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643626182308467e-05,
+      "loss": 0.3139,
+      "step": 22214
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964358271308011e-05,
+      "loss": 0.3016,
+      "step": 22215
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643539241248904e-05,
+      "loss": 0.3241,
+      "step": 22216
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643495766814867e-05,
+      "loss": 0.2669,
+      "step": 22217
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643452289778002e-05,
+      "loss": 0.2992,
+      "step": 22218
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643408810138324e-05,
+      "loss": 0.2826,
+      "step": 22219
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643365327895846e-05,
+      "loss": 0.2944,
+      "step": 22220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643321843050577e-05,
+      "loss": 0.3159,
+      "step": 22221
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964327835560253e-05,
+      "loss": 0.2987,
+      "step": 22222
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964323486555172e-05,
+      "loss": 0.2855,
+      "step": 22223
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643191372898155e-05,
+      "loss": 0.2766,
+      "step": 22224
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643147877641845e-05,
+      "loss": 0.3115,
+      "step": 22225
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9643104379782807e-05,
+      "loss": 0.3431,
+      "step": 22226
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.964306087932105e-05,
+      "loss": 0.3176,
+      "step": 22227
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9643017376256587e-05,
+      "loss": 0.294,
+      "step": 22228
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642973870589426e-05,
+      "loss": 0.31,
+      "step": 22229
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964293036231958e-05,
+      "loss": 0.2741,
+      "step": 22230
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642886851447066e-05,
+      "loss": 0.2961,
+      "step": 22231
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642843337971887e-05,
+      "loss": 0.2873,
+      "step": 22232
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964279982189406e-05,
+      "loss": 0.3108,
+      "step": 22233
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.96427563032136e-05,
+      "loss": 0.3178,
+      "step": 22234
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642712781930514e-05,
+      "loss": 0.2684,
+      "step": 22235
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642669258044816e-05,
+      "loss": 0.2483,
+      "step": 22236
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642625731556514e-05,
+      "loss": 0.2436,
+      "step": 22237
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642582202465623e-05,
+      "loss": 0.3505,
+      "step": 22238
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642538670772155e-05,
+      "loss": 0.3057,
+      "step": 22239
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642495136476117e-05,
+      "loss": 0.3514,
+      "step": 22240
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964245159957753e-05,
+      "loss": 0.2699,
+      "step": 22241
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642408060076397e-05,
+      "loss": 0.2627,
+      "step": 22242
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964236451797273e-05,
+      "loss": 0.3196,
+      "step": 22243
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964232097326655e-05,
+      "loss": 0.2602,
+      "step": 22244
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964227742595786e-05,
+      "loss": 0.3015,
+      "step": 22245
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642233876046672e-05,
+      "loss": 0.2905,
+      "step": 22246
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642190323533002e-05,
+      "loss": 0.262,
+      "step": 22247
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964214676841686e-05,
+      "loss": 0.2876,
+      "step": 22248
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642103210698254e-05,
+      "loss": 0.2565,
+      "step": 22249
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642059650377207e-05,
+      "loss": 0.2766,
+      "step": 22250
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9642016087453715e-05,
+      "loss": 0.2931,
+      "step": 22251
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641972521927803e-05,
+      "loss": 0.2362,
+      "step": 22252
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641928953799477e-05,
+      "loss": 0.2967,
+      "step": 22253
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641885383068748e-05,
+      "loss": 0.3036,
+      "step": 22254
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641841809735628e-05,
+      "loss": 0.2679,
+      "step": 22255
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964179823380013e-05,
+      "loss": 0.2748,
+      "step": 22256
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641754655262265e-05,
+      "loss": 0.2837,
+      "step": 22257
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964171107412205e-05,
+      "loss": 0.386,
+      "step": 22258
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641667490379488e-05,
+      "loss": 0.3025,
+      "step": 22259
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641623904034593e-05,
+      "loss": 0.3302,
+      "step": 22260
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641580315087383e-05,
+      "loss": 0.2818,
+      "step": 22261
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641536723537868e-05,
+      "loss": 0.3189,
+      "step": 22262
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964149312938605e-05,
+      "loss": 0.2965,
+      "step": 22263
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964144953263195e-05,
+      "loss": 0.2872,
+      "step": 22264
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964140593327558e-05,
+      "loss": 0.3396,
+      "step": 22265
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641362331316948e-05,
+      "loss": 0.3261,
+      "step": 22266
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641318726756068e-05,
+      "loss": 0.3466,
+      "step": 22267
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964127511959295e-05,
+      "loss": 0.2887,
+      "step": 22268
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964123150982761e-05,
+      "loss": 0.3032,
+      "step": 22269
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641187897460053e-05,
+      "loss": 0.3078,
+      "step": 22270
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641144282490293e-05,
+      "loss": 0.278,
+      "step": 22271
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641100664918344e-05,
+      "loss": 0.3058,
+      "step": 22272
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964105704474422e-05,
+      "loss": 0.2736,
+      "step": 22273
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9641013421967927e-05,
+      "loss": 0.2807,
+      "step": 22274
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964096979658948e-05,
+      "loss": 0.2931,
+      "step": 22275
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640926168608893e-05,
+      "loss": 0.3117,
+      "step": 22276
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640882538026168e-05,
+      "loss": 0.2997,
+      "step": 22277
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640838904841327e-05,
+      "loss": 0.3235,
+      "step": 22278
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964079526905438e-05,
+      "loss": 0.2862,
+      "step": 22279
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964075163066534e-05,
+      "loss": 0.3576,
+      "step": 22280
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640707989674213e-05,
+      "loss": 0.2593,
+      "step": 22281
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964066434608101e-05,
+      "loss": 0.3144,
+      "step": 22282
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640620699885754e-05,
+      "loss": 0.2918,
+      "step": 22283
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640577051088443e-05,
+      "loss": 0.2922,
+      "step": 22284
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640533399689098e-05,
+      "loss": 0.2927,
+      "step": 22285
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964048974568773e-05,
+      "loss": 0.3195,
+      "step": 22286
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640446089084344e-05,
+      "loss": 0.2855,
+      "step": 22287
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964040242987896e-05,
+      "loss": 0.3221,
+      "step": 22288
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640358768071585e-05,
+      "loss": 0.3116,
+      "step": 22289
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640315103662233e-05,
+      "loss": 0.2891,
+      "step": 22290
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640271436650915e-05,
+      "loss": 0.2816,
+      "step": 22291
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964022776703764e-05,
+      "loss": 0.3404,
+      "step": 22292
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640184094822426e-05,
+      "loss": 0.2456,
+      "step": 22293
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964014042000528e-05,
+      "loss": 0.3262,
+      "step": 22294
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640096742586213e-05,
+      "loss": 0.3079,
+      "step": 22295
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.964005306256524e-05,
+      "loss": 0.264,
+      "step": 22296
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9640009379942374e-05,
+      "loss": 0.2674,
+      "step": 22297
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639965694717625e-05,
+      "loss": 0.2961,
+      "step": 22298
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639922006890998e-05,
+      "loss": 0.301,
+      "step": 22299
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.963987831646252e-05,
+      "loss": 0.2682,
+      "step": 22300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639834623432184e-05,
+      "loss": 0.356,
+      "step": 22301
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639790927800016e-05,
+      "loss": 0.2866,
+      "step": 22302
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639747229566027e-05,
+      "loss": 0.3243,
+      "step": 22303
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.963970352873022e-05,
+      "loss": 0.38,
+      "step": 22304
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639659825292616e-05,
+      "loss": 0.281,
+      "step": 22305
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.963961611925322e-05,
+      "loss": 0.2673,
+      "step": 22306
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639572410612046e-05,
+      "loss": 0.412,
+      "step": 22307
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639528699369108e-05,
+      "loss": 0.2872,
+      "step": 22308
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639484985524417e-05,
+      "loss": 0.2715,
+      "step": 22309
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639441269077982e-05,
+      "loss": 0.3061,
+      "step": 22310
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639397550029815e-05,
+      "loss": 0.3154,
+      "step": 22311
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639353828379932e-05,
+      "loss": 0.31,
+      "step": 22312
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639310104128343e-05,
+      "loss": 0.2961,
+      "step": 22313
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.963926637727506e-05,
+      "loss": 0.3104,
+      "step": 22314
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639222647820092e-05,
+      "loss": 0.285,
+      "step": 22315
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639178915763453e-05,
+      "loss": 0.2956,
+      "step": 22316
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639135181105153e-05,
+      "loss": 0.2994,
+      "step": 22317
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639091443845205e-05,
+      "loss": 0.2928,
+      "step": 22318
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9639047703983626e-05,
+      "loss": 0.2848,
+      "step": 22319
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.963900396152042e-05,
+      "loss": 0.2807,
+      "step": 22320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.96389602164556e-05,
+      "loss": 0.2608,
+      "step": 22321
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.963891646878918e-05,
+      "loss": 0.3008,
+      "step": 22322
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9638872718521176e-05,
+      "loss": 0.3282,
+      "step": 22323
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.963882896565159e-05,
+      "loss": 0.2989,
+      "step": 22324
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9638785210180443e-05,
+      "loss": 0.2559,
+      "step": 22325
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.963874145210774e-05,
+      "loss": 0.3465,
+      "step": 22326
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638697691433494e-05,
+      "loss": 0.2922,
+      "step": 22327
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963865392815772e-05,
+      "loss": 0.3157,
+      "step": 22328
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638610162280433e-05,
+      "loss": 0.3827,
+      "step": 22329
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638566393801635e-05,
+      "loss": 0.2824,
+      "step": 22330
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638522622721342e-05,
+      "loss": 0.3054,
+      "step": 22331
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963847884903957e-05,
+      "loss": 0.3067,
+      "step": 22332
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638435072756326e-05,
+      "loss": 0.2945,
+      "step": 22333
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638391293871624e-05,
+      "loss": 0.3004,
+      "step": 22334
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638347512385473e-05,
+      "loss": 0.3503,
+      "step": 22335
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963830372829789e-05,
+      "loss": 0.3091,
+      "step": 22336
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638259941608884e-05,
+      "loss": 0.3278,
+      "step": 22337
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638216152318464e-05,
+      "loss": 0.3165,
+      "step": 22338
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638172360426645e-05,
+      "loss": 0.2518,
+      "step": 22339
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963812856593344e-05,
+      "loss": 0.2461,
+      "step": 22340
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9638084768838857e-05,
+      "loss": 0.2702,
+      "step": 22341
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963804096914291e-05,
+      "loss": 0.2654,
+      "step": 22342
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963799716684561e-05,
+      "loss": 0.2912,
+      "step": 22343
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637953361946972e-05,
+      "loss": 0.3123,
+      "step": 22344
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637909554447003e-05,
+      "loss": 0.3294,
+      "step": 22345
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637865744345718e-05,
+      "loss": 0.2888,
+      "step": 22346
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637821931643127e-05,
+      "loss": 0.2892,
+      "step": 22347
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637778116339247e-05,
+      "loss": 0.3023,
+      "step": 22348
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963773429843408e-05,
+      "loss": 0.2885,
+      "step": 22349
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637690477927647e-05,
+      "loss": 0.2304,
+      "step": 22350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637646654819957e-05,
+      "loss": 0.277,
+      "step": 22351
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963760282911102e-05,
+      "loss": 0.2855,
+      "step": 22352
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637559000800847e-05,
+      "loss": 0.2415,
+      "step": 22353
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637515169889453e-05,
+      "loss": 0.2819,
+      "step": 22354
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637471336376848e-05,
+      "loss": 0.3077,
+      "step": 22355
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637427500263046e-05,
+      "loss": 0.3172,
+      "step": 22356
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637383661548056e-05,
+      "loss": 0.3147,
+      "step": 22357
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637339820231892e-05,
+      "loss": 0.2918,
+      "step": 22358
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963729597631456e-05,
+      "loss": 0.2876,
+      "step": 22359
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637252129796084e-05,
+      "loss": 0.3026,
+      "step": 22360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637208280676467e-05,
+      "loss": 0.3774,
+      "step": 22361
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637164428955723e-05,
+      "loss": 0.2961,
+      "step": 22362
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963712057463386e-05,
+      "loss": 0.3393,
+      "step": 22363
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9637076717710894e-05,
+      "loss": 0.2683,
+      "step": 22364
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963703285818684e-05,
+      "loss": 0.2832,
+      "step": 22365
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636988996061702e-05,
+      "loss": 0.2681,
+      "step": 22366
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636945131335496e-05,
+      "loss": 0.3325,
+      "step": 22367
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963690126400823e-05,
+      "loss": 0.311,
+      "step": 22368
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636857394079928e-05,
+      "loss": 0.2734,
+      "step": 22369
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636813521550583e-05,
+      "loss": 0.3149,
+      "step": 22370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636769646420224e-05,
+      "loss": 0.3031,
+      "step": 22371
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636725768688857e-05,
+      "loss": 0.3199,
+      "step": 22372
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963668188835649e-05,
+      "loss": 0.264,
+      "step": 22373
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636638005423133e-05,
+      "loss": 0.2891,
+      "step": 22374
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636594119888807e-05,
+      "loss": 0.2683,
+      "step": 22375
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636550231753518e-05,
+      "loss": 0.3427,
+      "step": 22376
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963650634101728e-05,
+      "loss": 0.2858,
+      "step": 22377
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.96364624476801e-05,
+      "loss": 0.3192,
+      "step": 22378
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636418551742e-05,
+      "loss": 0.293,
+      "step": 22379
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636374653202982e-05,
+      "loss": 0.2589,
+      "step": 22380
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636330752063063e-05,
+      "loss": 0.3249,
+      "step": 22381
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636286848322252e-05,
+      "loss": 0.2906,
+      "step": 22382
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636242941980562e-05,
+      "loss": 0.3069,
+      "step": 22383
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963619903303801e-05,
+      "loss": 0.275,
+      "step": 22384
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636155121494595e-05,
+      "loss": 0.3318,
+      "step": 22385
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636111207350342e-05,
+      "loss": 0.2676,
+      "step": 22386
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636067290605254e-05,
+      "loss": 0.3246,
+      "step": 22387
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963602337125935e-05,
+      "loss": 0.2696,
+      "step": 22388
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635979449312637e-05,
+      "loss": 0.2864,
+      "step": 22389
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635935524765126e-05,
+      "loss": 0.2471,
+      "step": 22390
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635891597616833e-05,
+      "loss": 0.2528,
+      "step": 22391
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635847667867767e-05,
+      "loss": 0.3171,
+      "step": 22392
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963580373551794e-05,
+      "loss": 0.351,
+      "step": 22393
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635759800567363e-05,
+      "loss": 0.3052,
+      "step": 22394
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635715863016056e-05,
+      "loss": 0.2998,
+      "step": 22395
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963567192286402e-05,
+      "loss": 0.2873,
+      "step": 22396
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963562798011127e-05,
+      "loss": 0.2765,
+      "step": 22397
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963558403475782e-05,
+      "loss": 0.3056,
+      "step": 22398
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963554008680368e-05,
+      "loss": 0.3067,
+      "step": 22399
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635496136248866e-05,
+      "loss": 0.3307,
+      "step": 22400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635452183093382e-05,
+      "loss": 0.3283,
+      "step": 22401
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635408227337247e-05,
+      "loss": 0.3312,
+      "step": 22402
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963536426898047e-05,
+      "loss": 0.2948,
+      "step": 22403
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635320308023063e-05,
+      "loss": 0.2957,
+      "step": 22404
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963527634446504e-05,
+      "loss": 0.2624,
+      "step": 22405
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635232378306407e-05,
+      "loss": 0.3584,
+      "step": 22406
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963518840954718e-05,
+      "loss": 0.3105,
+      "step": 22407
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635144438187374e-05,
+      "loss": 0.32,
+      "step": 22408
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635100464226996e-05,
+      "loss": 0.2684,
+      "step": 22409
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963505648766606e-05,
+      "loss": 0.3019,
+      "step": 22410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9635012508504575e-05,
+      "loss": 0.3228,
+      "step": 22411
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9634968526742556e-05,
+      "loss": 0.2961,
+      "step": 22412
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9634924542380017e-05,
+      "loss": 0.2893,
+      "step": 22413
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9634880555416962e-05,
+      "loss": 0.3478,
+      "step": 22414
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963483656585341e-05,
+      "loss": 0.3904,
+      "step": 22415
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963479257368937e-05,
+      "loss": 0.2901,
+      "step": 22416
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9634748578924856e-05,
+      "loss": 0.3632,
+      "step": 22417
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9634704581559877e-05,
+      "loss": 0.2937,
+      "step": 22418
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9634660581594445e-05,
+      "loss": 0.3053,
+      "step": 22419
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9634616579028576e-05,
+      "loss": 0.2688,
+      "step": 22420
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9634572573862277e-05,
+      "loss": 0.2498,
+      "step": 22421
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963452856609556e-05,
+      "loss": 0.2896,
+      "step": 22422
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9634484555728443e-05,
+      "loss": 0.2832,
+      "step": 22423
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.963444054276093e-05,
+      "loss": 0.326,
+      "step": 22424
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963439652719304e-05,
+      "loss": 0.2876,
+      "step": 22425
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963435250902478e-05,
+      "loss": 0.2718,
+      "step": 22426
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9634308488256162e-05,
+      "loss": 0.282,
+      "step": 22427
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.96342644648872e-05,
+      "loss": 0.3365,
+      "step": 22428
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9634220438917905e-05,
+      "loss": 0.2936,
+      "step": 22429
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9634176410348286e-05,
+      "loss": 0.3261,
+      "step": 22430
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9634132379178364e-05,
+      "loss": 0.3423,
+      "step": 22431
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9634088345408136e-05,
+      "loss": 0.302,
+      "step": 22432
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9634044309037632e-05,
+      "loss": 0.2915,
+      "step": 22433
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963400027006685e-05,
+      "loss": 0.3039,
+      "step": 22434
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633956228495803e-05,
+      "loss": 0.2983,
+      "step": 22435
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963391218432451e-05,
+      "loss": 0.3374,
+      "step": 22436
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633868137552978e-05,
+      "loss": 0.281,
+      "step": 22437
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633824088181223e-05,
+      "loss": 0.2784,
+      "step": 22438
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633780036209253e-05,
+      "loss": 0.2632,
+      "step": 22439
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633735981637074e-05,
+      "loss": 0.2789,
+      "step": 22440
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963369192446471e-05,
+      "loss": 0.2857,
+      "step": 22441
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633647864692168e-05,
+      "loss": 0.3343,
+      "step": 22442
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963360380231946e-05,
+      "loss": 0.2713,
+      "step": 22443
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633559737346594e-05,
+      "loss": 0.2423,
+      "step": 22444
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963351566977359e-05,
+      "loss": 0.338,
+      "step": 22445
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633471599600452e-05,
+      "loss": 0.2994,
+      "step": 22446
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633427526827193e-05,
+      "loss": 0.3047,
+      "step": 22447
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963338345145383e-05,
+      "loss": 0.3004,
+      "step": 22448
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633339373480368e-05,
+      "loss": 0.2931,
+      "step": 22449
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633295292906826e-05,
+      "loss": 0.3162,
+      "step": 22450
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963325120973321e-05,
+      "loss": 0.2716,
+      "step": 22451
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633207123959538e-05,
+      "loss": 0.3386,
+      "step": 22452
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963316303558582e-05,
+      "loss": 0.3152,
+      "step": 22453
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963311894461206e-05,
+      "loss": 0.2743,
+      "step": 22454
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633074851038277e-05,
+      "loss": 0.3133,
+      "step": 22455
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9633030754864485e-05,
+      "loss": 0.2694,
+      "step": 22456
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632986656090692e-05,
+      "loss": 0.2441,
+      "step": 22457
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632942554716915e-05,
+      "loss": 0.2926,
+      "step": 22458
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632898450743156e-05,
+      "loss": 0.2882,
+      "step": 22459
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632854344169433e-05,
+      "loss": 0.3041,
+      "step": 22460
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963281023499576e-05,
+      "loss": 0.3336,
+      "step": 22461
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632766123222148e-05,
+      "loss": 0.3019,
+      "step": 22462
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632722008848604e-05,
+      "loss": 0.2998,
+      "step": 22463
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632677891875142e-05,
+      "loss": 0.3078,
+      "step": 22464
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632633772301777e-05,
+      "loss": 0.3388,
+      "step": 22465
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632589650128523e-05,
+      "loss": 0.3084,
+      "step": 22466
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632545525355386e-05,
+      "loss": 0.3514,
+      "step": 22467
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632501397982376e-05,
+      "loss": 0.3365,
+      "step": 22468
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632457268009513e-05,
+      "loss": 0.3059,
+      "step": 22469
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.96324131354368e-05,
+      "loss": 0.3223,
+      "step": 22470
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963236900026426e-05,
+      "loss": 0.2813,
+      "step": 22471
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632324862491896e-05,
+      "loss": 0.344,
+      "step": 22472
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963228072211972e-05,
+      "loss": 0.3087,
+      "step": 22473
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963223657914775e-05,
+      "loss": 0.2817,
+      "step": 22474
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632192433575992e-05,
+      "loss": 0.312,
+      "step": 22475
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963214828540446e-05,
+      "loss": 0.3349,
+      "step": 22476
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963210413463317e-05,
+      "loss": 0.2791,
+      "step": 22477
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632059981262125e-05,
+      "loss": 0.2965,
+      "step": 22478
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9632015825291345e-05,
+      "loss": 0.283,
+      "step": 22479
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631971666720837e-05,
+      "loss": 0.3751,
+      "step": 22480
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631927505550616e-05,
+      "loss": 0.3295,
+      "step": 22481
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631883341780692e-05,
+      "loss": 0.3366,
+      "step": 22482
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963183917541108e-05,
+      "loss": 0.2461,
+      "step": 22483
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631795006441788e-05,
+      "loss": 0.3078,
+      "step": 22484
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631750834872828e-05,
+      "loss": 0.2411,
+      "step": 22485
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631706660704215e-05,
+      "loss": 0.3035,
+      "step": 22486
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631662483935957e-05,
+      "loss": 0.3135,
+      "step": 22487
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631618304568073e-05,
+      "loss": 0.329,
+      "step": 22488
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631574122600568e-05,
+      "loss": 0.2765,
+      "step": 22489
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631529938033457e-05,
+      "loss": 0.314,
+      "step": 22490
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631485750866748e-05,
+      "loss": 0.3499,
+      "step": 22491
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631441561100457e-05,
+      "loss": 0.2761,
+      "step": 22492
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631397368734596e-05,
+      "loss": 0.3198,
+      "step": 22493
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631353173769176e-05,
+      "loss": 0.2711,
+      "step": 22494
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963130897620421e-05,
+      "loss": 0.3439,
+      "step": 22495
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631264776039705e-05,
+      "loss": 0.2661,
+      "step": 22496
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963122057327568e-05,
+      "loss": 0.3387,
+      "step": 22497
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631176367912142e-05,
+      "loss": 0.3572,
+      "step": 22498
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631132159949104e-05,
+      "loss": 0.3203,
+      "step": 22499
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963108794938658e-05,
+      "loss": 0.3071,
+      "step": 22500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9631043736224578e-05,
+      "loss": 0.273,
+      "step": 22501
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630999520463114e-05,
+      "loss": 0.3001,
+      "step": 22502
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630955302102198e-05,
+      "loss": 0.2949,
+      "step": 22503
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630911081141842e-05,
+      "loss": 0.3154,
+      "step": 22504
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630866857582057e-05,
+      "loss": 0.3228,
+      "step": 22505
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630822631422857e-05,
+      "loss": 0.3036,
+      "step": 22506
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630778402664252e-05,
+      "loss": 0.256,
+      "step": 22507
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963073417130626e-05,
+      "loss": 0.3419,
+      "step": 22508
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963068993734888e-05,
+      "loss": 0.267,
+      "step": 22509
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630645700792134e-05,
+      "loss": 0.2735,
+      "step": 22510
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630601461636037e-05,
+      "loss": 0.3264,
+      "step": 22511
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630557219880592e-05,
+      "loss": 0.3373,
+      "step": 22512
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630512975525813e-05,
+      "loss": 0.3345,
+      "step": 22513
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630468728571714e-05,
+      "loss": 0.3185,
+      "step": 22514
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630424479018307e-05,
+      "loss": 0.2483,
+      "step": 22515
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630380226865604e-05,
+      "loss": 0.2744,
+      "step": 22516
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630335972113618e-05,
+      "loss": 0.3138,
+      "step": 22517
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.963029171476236e-05,
+      "loss": 0.2461,
+      "step": 22518
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630247454811836e-05,
+      "loss": 0.2662,
+      "step": 22519
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630203192262065e-05,
+      "loss": 0.3203,
+      "step": 22520
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630158927113057e-05,
+      "loss": 0.278,
+      "step": 22521
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630114659364827e-05,
+      "loss": 0.293,
+      "step": 22522
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9630070389017382e-05,
+      "loss": 0.332,
+      "step": 22523
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9630026116070735e-05,
+      "loss": 0.3455,
+      "step": 22524
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.96299818405249e-05,
+      "loss": 0.3327,
+      "step": 22525
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962993756237989e-05,
+      "loss": 0.3332,
+      "step": 22526
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962989328163571e-05,
+      "loss": 0.2883,
+      "step": 22527
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962984899829238e-05,
+      "loss": 0.2536,
+      "step": 22528
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629804712349908e-05,
+      "loss": 0.332,
+      "step": 22529
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629760423808307e-05,
+      "loss": 0.3608,
+      "step": 22530
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962971613266759e-05,
+      "loss": 0.2659,
+      "step": 22531
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629671838927767e-05,
+      "loss": 0.3703,
+      "step": 22532
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629627542588848e-05,
+      "loss": 0.31,
+      "step": 22533
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629583243650853e-05,
+      "loss": 0.2968,
+      "step": 22534
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962953894211378e-05,
+      "loss": 0.2745,
+      "step": 22535
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629494637977656e-05,
+      "loss": 0.2643,
+      "step": 22536
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629450331242485e-05,
+      "loss": 0.306,
+      "step": 22537
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629406021908282e-05,
+      "loss": 0.2566,
+      "step": 22538
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629361709975053e-05,
+      "loss": 0.316,
+      "step": 22539
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962931739544282e-05,
+      "loss": 0.3498,
+      "step": 22540
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629273078311582e-05,
+      "loss": 0.398,
+      "step": 22541
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629228758581364e-05,
+      "loss": 0.2894,
+      "step": 22542
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629184436252168e-05,
+      "loss": 0.3019,
+      "step": 22543
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629140111324013e-05,
+      "loss": 0.3059,
+      "step": 22544
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629095783796908e-05,
+      "loss": 0.3518,
+      "step": 22545
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629051453670864e-05,
+      "loss": 0.339,
+      "step": 22546
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9629007120945897e-05,
+      "loss": 0.2784,
+      "step": 22547
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628962785622012e-05,
+      "loss": 0.2641,
+      "step": 22548
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628918447699227e-05,
+      "loss": 0.3185,
+      "step": 22549
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962887410717755e-05,
+      "loss": 0.2717,
+      "step": 22550
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628829764056997e-05,
+      "loss": 0.305,
+      "step": 22551
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628785418337575e-05,
+      "loss": 0.3021,
+      "step": 22552
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.96287410700193e-05,
+      "loss": 0.2904,
+      "step": 22553
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628696719102186e-05,
+      "loss": 0.3257,
+      "step": 22554
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962865236558624e-05,
+      "loss": 0.2827,
+      "step": 22555
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628608009471474e-05,
+      "loss": 0.2736,
+      "step": 22556
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628563650757903e-05,
+      "loss": 0.325,
+      "step": 22557
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628519289445536e-05,
+      "loss": 0.3795,
+      "step": 22558
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628474925534388e-05,
+      "loss": 0.3179,
+      "step": 22559
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628430559024468e-05,
+      "loss": 0.2824,
+      "step": 22560
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962838618991579e-05,
+      "loss": 0.2922,
+      "step": 22561
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962834181820837e-05,
+      "loss": 0.2914,
+      "step": 22562
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962829744390221e-05,
+      "loss": 0.2851,
+      "step": 22563
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628253066997333e-05,
+      "loss": 0.2798,
+      "step": 22564
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962820868749374e-05,
+      "loss": 0.3247,
+      "step": 22565
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628164305391452e-05,
+      "loss": 0.2938,
+      "step": 22566
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628119920690473e-05,
+      "loss": 0.2956,
+      "step": 22567
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9628075533390824e-05,
+      "loss": 0.2557,
+      "step": 22568
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962803114349251e-05,
+      "loss": 0.3662,
+      "step": 22569
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962798675099555e-05,
+      "loss": 0.3228,
+      "step": 22570
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627942355899945e-05,
+      "loss": 0.339,
+      "step": 22571
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627897958205717e-05,
+      "loss": 0.3158,
+      "step": 22572
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627853557912874e-05,
+      "loss": 0.2901,
+      "step": 22573
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627809155021425e-05,
+      "loss": 0.277,
+      "step": 22574
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962776474953139e-05,
+      "loss": 0.304,
+      "step": 22575
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627720341442776e-05,
+      "loss": 0.315,
+      "step": 22576
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962767593075559e-05,
+      "loss": 0.2737,
+      "step": 22577
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627631517469852e-05,
+      "loss": 0.3054,
+      "step": 22578
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627587101585576e-05,
+      "loss": 0.3658,
+      "step": 22579
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627542683102764e-05,
+      "loss": 0.2713,
+      "step": 22580
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627498262021435e-05,
+      "loss": 0.3073,
+      "step": 22581
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627453838341596e-05,
+      "loss": 0.248,
+      "step": 22582
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627409412063266e-05,
+      "loss": 0.295,
+      "step": 22583
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627364983186452e-05,
+      "loss": 0.33,
+      "step": 22584
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962732055171117e-05,
+      "loss": 0.3439,
+      "step": 22585
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627276117637425e-05,
+      "loss": 0.3591,
+      "step": 22586
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627231680965235e-05,
+      "loss": 0.2864,
+      "step": 22587
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962718724169461e-05,
+      "loss": 0.3082,
+      "step": 22588
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962714279982556e-05,
+      "loss": 0.3268,
+      "step": 22589
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627098355358102e-05,
+      "loss": 0.2969,
+      "step": 22590
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627053908292246e-05,
+      "loss": 0.3102,
+      "step": 22591
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9627009458628e-05,
+      "loss": 0.2739,
+      "step": 22592
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626965006365383e-05,
+      "loss": 0.3042,
+      "step": 22593
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.96269205515044e-05,
+      "loss": 0.3305,
+      "step": 22594
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626876094045066e-05,
+      "loss": 0.3054,
+      "step": 22595
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626831633987395e-05,
+      "loss": 0.3101,
+      "step": 22596
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626787171331395e-05,
+      "loss": 0.2933,
+      "step": 22597
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962674270607708e-05,
+      "loss": 0.3091,
+      "step": 22598
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626698238224463e-05,
+      "loss": 0.259,
+      "step": 22599
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626653767773558e-05,
+      "loss": 0.3085,
+      "step": 22600
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962660929472437e-05,
+      "loss": 0.3101,
+      "step": 22601
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626564819076917e-05,
+      "loss": 0.2885,
+      "step": 22602
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626520340831212e-05,
+      "loss": 0.3183,
+      "step": 22603
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962647585998726e-05,
+      "loss": 0.3393,
+      "step": 22604
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962643137654508e-05,
+      "loss": 0.3304,
+      "step": 22605
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626386890504678e-05,
+      "loss": 0.3005,
+      "step": 22606
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962634240186607e-05,
+      "loss": 0.2911,
+      "step": 22607
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962629791062927e-05,
+      "loss": 0.3526,
+      "step": 22608
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626253416794285e-05,
+      "loss": 0.279,
+      "step": 22609
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962620892036113e-05,
+      "loss": 0.2732,
+      "step": 22610
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626164421329814e-05,
+      "loss": 0.3152,
+      "step": 22611
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962611991970035e-05,
+      "loss": 0.2792,
+      "step": 22612
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9626075415472757e-05,
+      "loss": 0.2953,
+      "step": 22613
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962603090864704e-05,
+      "loss": 0.3073,
+      "step": 22614
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962598639922321e-05,
+      "loss": 0.3073,
+      "step": 22615
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962594188720128e-05,
+      "loss": 0.2659,
+      "step": 22616
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9625897372581265e-05,
+      "loss": 0.2931,
+      "step": 22617
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9625852855363177e-05,
+      "loss": 0.309,
+      "step": 22618
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9625808335547023e-05,
+      "loss": 0.2935,
+      "step": 22619
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962576381313282e-05,
+      "loss": 0.2786,
+      "step": 22620
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.962571928812058e-05,
+      "loss": 0.3315,
+      "step": 22621
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962567476051031e-05,
+      "loss": 0.2865,
+      "step": 22622
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625630230302027e-05,
+      "loss": 0.3067,
+      "step": 22623
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625585697495745e-05,
+      "loss": 0.2799,
+      "step": 22624
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625541162091466e-05,
+      "loss": 0.36,
+      "step": 22625
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625496624089213e-05,
+      "loss": 0.2929,
+      "step": 22626
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962545208348899e-05,
+      "loss": 0.3062,
+      "step": 22627
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625407540290814e-05,
+      "loss": 0.3139,
+      "step": 22628
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625362994494698e-05,
+      "loss": 0.3845,
+      "step": 22629
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625318446100648e-05,
+      "loss": 0.3533,
+      "step": 22630
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625273895108683e-05,
+      "loss": 0.2376,
+      "step": 22631
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625229341518808e-05,
+      "loss": 0.2847,
+      "step": 22632
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962518478533104e-05,
+      "loss": 0.3305,
+      "step": 22633
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625140226545388e-05,
+      "loss": 0.2788,
+      "step": 22634
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625095665161867e-05,
+      "loss": 0.3301,
+      "step": 22635
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625051101180493e-05,
+      "loss": 0.2867,
+      "step": 22636
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9625006534601266e-05,
+      "loss": 0.3405,
+      "step": 22637
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624961965424205e-05,
+      "loss": 0.3303,
+      "step": 22638
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624917393649325e-05,
+      "loss": 0.3028,
+      "step": 22639
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624872819276636e-05,
+      "loss": 0.3494,
+      "step": 22640
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624828242306143e-05,
+      "loss": 0.2911,
+      "step": 22641
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962478366273787e-05,
+      "loss": 0.2714,
+      "step": 22642
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962473908057182e-05,
+      "loss": 0.2892,
+      "step": 22643
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624694495808004e-05,
+      "loss": 0.3202,
+      "step": 22644
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624649908446443e-05,
+      "loss": 0.3227,
+      "step": 22645
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962460531848714e-05,
+      "loss": 0.2946,
+      "step": 22646
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624560725930118e-05,
+      "loss": 0.3051,
+      "step": 22647
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624516130775374e-05,
+      "loss": 0.2992,
+      "step": 22648
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962447153302293e-05,
+      "loss": 0.3254,
+      "step": 22649
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624426932672802e-05,
+      "loss": 0.2811,
+      "step": 22650
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624382329724995e-05,
+      "loss": 0.3337,
+      "step": 22651
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624337724179518e-05,
+      "loss": 0.3109,
+      "step": 22652
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624293116036387e-05,
+      "loss": 0.3332,
+      "step": 22653
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624248505295616e-05,
+      "loss": 0.3292,
+      "step": 22654
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624203891957216e-05,
+      "loss": 0.2901,
+      "step": 22655
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.96241592760212e-05,
+      "loss": 0.2876,
+      "step": 22656
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624114657487572e-05,
+      "loss": 0.2854,
+      "step": 22657
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624070036356357e-05,
+      "loss": 0.3087,
+      "step": 22658
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9624025412627556e-05,
+      "loss": 0.361,
+      "step": 22659
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962398078630119e-05,
+      "loss": 0.3155,
+      "step": 22660
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623936157377263e-05,
+      "loss": 0.2982,
+      "step": 22661
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623891525855793e-05,
+      "loss": 0.2767,
+      "step": 22662
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623846891736787e-05,
+      "loss": 0.2808,
+      "step": 22663
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962380225502026e-05,
+      "loss": 0.2845,
+      "step": 22664
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623757615706226e-05,
+      "loss": 0.293,
+      "step": 22665
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623712973794697e-05,
+      "loss": 0.3296,
+      "step": 22666
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962366832928568e-05,
+      "loss": 0.3822,
+      "step": 22667
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623623682179186e-05,
+      "loss": 0.3075,
+      "step": 22668
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962357903247524e-05,
+      "loss": 0.3072,
+      "step": 22669
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623534380173837e-05,
+      "loss": 0.3006,
+      "step": 22670
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623489725275e-05,
+      "loss": 0.3,
+      "step": 22671
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962344506777874e-05,
+      "loss": 0.2872,
+      "step": 22672
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962340040768506e-05,
+      "loss": 0.3145,
+      "step": 22673
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623355744993988e-05,
+      "loss": 0.3228,
+      "step": 22674
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962331107970552e-05,
+      "loss": 0.2929,
+      "step": 22675
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962326641181968e-05,
+      "loss": 0.2954,
+      "step": 22676
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623221741336476e-05,
+      "loss": 0.3493,
+      "step": 22677
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962317706825592e-05,
+      "loss": 0.2793,
+      "step": 22678
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623132392578022e-05,
+      "loss": 0.3242,
+      "step": 22679
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623087714302793e-05,
+      "loss": 0.2743,
+      "step": 22680
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9623043033430252e-05,
+      "loss": 0.3201,
+      "step": 22681
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622998349960403e-05,
+      "loss": 0.2888,
+      "step": 22682
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622953663893263e-05,
+      "loss": 0.3085,
+      "step": 22683
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622908975228843e-05,
+      "loss": 0.2686,
+      "step": 22684
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622864283967155e-05,
+      "loss": 0.2967,
+      "step": 22685
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962281959010821e-05,
+      "loss": 0.327,
+      "step": 22686
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962277489365202e-05,
+      "loss": 0.3006,
+      "step": 22687
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622730194598603e-05,
+      "loss": 0.3203,
+      "step": 22688
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622685492947964e-05,
+      "loss": 0.3233,
+      "step": 22689
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622640788700116e-05,
+      "loss": 0.2787,
+      "step": 22690
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622596081855072e-05,
+      "loss": 0.2652,
+      "step": 22691
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622551372412845e-05,
+      "loss": 0.2757,
+      "step": 22692
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622506660373446e-05,
+      "loss": 0.312,
+      "step": 22693
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622461945736887e-05,
+      "loss": 0.3343,
+      "step": 22694
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962241722850318e-05,
+      "loss": 0.3012,
+      "step": 22695
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622372508672342e-05,
+      "loss": 0.3531,
+      "step": 22696
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622327786244378e-05,
+      "loss": 0.366,
+      "step": 22697
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622283061219303e-05,
+      "loss": 0.2641,
+      "step": 22698
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622238333597127e-05,
+      "loss": 0.3703,
+      "step": 22699
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622193603377866e-05,
+      "loss": 0.2958,
+      "step": 22700
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622148870561525e-05,
+      "loss": 0.2554,
+      "step": 22701
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962210413514813e-05,
+      "loss": 0.2791,
+      "step": 22702
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962205939713768e-05,
+      "loss": 0.3163,
+      "step": 22703
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9622014656530187e-05,
+      "loss": 0.3508,
+      "step": 22704
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962196991332567e-05,
+      "loss": 0.2822,
+      "step": 22705
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962192516752414e-05,
+      "loss": 0.3524,
+      "step": 22706
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9621880419125605e-05,
+      "loss": 0.2729,
+      "step": 22707
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9621835668130083e-05,
+      "loss": 0.2606,
+      "step": 22708
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962179091453758e-05,
+      "loss": 0.3144,
+      "step": 22709
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962174615834811e-05,
+      "loss": 0.3412,
+      "step": 22710
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962170139956169e-05,
+      "loss": 0.3562,
+      "step": 22711
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962165663817832e-05,
+      "loss": 0.2764,
+      "step": 22712
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9621611874198028e-05,
+      "loss": 0.2628,
+      "step": 22713
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9621567107620813e-05,
+      "loss": 0.285,
+      "step": 22714
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9621522338446693e-05,
+      "loss": 0.311,
+      "step": 22715
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9621477566675683e-05,
+      "loss": 0.3277,
+      "step": 22716
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9621432792307787e-05,
+      "loss": 0.2886,
+      "step": 22717
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962138801534302e-05,
+      "loss": 0.2823,
+      "step": 22718
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9621343235781398e-05,
+      "loss": 0.3059,
+      "step": 22719
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.962129845362293e-05,
+      "loss": 0.2653,
+      "step": 22720
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962125366886763e-05,
+      "loss": 0.3269,
+      "step": 22721
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962120888151551e-05,
+      "loss": 0.2776,
+      "step": 22722
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9621164091566575e-05,
+      "loss": 0.3621,
+      "step": 22723
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9621119299020848e-05,
+      "loss": 0.342,
+      "step": 22724
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9621074503878335e-05,
+      "loss": 0.3257,
+      "step": 22725
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962102970613905e-05,
+      "loss": 0.3051,
+      "step": 22726
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620984905802997e-05,
+      "loss": 0.339,
+      "step": 22727
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620940102870203e-05,
+      "loss": 0.3268,
+      "step": 22728
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620895297340667e-05,
+      "loss": 0.301,
+      "step": 22729
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962085048921441e-05,
+      "loss": 0.3162,
+      "step": 22730
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962080567849144e-05,
+      "loss": 0.3242,
+      "step": 22731
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962076086517177e-05,
+      "loss": 0.2891,
+      "step": 22732
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620716049255407e-05,
+      "loss": 0.3036,
+      "step": 22733
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620671230742374e-05,
+      "loss": 0.3354,
+      "step": 22734
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620626409632676e-05,
+      "loss": 0.3159,
+      "step": 22735
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620581585926325e-05,
+      "loss": 0.3205,
+      "step": 22736
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962053675962333e-05,
+      "loss": 0.3509,
+      "step": 22737
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962049193072371e-05,
+      "loss": 0.3345,
+      "step": 22738
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620447099227475e-05,
+      "loss": 0.2512,
+      "step": 22739
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620402265134636e-05,
+      "loss": 0.3084,
+      "step": 22740
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620357428445208e-05,
+      "loss": 0.2699,
+      "step": 22741
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620312589159196e-05,
+      "loss": 0.2952,
+      "step": 22742
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962026774727662e-05,
+      "loss": 0.3095,
+      "step": 22743
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962022290279749e-05,
+      "loss": 0.2614,
+      "step": 22744
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620178055721814e-05,
+      "loss": 0.2787,
+      "step": 22745
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620133206049606e-05,
+      "loss": 0.3233,
+      "step": 22746
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.962008835378088e-05,
+      "loss": 0.304,
+      "step": 22747
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9620043498915646e-05,
+      "loss": 0.2751,
+      "step": 22748
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961999864145392e-05,
+      "loss": 0.2773,
+      "step": 22749
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961995378139571e-05,
+      "loss": 0.3174,
+      "step": 22750
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619908918741032e-05,
+      "loss": 0.3655,
+      "step": 22751
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619864053489893e-05,
+      "loss": 0.26,
+      "step": 22752
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619819185642307e-05,
+      "loss": 0.295,
+      "step": 22753
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961977431519829e-05,
+      "loss": 0.2879,
+      "step": 22754
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619729442157847e-05,
+      "loss": 0.2983,
+      "step": 22755
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619684566520996e-05,
+      "loss": 0.3011,
+      "step": 22756
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961963968828775e-05,
+      "loss": 0.2942,
+      "step": 22757
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619594807458112e-05,
+      "loss": 0.3323,
+      "step": 22758
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619549924032103e-05,
+      "loss": 0.3473,
+      "step": 22759
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619505038009737e-05,
+      "loss": 0.3311,
+      "step": 22760
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961946014939102e-05,
+      "loss": 0.2496,
+      "step": 22761
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619415258175962e-05,
+      "loss": 0.2971,
+      "step": 22762
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619370364364577e-05,
+      "loss": 0.3182,
+      "step": 22763
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619325467956886e-05,
+      "loss": 0.3353,
+      "step": 22764
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961928056895289e-05,
+      "loss": 0.2759,
+      "step": 22765
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619235667352606e-05,
+      "loss": 0.2655,
+      "step": 22766
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619190763156048e-05,
+      "loss": 0.422,
+      "step": 22767
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619145856363218e-05,
+      "loss": 0.2679,
+      "step": 22768
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961910094697414e-05,
+      "loss": 0.2764,
+      "step": 22769
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619056034988823e-05,
+      "loss": 0.2833,
+      "step": 22770
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9619011120407275e-05,
+      "loss": 0.3095,
+      "step": 22771
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9618966203229516e-05,
+      "loss": 0.348,
+      "step": 22772
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961892128345555e-05,
+      "loss": 0.2795,
+      "step": 22773
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961887636108539e-05,
+      "loss": 0.3137,
+      "step": 22774
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9618831436119054e-05,
+      "loss": 0.3048,
+      "step": 22775
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961878650855655e-05,
+      "loss": 0.3333,
+      "step": 22776
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961874157839789e-05,
+      "loss": 0.3294,
+      "step": 22777
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9618696645643083e-05,
+      "loss": 0.3342,
+      "step": 22778
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9618651710292148e-05,
+      "loss": 0.2923,
+      "step": 22779
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9618606772345094e-05,
+      "loss": 0.2965,
+      "step": 22780
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961856183180193e-05,
+      "loss": 0.249,
+      "step": 22781
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9618516888662674e-05,
+      "loss": 0.2858,
+      "step": 22782
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961847194292734e-05,
+      "loss": 0.3807,
+      "step": 22783
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961842699459593e-05,
+      "loss": 0.3156,
+      "step": 22784
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961838204366846e-05,
+      "loss": 0.3192,
+      "step": 22785
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961833709014495e-05,
+      "loss": 0.3246,
+      "step": 22786
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.96182921340254e-05,
+      "loss": 0.3493,
+      "step": 22787
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961824717530983e-05,
+      "loss": 0.3674,
+      "step": 22788
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9618202213998246e-05,
+      "loss": 0.3223,
+      "step": 22789
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961815725009067e-05,
+      "loss": 0.2778,
+      "step": 22790
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961811228358711e-05,
+      "loss": 0.2956,
+      "step": 22791
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961806731448757e-05,
+      "loss": 0.2744,
+      "step": 22792
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9618022342792073e-05,
+      "loss": 0.3143,
+      "step": 22793
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617977368500625e-05,
+      "loss": 0.267,
+      "step": 22794
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961793239161324e-05,
+      "loss": 0.4017,
+      "step": 22795
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617887412129933e-05,
+      "loss": 0.2697,
+      "step": 22796
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617842430050708e-05,
+      "loss": 0.2814,
+      "step": 22797
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617797445375588e-05,
+      "loss": 0.2867,
+      "step": 22798
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617752458104575e-05,
+      "loss": 0.2976,
+      "step": 22799
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617707468237688e-05,
+      "loss": 0.3209,
+      "step": 22800
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617662475774937e-05,
+      "loss": 0.3046,
+      "step": 22801
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617617480716334e-05,
+      "loss": 0.3087,
+      "step": 22802
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617572483061887e-05,
+      "loss": 0.2908,
+      "step": 22803
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961752748281162e-05,
+      "loss": 0.3044,
+      "step": 22804
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961748247996553e-05,
+      "loss": 0.3175,
+      "step": 22805
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961743747452364e-05,
+      "loss": 0.3133,
+      "step": 22806
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617392466485958e-05,
+      "loss": 0.3542,
+      "step": 22807
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617347455852496e-05,
+      "loss": 0.2421,
+      "step": 22808
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961730244262327e-05,
+      "loss": 0.3138,
+      "step": 22809
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617257426798284e-05,
+      "loss": 0.2935,
+      "step": 22810
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961721240837756e-05,
+      "loss": 0.2855,
+      "step": 22811
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.96171673873611e-05,
+      "loss": 0.2962,
+      "step": 22812
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617122363748927e-05,
+      "loss": 0.294,
+      "step": 22813
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961707733754105e-05,
+      "loss": 0.3046,
+      "step": 22814
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9617032308737475e-05,
+      "loss": 0.2862,
+      "step": 22815
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9616987277338214e-05,
+      "loss": 0.2539,
+      "step": 22816
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.961694224334329e-05,
+      "loss": 0.2979,
+      "step": 22817
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9616897206752706e-05,
+      "loss": 0.2601,
+      "step": 22818
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616852167566472e-05,
+      "loss": 0.2845,
+      "step": 22819
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961680712578461e-05,
+      "loss": 0.2828,
+      "step": 22820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616762081407125e-05,
+      "loss": 0.2887,
+      "step": 22821
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616717034434032e-05,
+      "loss": 0.2849,
+      "step": 22822
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616671984865342e-05,
+      "loss": 0.3018,
+      "step": 22823
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616626932701066e-05,
+      "loss": 0.2887,
+      "step": 22824
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961658187794122e-05,
+      "loss": 0.3046,
+      "step": 22825
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961653682058581e-05,
+      "loss": 0.2813,
+      "step": 22826
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961649176063485e-05,
+      "loss": 0.2783,
+      "step": 22827
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961644669808836e-05,
+      "loss": 0.2422,
+      "step": 22828
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616401632946346e-05,
+      "loss": 0.3187,
+      "step": 22829
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616356565208816e-05,
+      "loss": 0.3147,
+      "step": 22830
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616311494875788e-05,
+      "loss": 0.2704,
+      "step": 22831
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961626642194727e-05,
+      "loss": 0.317,
+      "step": 22832
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616221346423283e-05,
+      "loss": 0.2797,
+      "step": 22833
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616176268303827e-05,
+      "loss": 0.305,
+      "step": 22834
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616131187588926e-05,
+      "loss": 0.322,
+      "step": 22835
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9616086104278584e-05,
+      "loss": 0.2529,
+      "step": 22836
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961604101837281e-05,
+      "loss": 0.3503,
+      "step": 22837
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615995929871627e-05,
+      "loss": 0.3179,
+      "step": 22838
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961595083877504e-05,
+      "loss": 0.3593,
+      "step": 22839
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615905745083062e-05,
+      "loss": 0.3124,
+      "step": 22840
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615860648795708e-05,
+      "loss": 0.2911,
+      "step": 22841
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961581554991299e-05,
+      "loss": 0.2504,
+      "step": 22842
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615770448434913e-05,
+      "loss": 0.3432,
+      "step": 22843
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.96157253443615e-05,
+      "loss": 0.2999,
+      "step": 22844
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615680237692754e-05,
+      "loss": 0.3043,
+      "step": 22845
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615635128428692e-05,
+      "loss": 0.3071,
+      "step": 22846
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615590016569328e-05,
+      "loss": 0.3207,
+      "step": 22847
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615544902114665e-05,
+      "loss": 0.3625,
+      "step": 22848
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615499785064728e-05,
+      "loss": 0.3263,
+      "step": 22849
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961545466541952e-05,
+      "loss": 0.3603,
+      "step": 22850
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615409543179058e-05,
+      "loss": 0.3457,
+      "step": 22851
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961536441834335e-05,
+      "loss": 0.2895,
+      "step": 22852
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961531929091241e-05,
+      "loss": 0.3334,
+      "step": 22853
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961527416088625e-05,
+      "loss": 0.2983,
+      "step": 22854
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961522902826488e-05,
+      "loss": 0.3296,
+      "step": 22855
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961518389304832e-05,
+      "loss": 0.3153,
+      "step": 22856
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615138755236573e-05,
+      "loss": 0.2813,
+      "step": 22857
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615093614829658e-05,
+      "loss": 0.3061,
+      "step": 22858
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9615048471827586e-05,
+      "loss": 0.2932,
+      "step": 22859
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961500332623036e-05,
+      "loss": 0.2967,
+      "step": 22860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614958178038007e-05,
+      "loss": 0.3118,
+      "step": 22861
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614913027250527e-05,
+      "loss": 0.3217,
+      "step": 22862
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961486787386794e-05,
+      "loss": 0.2748,
+      "step": 22863
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614822717890253e-05,
+      "loss": 0.3068,
+      "step": 22864
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614777559317483e-05,
+      "loss": 0.2774,
+      "step": 22865
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614732398149637e-05,
+      "loss": 0.3263,
+      "step": 22866
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961468723438673e-05,
+      "loss": 0.2923,
+      "step": 22867
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961464206802878e-05,
+      "loss": 0.2665,
+      "step": 22868
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614596899075782e-05,
+      "loss": 0.3339,
+      "step": 22869
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614551727527772e-05,
+      "loss": 0.3108,
+      "step": 22870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961450655338474e-05,
+      "loss": 0.3188,
+      "step": 22871
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961446137664671e-05,
+      "loss": 0.2729,
+      "step": 22872
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614416197313693e-05,
+      "loss": 0.3277,
+      "step": 22873
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.96143710153857e-05,
+      "loss": 0.3094,
+      "step": 22874
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614325830862744e-05,
+      "loss": 0.307,
+      "step": 22875
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614280643744835e-05,
+      "loss": 0.3244,
+      "step": 22876
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614235454031986e-05,
+      "loss": 0.3008,
+      "step": 22877
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614190261724214e-05,
+      "loss": 0.332,
+      "step": 22878
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614145066821523e-05,
+      "loss": 0.2971,
+      "step": 22879
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961409986932393e-05,
+      "loss": 0.2848,
+      "step": 22880
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614054669231448e-05,
+      "loss": 0.3577,
+      "step": 22881
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9614009466544087e-05,
+      "loss": 0.2864,
+      "step": 22882
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961396426126186e-05,
+      "loss": 0.3528,
+      "step": 22883
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961391905338478e-05,
+      "loss": 0.3204,
+      "step": 22884
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961387384291286e-05,
+      "loss": 0.2825,
+      "step": 22885
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613828629846105e-05,
+      "loss": 0.2641,
+      "step": 22886
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613783414184534e-05,
+      "loss": 0.3151,
+      "step": 22887
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961373819592816e-05,
+      "loss": 0.3135,
+      "step": 22888
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613692975076993e-05,
+      "loss": 0.2672,
+      "step": 22889
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613647751631047e-05,
+      "loss": 0.2912,
+      "step": 22890
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961360252559033e-05,
+      "loss": 0.2875,
+      "step": 22891
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613557296954858e-05,
+      "loss": 0.2931,
+      "step": 22892
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613512065724638e-05,
+      "loss": 0.2854,
+      "step": 22893
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961346683189969e-05,
+      "loss": 0.2657,
+      "step": 22894
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613421595480024e-05,
+      "loss": 0.2854,
+      "step": 22895
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961337635646565e-05,
+      "loss": 0.293,
+      "step": 22896
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613331114856576e-05,
+      "loss": 0.2829,
+      "step": 22897
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613285870652823e-05,
+      "loss": 0.2793,
+      "step": 22898
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613240623854396e-05,
+      "loss": 0.258,
+      "step": 22899
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613195374461313e-05,
+      "loss": 0.2702,
+      "step": 22900
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613150122473582e-05,
+      "loss": 0.2537,
+      "step": 22901
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961310486789122e-05,
+      "loss": 0.3101,
+      "step": 22902
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613059610714233e-05,
+      "loss": 0.3285,
+      "step": 22903
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9613014350942638e-05,
+      "loss": 0.2732,
+      "step": 22904
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9612969088576446e-05,
+      "loss": 0.2947,
+      "step": 22905
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9612923823615667e-05,
+      "loss": 0.2633,
+      "step": 22906
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9612878556060314e-05,
+      "loss": 0.3154,
+      "step": 22907
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.96128332859104e-05,
+      "loss": 0.3471,
+      "step": 22908
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961278801316594e-05,
+      "loss": 0.2409,
+      "step": 22909
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9612742737826945e-05,
+      "loss": 0.2915,
+      "step": 22910
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9612697459893423e-05,
+      "loss": 0.2962,
+      "step": 22911
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961265217936539e-05,
+      "loss": 0.2631,
+      "step": 22912
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9612606896242857e-05,
+      "loss": 0.2996,
+      "step": 22913
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9612561610525837e-05,
+      "loss": 0.2896,
+      "step": 22914
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9612516322214338e-05,
+      "loss": 0.3564,
+      "step": 22915
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961247103130838e-05,
+      "loss": 0.3189,
+      "step": 22916
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.961242573780797e-05,
+      "loss": 0.2792,
+      "step": 22917
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961238044171312e-05,
+      "loss": 0.35,
+      "step": 22918
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9612335143023845e-05,
+      "loss": 0.2595,
+      "step": 22919
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9612289841740155e-05,
+      "loss": 0.3021,
+      "step": 22920
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9612244537862068e-05,
+      "loss": 0.2724,
+      "step": 22921
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9612199231389586e-05,
+      "loss": 0.3067,
+      "step": 22922
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9612153922322728e-05,
+      "loss": 0.3214,
+      "step": 22923
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9612108610661502e-05,
+      "loss": 0.2957,
+      "step": 22924
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9612063296405926e-05,
+      "loss": 0.3026,
+      "step": 22925
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961201797955601e-05,
+      "loss": 0.3417,
+      "step": 22926
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961197266011176e-05,
+      "loss": 0.2586,
+      "step": 22927
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.96119273380732e-05,
+      "loss": 0.2737,
+      "step": 22928
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961188201344033e-05,
+      "loss": 0.2977,
+      "step": 22929
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611836686213174e-05,
+      "loss": 0.314,
+      "step": 22930
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611791356391735e-05,
+      "loss": 0.3042,
+      "step": 22931
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961174602397603e-05,
+      "loss": 0.3811,
+      "step": 22932
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961170068896607e-05,
+      "loss": 0.2924,
+      "step": 22933
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611655351361864e-05,
+      "loss": 0.3148,
+      "step": 22934
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611610011163428e-05,
+      "loss": 0.2848,
+      "step": 22935
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611564668370777e-05,
+      "loss": 0.3079,
+      "step": 22936
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611519322983917e-05,
+      "loss": 0.2634,
+      "step": 22937
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611473975002868e-05,
+      "loss": 0.3257,
+      "step": 22938
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961142862442763e-05,
+      "loss": 0.3362,
+      "step": 22939
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611383271258226e-05,
+      "loss": 0.2792,
+      "step": 22940
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611337915494664e-05,
+      "loss": 0.3343,
+      "step": 22941
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611292557136956e-05,
+      "loss": 0.3044,
+      "step": 22942
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611247196185117e-05,
+      "loss": 0.3127,
+      "step": 22943
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611201832639156e-05,
+      "loss": 0.3029,
+      "step": 22944
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961115646649909e-05,
+      "loss": 0.3071,
+      "step": 22945
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961111109776492e-05,
+      "loss": 0.347,
+      "step": 22946
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611065726436676e-05,
+      "loss": 0.3175,
+      "step": 22947
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9611020352514357e-05,
+      "loss": 0.2911,
+      "step": 22948
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610974975997977e-05,
+      "loss": 0.2767,
+      "step": 22949
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961092959688755e-05,
+      "loss": 0.2688,
+      "step": 22950
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961088421518309e-05,
+      "loss": 0.2887,
+      "step": 22951
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610838830884606e-05,
+      "loss": 0.2715,
+      "step": 22952
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610793443992112e-05,
+      "loss": 0.2639,
+      "step": 22953
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610748054505618e-05,
+      "loss": 0.3097,
+      "step": 22954
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610702662425143e-05,
+      "loss": 0.2996,
+      "step": 22955
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961065726775069e-05,
+      "loss": 0.2535,
+      "step": 22956
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610611870482277e-05,
+      "loss": 0.2735,
+      "step": 22957
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610566470619913e-05,
+      "loss": 0.3104,
+      "step": 22958
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610521068163617e-05,
+      "loss": 0.2694,
+      "step": 22959
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610475663113393e-05,
+      "loss": 0.2855,
+      "step": 22960
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610430255469257e-05,
+      "loss": 0.3059,
+      "step": 22961
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.961038484523122e-05,
+      "loss": 0.2709,
+      "step": 22962
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.96103394323993e-05,
+      "loss": 0.308,
+      "step": 22963
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.96102940169735e-05,
+      "loss": 0.3145,
+      "step": 22964
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610248598953838e-05,
+      "loss": 0.2743,
+      "step": 22965
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610203178340326e-05,
+      "loss": 0.2933,
+      "step": 22966
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610157755132972e-05,
+      "loss": 0.2879,
+      "step": 22967
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610112329331794e-05,
+      "loss": 0.284,
+      "step": 22968
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.96100669009368e-05,
+      "loss": 0.2504,
+      "step": 22969
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9610021469948005e-05,
+      "loss": 0.2911,
+      "step": 22970
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609976036365423e-05,
+      "loss": 0.3072,
+      "step": 22971
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960993060018906e-05,
+      "loss": 0.2497,
+      "step": 22972
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609885161418932e-05,
+      "loss": 0.2417,
+      "step": 22973
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960983972005505e-05,
+      "loss": 0.3018,
+      "step": 22974
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609794276097433e-05,
+      "loss": 0.3386,
+      "step": 22975
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609748829546085e-05,
+      "loss": 0.3023,
+      "step": 22976
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960970338040102e-05,
+      "loss": 0.3418,
+      "step": 22977
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960965792866225e-05,
+      "loss": 0.2759,
+      "step": 22978
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960961247432979e-05,
+      "loss": 0.2757,
+      "step": 22979
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960956701740365e-05,
+      "loss": 0.3528,
+      "step": 22980
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960952155788384e-05,
+      "loss": 0.2999,
+      "step": 22981
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960947609577038e-05,
+      "loss": 0.2952,
+      "step": 22982
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609430631063274e-05,
+      "loss": 0.2892,
+      "step": 22983
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960938516376254e-05,
+      "loss": 0.2702,
+      "step": 22984
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609339693868186e-05,
+      "loss": 0.2589,
+      "step": 22985
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609294221380232e-05,
+      "loss": 0.3178,
+      "step": 22986
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960924874629868e-05,
+      "loss": 0.2982,
+      "step": 22987
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609203268623543e-05,
+      "loss": 0.3021,
+      "step": 22988
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609157788354846e-05,
+      "loss": 0.3424,
+      "step": 22989
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609112305492584e-05,
+      "loss": 0.2908,
+      "step": 22990
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9609066820036783e-05,
+      "loss": 0.289,
+      "step": 22991
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960902133198745e-05,
+      "loss": 0.2548,
+      "step": 22992
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608975841344592e-05,
+      "loss": 0.293,
+      "step": 22993
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608930348108234e-05,
+      "loss": 0.2973,
+      "step": 22994
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608884852278372e-05,
+      "loss": 0.36,
+      "step": 22995
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608839353855035e-05,
+      "loss": 0.3338,
+      "step": 22996
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608793852838225e-05,
+      "loss": 0.3395,
+      "step": 22997
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608748349227956e-05,
+      "loss": 0.2848,
+      "step": 22998
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608702843024238e-05,
+      "loss": 0.3077,
+      "step": 22999
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960865733422709e-05,
+      "loss": 0.3185,
+      "step": 23000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960861182283652e-05,
+      "loss": 0.2905,
+      "step": 23001
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960856630885254e-05,
+      "loss": 0.3095,
+      "step": 23002
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608520792275162e-05,
+      "loss": 0.3099,
+      "step": 23003
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.96084752731044e-05,
+      "loss": 0.305,
+      "step": 23004
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608429751340267e-05,
+      "loss": 0.2966,
+      "step": 23005
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608384226982773e-05,
+      "loss": 0.3333,
+      "step": 23006
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960833870003193e-05,
+      "loss": 0.2871,
+      "step": 23007
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608293170487753e-05,
+      "loss": 0.298,
+      "step": 23008
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960824763835025e-05,
+      "loss": 0.2942,
+      "step": 23009
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608202103619443e-05,
+      "loss": 0.319,
+      "step": 23010
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608156566295335e-05,
+      "loss": 0.2951,
+      "step": 23011
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608111026377937e-05,
+      "loss": 0.248,
+      "step": 23012
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608065483867267e-05,
+      "loss": 0.3236,
+      "step": 23013
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9608019938763334e-05,
+      "loss": 0.3143,
+      "step": 23014
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960797439106615e-05,
+      "loss": 0.3346,
+      "step": 23015
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.960792884077573e-05,
+      "loss": 0.3159,
+      "step": 23016
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607883287892088e-05,
+      "loss": 0.2991,
+      "step": 23017
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960783773241523e-05,
+      "loss": 0.3167,
+      "step": 23018
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960779217434517e-05,
+      "loss": 0.24,
+      "step": 23019
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960774661368193e-05,
+      "loss": 0.3222,
+      "step": 23020
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960770105042551e-05,
+      "loss": 0.307,
+      "step": 23021
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607655484575923e-05,
+      "loss": 0.304,
+      "step": 23022
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607609916133187e-05,
+      "loss": 0.2786,
+      "step": 23023
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607564345097313e-05,
+      "loss": 0.2733,
+      "step": 23024
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607518771468316e-05,
+      "loss": 0.2433,
+      "step": 23025
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607473195246198e-05,
+      "loss": 0.338,
+      "step": 23026
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607427616430984e-05,
+      "loss": 0.3104,
+      "step": 23027
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607382035022677e-05,
+      "loss": 0.2763,
+      "step": 23028
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607336451021293e-05,
+      "loss": 0.2945,
+      "step": 23029
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607290864426844e-05,
+      "loss": 0.2958,
+      "step": 23030
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607245275239345e-05,
+      "loss": 0.2595,
+      "step": 23031
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.96071996834588e-05,
+      "loss": 0.278,
+      "step": 23032
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607154089085232e-05,
+      "loss": 0.307,
+      "step": 23033
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607108492118644e-05,
+      "loss": 0.3118,
+      "step": 23034
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607062892559057e-05,
+      "loss": 0.3486,
+      "step": 23035
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9607017290406476e-05,
+      "loss": 0.2898,
+      "step": 23036
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606971685660917e-05,
+      "loss": 0.283,
+      "step": 23037
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960692607832239e-05,
+      "loss": 0.3034,
+      "step": 23038
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606880468390913e-05,
+      "loss": 0.3023,
+      "step": 23039
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606834855866488e-05,
+      "loss": 0.2997,
+      "step": 23040
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960678924074914e-05,
+      "loss": 0.2787,
+      "step": 23041
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606743623038866e-05,
+      "loss": 0.2871,
+      "step": 23042
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606698002735694e-05,
+      "loss": 0.3026,
+      "step": 23043
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606652379839625e-05,
+      "loss": 0.2962,
+      "step": 23044
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960660675435068e-05,
+      "loss": 0.3048,
+      "step": 23045
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606561126268865e-05,
+      "loss": 0.3377,
+      "step": 23046
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606515495594194e-05,
+      "loss": 0.3003,
+      "step": 23047
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606469862326677e-05,
+      "loss": 0.3108,
+      "step": 23048
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606424226466335e-05,
+      "loss": 0.3379,
+      "step": 23049
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960637858801317e-05,
+      "loss": 0.314,
+      "step": 23050
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.96063329469672e-05,
+      "loss": 0.3172,
+      "step": 23051
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606287303328434e-05,
+      "loss": 0.2872,
+      "step": 23052
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606241657096886e-05,
+      "loss": 0.3165,
+      "step": 23053
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606196008272572e-05,
+      "loss": 0.2974,
+      "step": 23054
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606150356855498e-05,
+      "loss": 0.2616,
+      "step": 23055
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960610470284568e-05,
+      "loss": 0.2919,
+      "step": 23056
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606059046243126e-05,
+      "loss": 0.299,
+      "step": 23057
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9606013387047857e-05,
+      "loss": 0.3144,
+      "step": 23058
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605967725259877e-05,
+      "loss": 0.3152,
+      "step": 23059
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605922060879204e-05,
+      "loss": 0.3183,
+      "step": 23060
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605876393905844e-05,
+      "loss": 0.3643,
+      "step": 23061
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605830724339815e-05,
+      "loss": 0.3101,
+      "step": 23062
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605785052181126e-05,
+      "loss": 0.3023,
+      "step": 23063
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605739377429795e-05,
+      "loss": 0.3298,
+      "step": 23064
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605693700085825e-05,
+      "loss": 0.2807,
+      "step": 23065
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960564802014924e-05,
+      "loss": 0.292,
+      "step": 23066
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605602337620038e-05,
+      "loss": 0.2892,
+      "step": 23067
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960555665249824e-05,
+      "loss": 0.266,
+      "step": 23068
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605510964783864e-05,
+      "loss": 0.2768,
+      "step": 23069
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605465274476908e-05,
+      "loss": 0.3213,
+      "step": 23070
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605419581577395e-05,
+      "loss": 0.3083,
+      "step": 23071
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605373886085337e-05,
+      "loss": 0.2815,
+      "step": 23072
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960532818800074e-05,
+      "loss": 0.2692,
+      "step": 23073
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605282487323626e-05,
+      "loss": 0.2606,
+      "step": 23074
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605236784053993e-05,
+      "loss": 0.2982,
+      "step": 23075
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605191078191866e-05,
+      "loss": 0.2994,
+      "step": 23076
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605145369737254e-05,
+      "loss": 0.311,
+      "step": 23077
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605099658690164e-05,
+      "loss": 0.3646,
+      "step": 23078
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9605053945050615e-05,
+      "loss": 0.2988,
+      "step": 23079
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960500822881862e-05,
+      "loss": 0.3185,
+      "step": 23080
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604962509994186e-05,
+      "loss": 0.2861,
+      "step": 23081
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960491678857733e-05,
+      "loss": 0.337,
+      "step": 23082
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604871064568058e-05,
+      "loss": 0.3102,
+      "step": 23083
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604825337966387e-05,
+      "loss": 0.2905,
+      "step": 23084
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604779608772332e-05,
+      "loss": 0.292,
+      "step": 23085
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604733876985898e-05,
+      "loss": 0.2935,
+      "step": 23086
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604688142607103e-05,
+      "loss": 0.2844,
+      "step": 23087
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604642405635957e-05,
+      "loss": 0.2716,
+      "step": 23088
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604596666072477e-05,
+      "loss": 0.3206,
+      "step": 23089
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960455092391667e-05,
+      "loss": 0.3156,
+      "step": 23090
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604505179168548e-05,
+      "loss": 0.3059,
+      "step": 23091
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604459431828123e-05,
+      "loss": 0.2725,
+      "step": 23092
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604413681895414e-05,
+      "loss": 0.3096,
+      "step": 23093
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960436792937043e-05,
+      "loss": 0.3095,
+      "step": 23094
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604322174253178e-05,
+      "loss": 0.2951,
+      "step": 23095
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604276416543674e-05,
+      "loss": 0.3497,
+      "step": 23096
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604230656241933e-05,
+      "loss": 0.3075,
+      "step": 23097
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604184893347967e-05,
+      "loss": 0.3067,
+      "step": 23098
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604139127861783e-05,
+      "loss": 0.3013,
+      "step": 23099
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.96040933597834e-05,
+      "loss": 0.2818,
+      "step": 23100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604047589112824e-05,
+      "loss": 0.2643,
+      "step": 23101
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9604001815850073e-05,
+      "loss": 0.3582,
+      "step": 23102
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9603956039995155e-05,
+      "loss": 0.3003,
+      "step": 23103
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9603910261548088e-05,
+      "loss": 0.3231,
+      "step": 23104
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9603864480508877e-05,
+      "loss": 0.2776,
+      "step": 23105
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9603818696877543e-05,
+      "loss": 0.2744,
+      "step": 23106
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960377291065409e-05,
+      "loss": 0.3437,
+      "step": 23107
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9603727121838534e-05,
+      "loss": 0.3173,
+      "step": 23108
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9603681330430885e-05,
+      "loss": 0.2688,
+      "step": 23109
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9603635536431162e-05,
+      "loss": 0.2918,
+      "step": 23110
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960358973983937e-05,
+      "loss": 0.2582,
+      "step": 23111
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9603543940655527e-05,
+      "loss": 0.3458,
+      "step": 23112
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9603498138879638e-05,
+      "loss": 0.3043,
+      "step": 23113
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.960345233451172e-05,
+      "loss": 0.2902,
+      "step": 23114
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960340652755179e-05,
+      "loss": 0.2648,
+      "step": 23115
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960336071799985e-05,
+      "loss": 0.2831,
+      "step": 23116
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9603314905855924e-05,
+      "loss": 0.2966,
+      "step": 23117
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9603269091120018e-05,
+      "loss": 0.2928,
+      "step": 23118
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960322327379214e-05,
+      "loss": 0.2723,
+      "step": 23119
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960317745387231e-05,
+      "loss": 0.2936,
+      "step": 23120
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9603131631360538e-05,
+      "loss": 0.2952,
+      "step": 23121
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9603085806256833e-05,
+      "loss": 0.305,
+      "step": 23122
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9603039978561215e-05,
+      "loss": 0.2721,
+      "step": 23123
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602994148273685e-05,
+      "loss": 0.274,
+      "step": 23124
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960294831539427e-05,
+      "loss": 0.2917,
+      "step": 23125
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960290247992297e-05,
+      "loss": 0.3203,
+      "step": 23126
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.96028566418598e-05,
+      "loss": 0.3063,
+      "step": 23127
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602810801204777e-05,
+      "loss": 0.33,
+      "step": 23128
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960276495795791e-05,
+      "loss": 0.2848,
+      "step": 23129
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602719112119213e-05,
+      "loss": 0.2904,
+      "step": 23130
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602673263688692e-05,
+      "loss": 0.2637,
+      "step": 23131
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960262741266637e-05,
+      "loss": 0.3121,
+      "step": 23132
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602581559052254e-05,
+      "loss": 0.25,
+      "step": 23133
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602535702846354e-05,
+      "loss": 0.2958,
+      "step": 23134
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602489844048685e-05,
+      "loss": 0.2304,
+      "step": 23135
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960244398265926e-05,
+      "loss": 0.3354,
+      "step": 23136
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960239811867809e-05,
+      "loss": 0.3105,
+      "step": 23137
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602352252105187e-05,
+      "loss": 0.3161,
+      "step": 23138
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602306382940567e-05,
+      "loss": 0.3219,
+      "step": 23139
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602260511184236e-05,
+      "loss": 0.2834,
+      "step": 23140
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960221463683621e-05,
+      "loss": 0.3141,
+      "step": 23141
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602168759896506e-05,
+      "loss": 0.2945,
+      "step": 23142
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602122880365127e-05,
+      "loss": 0.3495,
+      "step": 23143
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602076998242095e-05,
+      "loss": 0.3199,
+      "step": 23144
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9602031113527414e-05,
+      "loss": 0.3062,
+      "step": 23145
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.96019852262211e-05,
+      "loss": 0.2714,
+      "step": 23146
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601939336323162e-05,
+      "loss": 0.3165,
+      "step": 23147
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601893443833623e-05,
+      "loss": 0.2917,
+      "step": 23148
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601847548752485e-05,
+      "loss": 0.3285,
+      "step": 23149
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960180165107976e-05,
+      "loss": 0.304,
+      "step": 23150
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960175575081547e-05,
+      "loss": 0.2846,
+      "step": 23151
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601709847959615e-05,
+      "loss": 0.295,
+      "step": 23152
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960166394251222e-05,
+      "loss": 0.3274,
+      "step": 23153
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601618034473287e-05,
+      "loss": 0.28,
+      "step": 23154
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601572123842833e-05,
+      "loss": 0.2974,
+      "step": 23155
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601526210620868e-05,
+      "loss": 0.2712,
+      "step": 23156
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960148029480741e-05,
+      "loss": 0.3147,
+      "step": 23157
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601434376402463e-05,
+      "loss": 0.2547,
+      "step": 23158
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960138845540605e-05,
+      "loss": 0.3176,
+      "step": 23159
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601342531818174e-05,
+      "loss": 0.2746,
+      "step": 23160
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601296605638848e-05,
+      "loss": 0.2979,
+      "step": 23161
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601250676868092e-05,
+      "loss": 0.3029,
+      "step": 23162
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601204745505914e-05,
+      "loss": 0.2766,
+      "step": 23163
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601158811552322e-05,
+      "loss": 0.3165,
+      "step": 23164
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601112875007332e-05,
+      "loss": 0.3764,
+      "step": 23165
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960106693587096e-05,
+      "loss": 0.2757,
+      "step": 23166
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9601020994143213e-05,
+      "loss": 0.3347,
+      "step": 23167
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600975049824107e-05,
+      "loss": 0.2806,
+      "step": 23168
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600929102913653e-05,
+      "loss": 0.287,
+      "step": 23169
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960088315341186e-05,
+      "loss": 0.2636,
+      "step": 23170
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600837201318747e-05,
+      "loss": 0.3054,
+      "step": 23171
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600791246634326e-05,
+      "loss": 0.2974,
+      "step": 23172
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.96007452893586e-05,
+      "loss": 0.3422,
+      "step": 23173
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600699329491592e-05,
+      "loss": 0.296,
+      "step": 23174
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960065336703331e-05,
+      "loss": 0.2443,
+      "step": 23175
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600607401983767e-05,
+      "loss": 0.2813,
+      "step": 23176
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600561434342974e-05,
+      "loss": 0.2568,
+      "step": 23177
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600515464110942e-05,
+      "loss": 0.3365,
+      "step": 23178
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600469491287693e-05,
+      "loss": 0.3506,
+      "step": 23179
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600423515873225e-05,
+      "loss": 0.2697,
+      "step": 23180
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600377537867565e-05,
+      "loss": 0.3429,
+      "step": 23181
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600331557270712e-05,
+      "loss": 0.2708,
+      "step": 23182
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960028557408269e-05,
+      "loss": 0.3242,
+      "step": 23183
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.96002395883035e-05,
+      "loss": 0.3144,
+      "step": 23184
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600193599933165e-05,
+      "loss": 0.3151,
+      "step": 23185
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600147608971692e-05,
+      "loss": 0.3377,
+      "step": 23186
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600101615419094e-05,
+      "loss": 0.3386,
+      "step": 23187
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600055619275382e-05,
+      "loss": 0.2866,
+      "step": 23188
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9600009620540572e-05,
+      "loss": 0.3774,
+      "step": 23189
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599963619214675e-05,
+      "loss": 0.2704,
+      "step": 23190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.95999176152977e-05,
+      "loss": 0.2635,
+      "step": 23191
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599871608789666e-05,
+      "loss": 0.2965,
+      "step": 23192
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599825599690582e-05,
+      "loss": 0.2597,
+      "step": 23193
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599779588000457e-05,
+      "loss": 0.3247,
+      "step": 23194
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599733573719307e-05,
+      "loss": 0.3113,
+      "step": 23195
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599687556847143e-05,
+      "loss": 0.3305,
+      "step": 23196
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599641537383983e-05,
+      "loss": 0.2524,
+      "step": 23197
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.959959551532983e-05,
+      "loss": 0.3153,
+      "step": 23198
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599549490684707e-05,
+      "loss": 0.3026,
+      "step": 23199
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599503463448618e-05,
+      "loss": 0.2948,
+      "step": 23200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599457433621577e-05,
+      "loss": 0.3409,
+      "step": 23201
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599411401203597e-05,
+      "loss": 0.266,
+      "step": 23202
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599365366194693e-05,
+      "loss": 0.2531,
+      "step": 23203
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.959931932859487e-05,
+      "loss": 0.3073,
+      "step": 23204
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599273288404154e-05,
+      "loss": 0.3215,
+      "step": 23205
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599227245622544e-05,
+      "loss": 0.2765,
+      "step": 23206
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599181200250063e-05,
+      "loss": 0.2679,
+      "step": 23207
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.959913515228671e-05,
+      "loss": 0.2919,
+      "step": 23208
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599089101732513e-05,
+      "loss": 0.2801,
+      "step": 23209
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9599043048587474e-05,
+      "loss": 0.302,
+      "step": 23210
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9598996992851606e-05,
+      "loss": 0.2859,
+      "step": 23211
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.959895093452493e-05,
+      "loss": 0.3598,
+      "step": 23212
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9598904873607448e-05,
+      "loss": 0.268,
+      "step": 23213
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598858810099175e-05,
+      "loss": 0.3529,
+      "step": 23214
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959881274400013e-05,
+      "loss": 0.2763,
+      "step": 23215
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598766675310315e-05,
+      "loss": 0.3379,
+      "step": 23216
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959872060402975e-05,
+      "loss": 0.3532,
+      "step": 23217
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959867453015845e-05,
+      "loss": 0.2955,
+      "step": 23218
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959862845369642e-05,
+      "loss": 0.2909,
+      "step": 23219
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598582374643672e-05,
+      "loss": 0.2927,
+      "step": 23220
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598536293000224e-05,
+      "loss": 0.2809,
+      "step": 23221
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598490208766088e-05,
+      "loss": 0.2851,
+      "step": 23222
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598444121941274e-05,
+      "loss": 0.3275,
+      "step": 23223
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598398032525794e-05,
+      "loss": 0.2804,
+      "step": 23224
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959835194051966e-05,
+      "loss": 0.2891,
+      "step": 23225
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598305845922886e-05,
+      "loss": 0.3112,
+      "step": 23226
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598259748735485e-05,
+      "loss": 0.2678,
+      "step": 23227
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598213648957474e-05,
+      "loss": 0.3887,
+      "step": 23228
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9598167546588854e-05,
+      "loss": 0.2877,
+      "step": 23229
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959812144162965e-05,
+      "loss": 0.3079,
+      "step": 23230
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959807533407986e-05,
+      "loss": 0.362,
+      "step": 23231
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959802922393951e-05,
+      "loss": 0.2356,
+      "step": 23232
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597983111208604e-05,
+      "loss": 0.257,
+      "step": 23233
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959793699588716e-05,
+      "loss": 0.253,
+      "step": 23234
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597890877975186e-05,
+      "loss": 0.3075,
+      "step": 23235
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597844757472695e-05,
+      "loss": 0.2583,
+      "step": 23236
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597798634379707e-05,
+      "loss": 0.2792,
+      "step": 23237
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597752508696224e-05,
+      "loss": 0.2969,
+      "step": 23238
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597706380422263e-05,
+      "loss": 0.2648,
+      "step": 23239
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597660249557835e-05,
+      "loss": 0.3236,
+      "step": 23240
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597614116102957e-05,
+      "loss": 0.281,
+      "step": 23241
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597567980057635e-05,
+      "loss": 0.3084,
+      "step": 23242
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597521841421887e-05,
+      "loss": 0.3052,
+      "step": 23243
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597475700195722e-05,
+      "loss": 0.2823,
+      "step": 23244
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597429556379154e-05,
+      "loss": 0.2742,
+      "step": 23245
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597383409972193e-05,
+      "loss": 0.3657,
+      "step": 23246
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597337260974853e-05,
+      "loss": 0.364,
+      "step": 23247
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959729110938715e-05,
+      "loss": 0.2701,
+      "step": 23248
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959724495520909e-05,
+      "loss": 0.2754,
+      "step": 23249
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959719879844069e-05,
+      "loss": 0.2988,
+      "step": 23250
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597152639081962e-05,
+      "loss": 0.3202,
+      "step": 23251
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959710647713292e-05,
+      "loss": 0.3075,
+      "step": 23252
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959706031259357e-05,
+      "loss": 0.2664,
+      "step": 23253
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959701414546393e-05,
+      "loss": 0.3216,
+      "step": 23254
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959696797574401e-05,
+      "loss": 0.3244,
+      "step": 23255
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596921803433823e-05,
+      "loss": 0.2806,
+      "step": 23256
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596875628533386e-05,
+      "loss": 0.3403,
+      "step": 23257
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.95968294510427e-05,
+      "loss": 0.2768,
+      "step": 23258
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959678327096179e-05,
+      "loss": 0.3197,
+      "step": 23259
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959673708829066e-05,
+      "loss": 0.2929,
+      "step": 23260
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959669090302933e-05,
+      "loss": 0.415,
+      "step": 23261
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959664471517781e-05,
+      "loss": 0.3284,
+      "step": 23262
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596598524736107e-05,
+      "loss": 0.2879,
+      "step": 23263
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596552331704238e-05,
+      "loss": 0.3374,
+      "step": 23264
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596506136082214e-05,
+      "loss": 0.2681,
+      "step": 23265
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959645993787005e-05,
+      "loss": 0.3039,
+      "step": 23266
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596413737067755e-05,
+      "loss": 0.3429,
+      "step": 23267
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596367533675342e-05,
+      "loss": 0.3297,
+      "step": 23268
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596321327692827e-05,
+      "loss": 0.3073,
+      "step": 23269
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596275119120218e-05,
+      "loss": 0.3196,
+      "step": 23270
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596228907957527e-05,
+      "loss": 0.28,
+      "step": 23271
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596182694204776e-05,
+      "loss": 0.3243,
+      "step": 23272
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596136477861967e-05,
+      "loss": 0.3273,
+      "step": 23273
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596090258929115e-05,
+      "loss": 0.372,
+      "step": 23274
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9596044037406236e-05,
+      "loss": 0.2703,
+      "step": 23275
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595997813293336e-05,
+      "loss": 0.2851,
+      "step": 23276
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959595158659043e-05,
+      "loss": 0.3007,
+      "step": 23277
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959590535729754e-05,
+      "loss": 0.2831,
+      "step": 23278
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595859125414664e-05,
+      "loss": 0.2758,
+      "step": 23279
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959581289094182e-05,
+      "loss": 0.2972,
+      "step": 23280
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595766653879023e-05,
+      "loss": 0.3542,
+      "step": 23281
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595720414226285e-05,
+      "loss": 0.2655,
+      "step": 23282
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595674171983618e-05,
+      "loss": 0.2959,
+      "step": 23283
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595627927151032e-05,
+      "loss": 0.3022,
+      "step": 23284
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959558167972854e-05,
+      "loss": 0.306,
+      "step": 23285
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595535429716157e-05,
+      "loss": 0.3037,
+      "step": 23286
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595489177113894e-05,
+      "loss": 0.3044,
+      "step": 23287
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959544292192176e-05,
+      "loss": 0.2936,
+      "step": 23288
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959539666413978e-05,
+      "loss": 0.328,
+      "step": 23289
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595350403767947e-05,
+      "loss": 0.3299,
+      "step": 23290
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959530414080629e-05,
+      "loss": 0.3087,
+      "step": 23291
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595257875254815e-05,
+      "loss": 0.3214,
+      "step": 23292
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595211607113537e-05,
+      "loss": 0.289,
+      "step": 23293
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595165336382465e-05,
+      "loss": 0.2989,
+      "step": 23294
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959511906306161e-05,
+      "loss": 0.3089,
+      "step": 23295
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959507278715099e-05,
+      "loss": 0.3158,
+      "step": 23296
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595026508650616e-05,
+      "loss": 0.2889,
+      "step": 23297
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594980227560497e-05,
+      "loss": 0.2606,
+      "step": 23298
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594933943880653e-05,
+      "loss": 0.2419,
+      "step": 23299
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594887657611083e-05,
+      "loss": 0.2854,
+      "step": 23300
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594841368751812e-05,
+      "loss": 0.317,
+      "step": 23301
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594795077302853e-05,
+      "loss": 0.2685,
+      "step": 23302
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959474878326421e-05,
+      "loss": 0.2819,
+      "step": 23303
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594702486635898e-05,
+      "loss": 0.2698,
+      "step": 23304
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594656187417932e-05,
+      "loss": 0.2944,
+      "step": 23305
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594609885610326e-05,
+      "loss": 0.3228,
+      "step": 23306
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959456358121309e-05,
+      "loss": 0.3162,
+      "step": 23307
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594517274226232e-05,
+      "loss": 0.2985,
+      "step": 23308
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594470964649772e-05,
+      "loss": 0.306,
+      "step": 23309
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959442465248372e-05,
+      "loss": 0.324,
+      "step": 23310
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9594378337728085e-05,
+      "loss": 0.2749,
+      "step": 23311
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9594332020382886e-05,
+      "loss": 0.2521,
+      "step": 23312
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9594285700448128e-05,
+      "loss": 0.3267,
+      "step": 23313
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959423937792383e-05,
+      "loss": 0.31,
+      "step": 23314
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959419305281e-05,
+      "loss": 0.3219,
+      "step": 23315
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9594146725106656e-05,
+      "loss": 0.3373,
+      "step": 23316
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9594100394813806e-05,
+      "loss": 0.2753,
+      "step": 23317
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959405406193146e-05,
+      "loss": 0.2655,
+      "step": 23318
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959400772645964e-05,
+      "loss": 0.2915,
+      "step": 23319
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593961388398348e-05,
+      "loss": 0.2506,
+      "step": 23320
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.95939150477476e-05,
+      "loss": 0.3164,
+      "step": 23321
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593868704507414e-05,
+      "loss": 0.3139,
+      "step": 23322
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593822358677794e-05,
+      "loss": 0.3595,
+      "step": 23323
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593776010258758e-05,
+      "loss": 0.3171,
+      "step": 23324
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593729659250315e-05,
+      "loss": 0.3059,
+      "step": 23325
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959368330565248e-05,
+      "loss": 0.306,
+      "step": 23326
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959363694946527e-05,
+      "loss": 0.3438,
+      "step": 23327
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593590590688686e-05,
+      "loss": 0.3124,
+      "step": 23328
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959354422932275e-05,
+      "loss": 0.272,
+      "step": 23329
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593497865367473e-05,
+      "loss": 0.2579,
+      "step": 23330
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593451498822863e-05,
+      "loss": 0.2781,
+      "step": 23331
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593405129688938e-05,
+      "loss": 0.3513,
+      "step": 23332
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593358757965708e-05,
+      "loss": 0.2689,
+      "step": 23333
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593312383653186e-05,
+      "loss": 0.3432,
+      "step": 23334
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593266006751382e-05,
+      "loss": 0.3284,
+      "step": 23335
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959321962726031e-05,
+      "loss": 0.2845,
+      "step": 23336
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593173245179985e-05,
+      "loss": 0.3037,
+      "step": 23337
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959312686051042e-05,
+      "loss": 0.2962,
+      "step": 23338
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959308047325162e-05,
+      "loss": 0.28,
+      "step": 23339
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593034083403607e-05,
+      "loss": 0.2791,
+      "step": 23340
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592987690966386e-05,
+      "loss": 0.3432,
+      "step": 23341
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959294129593998e-05,
+      "loss": 0.294,
+      "step": 23342
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592894898324384e-05,
+      "loss": 0.297,
+      "step": 23343
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592848498119627e-05,
+      "loss": 0.3191,
+      "step": 23344
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592802095325717e-05,
+      "loss": 0.2773,
+      "step": 23345
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959275568994266e-05,
+      "loss": 0.2962,
+      "step": 23346
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592709281970476e-05,
+      "loss": 0.2734,
+      "step": 23347
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592662871409172e-05,
+      "loss": 0.2853,
+      "step": 23348
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592616458258766e-05,
+      "loss": 0.2732,
+      "step": 23349
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959257004251927e-05,
+      "loss": 0.3116,
+      "step": 23350
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959252362419069e-05,
+      "loss": 0.3209,
+      "step": 23351
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592477203273047e-05,
+      "loss": 0.2625,
+      "step": 23352
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592430779766345e-05,
+      "loss": 0.2808,
+      "step": 23353
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592384353670603e-05,
+      "loss": 0.2894,
+      "step": 23354
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592337924985837e-05,
+      "loss": 0.3124,
+      "step": 23355
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592291493712046e-05,
+      "loss": 0.2957,
+      "step": 23356
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592245059849256e-05,
+      "loss": 0.3407,
+      "step": 23357
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592198623397475e-05,
+      "loss": 0.2614,
+      "step": 23358
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959215218435671e-05,
+      "loss": 0.2781,
+      "step": 23359
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959210574272698e-05,
+      "loss": 0.2395,
+      "step": 23360
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9592059298508298e-05,
+      "loss": 0.2665,
+      "step": 23361
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959201285170067e-05,
+      "loss": 0.3205,
+      "step": 23362
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591966402304118e-05,
+      "loss": 0.3031,
+      "step": 23363
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591919950318647e-05,
+      "loss": 0.3082,
+      "step": 23364
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959187349574427e-05,
+      "loss": 0.3527,
+      "step": 23365
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591827038581007e-05,
+      "loss": 0.3562,
+      "step": 23366
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591780578828858e-05,
+      "loss": 0.2964,
+      "step": 23367
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591734116487847e-05,
+      "loss": 0.2555,
+      "step": 23368
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591687651557982e-05,
+      "loss": 0.2782,
+      "step": 23369
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591641184039276e-05,
+      "loss": 0.2812,
+      "step": 23370
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959159471393174e-05,
+      "loss": 0.3461,
+      "step": 23371
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591548241235388e-05,
+      "loss": 0.2951,
+      "step": 23372
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959150176595023e-05,
+      "loss": 0.2734,
+      "step": 23373
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591455288076285e-05,
+      "loss": 0.2622,
+      "step": 23374
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591408807613557e-05,
+      "loss": 0.3297,
+      "step": 23375
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959136232456207e-05,
+      "loss": 0.3159,
+      "step": 23376
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591315838921822e-05,
+      "loss": 0.2611,
+      "step": 23377
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591269350692835e-05,
+      "loss": 0.287,
+      "step": 23378
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591222859875122e-05,
+      "loss": 0.3361,
+      "step": 23379
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959117636646869e-05,
+      "loss": 0.2582,
+      "step": 23380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591129870473556e-05,
+      "loss": 0.2959,
+      "step": 23381
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9591083371889734e-05,
+      "loss": 0.2836,
+      "step": 23382
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959103687071723e-05,
+      "loss": 0.2963,
+      "step": 23383
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959099036695606e-05,
+      "loss": 0.3146,
+      "step": 23384
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959094386060624e-05,
+      "loss": 0.3485,
+      "step": 23385
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959089735166778e-05,
+      "loss": 0.3015,
+      "step": 23386
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590850840140687e-05,
+      "loss": 0.2649,
+      "step": 23387
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959080432602498e-05,
+      "loss": 0.2735,
+      "step": 23388
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590757809320677e-05,
+      "loss": 0.2799,
+      "step": 23389
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590711290027776e-05,
+      "loss": 0.3118,
+      "step": 23390
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590664768146296e-05,
+      "loss": 0.3182,
+      "step": 23391
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590618243676255e-05,
+      "loss": 0.3195,
+      "step": 23392
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590571716617662e-05,
+      "loss": 0.2917,
+      "step": 23393
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590525186970526e-05,
+      "loss": 0.2619,
+      "step": 23394
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590478654734864e-05,
+      "loss": 0.2981,
+      "step": 23395
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590432119910686e-05,
+      "loss": 0.3143,
+      "step": 23396
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959038558249801e-05,
+      "loss": 0.351,
+      "step": 23397
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590339042496837e-05,
+      "loss": 0.2819,
+      "step": 23398
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959029249990719e-05,
+      "loss": 0.3275,
+      "step": 23399
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959024595472908e-05,
+      "loss": 0.4108,
+      "step": 23400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590199406962514e-05,
+      "loss": 0.3174,
+      "step": 23401
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959015285660751e-05,
+      "loss": 0.2942,
+      "step": 23402
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590106303664077e-05,
+      "loss": 0.2989,
+      "step": 23403
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590059748132234e-05,
+      "loss": 0.3315,
+      "step": 23404
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9590013190011987e-05,
+      "loss": 0.2742,
+      "step": 23405
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.958996662930335e-05,
+      "loss": 0.2827,
+      "step": 23406
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9589920066006335e-05,
+      "loss": 0.2844,
+      "step": 23407
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.958987350012096e-05,
+      "loss": 0.2756,
+      "step": 23408
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.958982693164723e-05,
+      "loss": 0.3106,
+      "step": 23409
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.958978036058516e-05,
+      "loss": 0.2845,
+      "step": 23410
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589733786934767e-05,
+      "loss": 0.2919,
+      "step": 23411
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589687210696055e-05,
+      "loss": 0.2907,
+      "step": 23412
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589640631869047e-05,
+      "loss": 0.3268,
+      "step": 23413
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589594050453748e-05,
+      "loss": 0.3096,
+      "step": 23414
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958954746645017e-05,
+      "loss": 0.3999,
+      "step": 23415
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589500879858333e-05,
+      "loss": 0.2916,
+      "step": 23416
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958945429067824e-05,
+      "loss": 0.2736,
+      "step": 23417
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589407698909912e-05,
+      "loss": 0.2669,
+      "step": 23418
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958936110455336e-05,
+      "loss": 0.3011,
+      "step": 23419
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589314507608587e-05,
+      "loss": 0.299,
+      "step": 23420
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958926790807562e-05,
+      "loss": 0.2953,
+      "step": 23421
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589221305954463e-05,
+      "loss": 0.3116,
+      "step": 23422
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589174701245127e-05,
+      "loss": 0.3657,
+      "step": 23423
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958912809394763e-05,
+      "loss": 0.3114,
+      "step": 23424
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589081484061985e-05,
+      "loss": 0.2493,
+      "step": 23425
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9589034871588203e-05,
+      "loss": 0.3153,
+      "step": 23426
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588988256526292e-05,
+      "loss": 0.3081,
+      "step": 23427
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958894163887627e-05,
+      "loss": 0.2797,
+      "step": 23428
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588895018638148e-05,
+      "loss": 0.2984,
+      "step": 23429
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588848395811936e-05,
+      "loss": 0.2551,
+      "step": 23430
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958880177039765e-05,
+      "loss": 0.3533,
+      "step": 23431
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588755142395303e-05,
+      "loss": 0.3121,
+      "step": 23432
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588708511804906e-05,
+      "loss": 0.2713,
+      "step": 23433
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958866187862647e-05,
+      "loss": 0.3053,
+      "step": 23434
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958861524286001e-05,
+      "loss": 0.2475,
+      "step": 23435
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958856860450554e-05,
+      "loss": 0.2924,
+      "step": 23436
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588521963563066e-05,
+      "loss": 0.3621,
+      "step": 23437
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958847532003261e-05,
+      "loss": 0.3065,
+      "step": 23438
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958842867391418e-05,
+      "loss": 0.3509,
+      "step": 23439
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588382025207783e-05,
+      "loss": 0.2865,
+      "step": 23440
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958833537391344e-05,
+      "loss": 0.3039,
+      "step": 23441
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588288720031162e-05,
+      "loss": 0.2977,
+      "step": 23442
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588242063560957e-05,
+      "loss": 0.3031,
+      "step": 23443
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588195404502843e-05,
+      "loss": 0.3236,
+      "step": 23444
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958814874285683e-05,
+      "loss": 0.2677,
+      "step": 23445
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958810207862293e-05,
+      "loss": 0.3105,
+      "step": 23446
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9588055411801154e-05,
+      "loss": 0.3417,
+      "step": 23447
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958800874239152e-05,
+      "loss": 0.2963,
+      "step": 23448
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587962070394037e-05,
+      "loss": 0.2758,
+      "step": 23449
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958791539580872e-05,
+      "loss": 0.2869,
+      "step": 23450
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587868718635578e-05,
+      "loss": 0.3305,
+      "step": 23451
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587822038874624e-05,
+      "loss": 0.3205,
+      "step": 23452
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587775356525872e-05,
+      "loss": 0.3177,
+      "step": 23453
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958772867158934e-05,
+      "loss": 0.3158,
+      "step": 23454
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958768198406503e-05,
+      "loss": 0.2642,
+      "step": 23455
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958763529395296e-05,
+      "loss": 0.3367,
+      "step": 23456
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587588601253147e-05,
+      "loss": 0.3559,
+      "step": 23457
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587541905965594e-05,
+      "loss": 0.3584,
+      "step": 23458
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958749520809032e-05,
+      "loss": 0.2997,
+      "step": 23459
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587448507627336e-05,
+      "loss": 0.2962,
+      "step": 23460
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587401804576655e-05,
+      "loss": 0.3218,
+      "step": 23461
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587355098938288e-05,
+      "loss": 0.2953,
+      "step": 23462
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587308390712253e-05,
+      "loss": 0.2919,
+      "step": 23463
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587261679898555e-05,
+      "loss": 0.3702,
+      "step": 23464
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587214966497213e-05,
+      "loss": 0.279,
+      "step": 23465
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587168250508236e-05,
+      "loss": 0.348,
+      "step": 23466
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587121531931637e-05,
+      "loss": 0.268,
+      "step": 23467
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958707481076743e-05,
+      "loss": 0.2975,
+      "step": 23468
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9587028087015624e-05,
+      "loss": 0.2854,
+      "step": 23469
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958698136067624e-05,
+      "loss": 0.3168,
+      "step": 23470
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586934631749278e-05,
+      "loss": 0.2924,
+      "step": 23471
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958688790023476e-05,
+      "loss": 0.3063,
+      "step": 23472
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586841166132698e-05,
+      "loss": 0.2753,
+      "step": 23473
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586794429443103e-05,
+      "loss": 0.2713,
+      "step": 23474
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586747690165985e-05,
+      "loss": 0.304,
+      "step": 23475
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958670094830136e-05,
+      "loss": 0.2849,
+      "step": 23476
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586654203849237e-05,
+      "loss": 0.3027,
+      "step": 23477
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586607456809638e-05,
+      "loss": 0.2764,
+      "step": 23478
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586560707182563e-05,
+      "loss": 0.2847,
+      "step": 23479
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586513954968033e-05,
+      "loss": 0.307,
+      "step": 23480
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586467200166055e-05,
+      "loss": 0.3665,
+      "step": 23481
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958642044277665e-05,
+      "loss": 0.2718,
+      "step": 23482
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958637368279982e-05,
+      "loss": 0.3395,
+      "step": 23483
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586326920235585e-05,
+      "loss": 0.3142,
+      "step": 23484
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586280155083958e-05,
+      "loss": 0.2754,
+      "step": 23485
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586233387344944e-05,
+      "loss": 0.2861,
+      "step": 23486
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586186617018563e-05,
+      "loss": 0.3255,
+      "step": 23487
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586139844104828e-05,
+      "loss": 0.2926,
+      "step": 23488
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586093068603746e-05,
+      "loss": 0.3027,
+      "step": 23489
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9586046290515334e-05,
+      "loss": 0.3364,
+      "step": 23490
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585999509839603e-05,
+      "loss": 0.3004,
+      "step": 23491
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585952726576565e-05,
+      "loss": 0.3074,
+      "step": 23492
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958590594072623e-05,
+      "loss": 0.2723,
+      "step": 23493
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585859152288626e-05,
+      "loss": 0.2881,
+      "step": 23494
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585812361263744e-05,
+      "loss": 0.2532,
+      "step": 23495
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585765567651607e-05,
+      "loss": 0.2807,
+      "step": 23496
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585718771452228e-05,
+      "loss": 0.3696,
+      "step": 23497
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585671972665625e-05,
+      "loss": 0.248,
+      "step": 23498
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585625171291796e-05,
+      "loss": 0.2925,
+      "step": 23499
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585578367330763e-05,
+      "loss": 0.2416,
+      "step": 23500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585531560782542e-05,
+      "loss": 0.3189,
+      "step": 23501
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585484751647138e-05,
+      "loss": 0.2996,
+      "step": 23502
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958543793992457e-05,
+      "loss": 0.2861,
+      "step": 23503
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585391125614843e-05,
+      "loss": 0.3067,
+      "step": 23504
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9585344308717974e-05,
+      "loss": 0.3704,
+      "step": 23505
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958529748923398e-05,
+      "loss": 0.306,
+      "step": 23506
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958525066716287e-05,
+      "loss": 0.2968,
+      "step": 23507
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958520384250465e-05,
+      "loss": 0.3317,
+      "step": 23508
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9585157015259342e-05,
+      "loss": 0.2769,
+      "step": 23509
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9585110185426958e-05,
+      "loss": 0.2779,
+      "step": 23510
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.95850633530075e-05,
+      "loss": 0.2998,
+      "step": 23511
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9585016518000997e-05,
+      "loss": 0.2849,
+      "step": 23512
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958496968040745e-05,
+      "loss": 0.2841,
+      "step": 23513
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584922840226875e-05,
+      "loss": 0.2996,
+      "step": 23514
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584875997459285e-05,
+      "loss": 0.2766,
+      "step": 23515
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584829152104693e-05,
+      "loss": 0.2985,
+      "step": 23516
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584782304163107e-05,
+      "loss": 0.3156,
+      "step": 23517
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584735453634548e-05,
+      "loss": 0.3683,
+      "step": 23518
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958468860051902e-05,
+      "loss": 0.2829,
+      "step": 23519
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584641744816543e-05,
+      "loss": 0.3484,
+      "step": 23520
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584594886527125e-05,
+      "loss": 0.3207,
+      "step": 23521
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958454802565078e-05,
+      "loss": 0.2549,
+      "step": 23522
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584501162187522e-05,
+      "loss": 0.2977,
+      "step": 23523
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958445429613736e-05,
+      "loss": 0.2947,
+      "step": 23524
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958440742750031e-05,
+      "loss": 0.3164,
+      "step": 23525
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584360556276385e-05,
+      "loss": 0.2623,
+      "step": 23526
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584313682465596e-05,
+      "loss": 0.2564,
+      "step": 23527
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584266806067952e-05,
+      "loss": 0.2993,
+      "step": 23528
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584219927083476e-05,
+      "loss": 0.2751,
+      "step": 23529
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584173045512172e-05,
+      "loss": 0.2875,
+      "step": 23530
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584126161354053e-05,
+      "loss": 0.3383,
+      "step": 23531
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584079274609132e-05,
+      "loss": 0.3379,
+      "step": 23532
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9584032385277427e-05,
+      "loss": 0.2946,
+      "step": 23533
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583985493358944e-05,
+      "loss": 0.2879,
+      "step": 23534
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.95839385988537e-05,
+      "loss": 0.3377,
+      "step": 23535
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583891701761707e-05,
+      "loss": 0.3696,
+      "step": 23536
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583844802082976e-05,
+      "loss": 0.3104,
+      "step": 23537
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583797899817522e-05,
+      "loss": 0.2887,
+      "step": 23538
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583750994965354e-05,
+      "loss": 0.3176,
+      "step": 23539
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583704087526487e-05,
+      "loss": 0.3277,
+      "step": 23540
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583657177500934e-05,
+      "loss": 0.2944,
+      "step": 23541
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583610264888708e-05,
+      "loss": 0.3133,
+      "step": 23542
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958356334968982e-05,
+      "loss": 0.2788,
+      "step": 23543
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583516431904282e-05,
+      "loss": 0.2638,
+      "step": 23544
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958346951153211e-05,
+      "loss": 0.2811,
+      "step": 23545
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958342258857331e-05,
+      "loss": 0.3023,
+      "step": 23546
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958337566302791e-05,
+      "loss": 0.2746,
+      "step": 23547
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.95833287348959e-05,
+      "loss": 0.2467,
+      "step": 23548
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583281804177313e-05,
+      "loss": 0.3376,
+      "step": 23549
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583234870872153e-05,
+      "loss": 0.2581,
+      "step": 23550
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958318793498043e-05,
+      "loss": 0.3377,
+      "step": 23551
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583140996502158e-05,
+      "loss": 0.3005,
+      "step": 23552
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583094055437353e-05,
+      "loss": 0.2884,
+      "step": 23553
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958304711178603e-05,
+      "loss": 0.3003,
+      "step": 23554
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9583000165548197e-05,
+      "loss": 0.2902,
+      "step": 23555
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9582953216723863e-05,
+      "loss": 0.2916,
+      "step": 23556
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958290626531305e-05,
+      "loss": 0.3121,
+      "step": 23557
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958285931131576e-05,
+      "loss": 0.3357,
+      "step": 23558
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9582812354732016e-05,
+      "loss": 0.3164,
+      "step": 23559
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9582765395561828e-05,
+      "loss": 0.3331,
+      "step": 23560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.95827184338052e-05,
+      "loss": 0.2999,
+      "step": 23561
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958267146946216e-05,
+      "loss": 0.3262,
+      "step": 23562
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9582624502532705e-05,
+      "loss": 0.2675,
+      "step": 23563
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958257753301686e-05,
+      "loss": 0.2877,
+      "step": 23564
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9582530560914626e-05,
+      "loss": 0.267,
+      "step": 23565
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958248358622603e-05,
+      "loss": 0.2896,
+      "step": 23566
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958243660895107e-05,
+      "loss": 0.3061,
+      "step": 23567
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958238962908977e-05,
+      "loss": 0.3059,
+      "step": 23568
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958234264664214e-05,
+      "loss": 0.3285,
+      "step": 23569
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9582295661608186e-05,
+      "loss": 0.2631,
+      "step": 23570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9582248673987926e-05,
+      "loss": 0.2562,
+      "step": 23571
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9582201683781377e-05,
+      "loss": 0.279,
+      "step": 23572
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958215469098854e-05,
+      "loss": 0.3052,
+      "step": 23573
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958210769560944e-05,
+      "loss": 0.3669,
+      "step": 23574
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958206069764408e-05,
+      "loss": 0.3373,
+      "step": 23575
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958201369709248e-05,
+      "loss": 0.2747,
+      "step": 23576
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581966693954653e-05,
+      "loss": 0.2818,
+      "step": 23577
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581919688230605e-05,
+      "loss": 0.3246,
+      "step": 23578
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958187267992035e-05,
+      "loss": 0.3104,
+      "step": 23579
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581825669023906e-05,
+      "loss": 0.3123,
+      "step": 23580
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958177865554128e-05,
+      "loss": 0.332,
+      "step": 23581
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958173163947249e-05,
+      "loss": 0.3139,
+      "step": 23582
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581684620817544e-05,
+      "loss": 0.3009,
+      "step": 23583
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581637599576456e-05,
+      "loss": 0.2877,
+      "step": 23584
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958159057574924e-05,
+      "loss": 0.2967,
+      "step": 23585
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958154354933591e-05,
+      "loss": 0.2505,
+      "step": 23586
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581496520336474e-05,
+      "loss": 0.3504,
+      "step": 23587
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581449488750946e-05,
+      "loss": 0.3259,
+      "step": 23588
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581402454579344e-05,
+      "loss": 0.3395,
+      "step": 23589
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581355417821673e-05,
+      "loss": 0.28,
+      "step": 23590
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958130837847795e-05,
+      "loss": 0.3289,
+      "step": 23591
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958126133654819e-05,
+      "loss": 0.2863,
+      "step": 23592
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.95812142920324e-05,
+      "loss": 0.3623,
+      "step": 23593
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581167244930596e-05,
+      "loss": 0.3347,
+      "step": 23594
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958112019524279e-05,
+      "loss": 0.2744,
+      "step": 23595
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581073142968994e-05,
+      "loss": 0.3137,
+      "step": 23596
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9581026088109227e-05,
+      "loss": 0.2943,
+      "step": 23597
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958097903066349e-05,
+      "loss": 0.26,
+      "step": 23598
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9580931970631804e-05,
+      "loss": 0.3307,
+      "step": 23599
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958088490801418e-05,
+      "loss": 0.2952,
+      "step": 23600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958083784281063e-05,
+      "loss": 0.2608,
+      "step": 23601
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9580790775021167e-05,
+      "loss": 0.33,
+      "step": 23602
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9580743704645802e-05,
+      "loss": 0.3531,
+      "step": 23603
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9580696631684554e-05,
+      "loss": 0.2938,
+      "step": 23604
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958064955613743e-05,
+      "loss": 0.2955,
+      "step": 23605
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.958060247800444e-05,
+      "loss": 0.306,
+      "step": 23606
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9580555397285602e-05,
+      "loss": 0.28,
+      "step": 23607
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.958050831398093e-05,
+      "loss": 0.3215,
+      "step": 23608
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.958046122809043e-05,
+      "loss": 0.3512,
+      "step": 23609
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9580414139614125e-05,
+      "loss": 0.332,
+      "step": 23610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.958036704855202e-05,
+      "loss": 0.3147,
+      "step": 23611
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9580319954904123e-05,
+      "loss": 0.3387,
+      "step": 23612
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9580272858670458e-05,
+      "loss": 0.3202,
+      "step": 23613
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9580225759851032e-05,
+      "loss": 0.3045,
+      "step": 23614
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9580178658445855e-05,
+      "loss": 0.2832,
+      "step": 23615
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9580131554454947e-05,
+      "loss": 0.2887,
+      "step": 23616
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9580084447878316e-05,
+      "loss": 0.3068,
+      "step": 23617
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9580037338715974e-05,
+      "loss": 0.2799,
+      "step": 23618
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579990226967935e-05,
+      "loss": 0.33,
+      "step": 23619
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579943112634213e-05,
+      "loss": 0.2711,
+      "step": 23620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957989599571482e-05,
+      "loss": 0.3095,
+      "step": 23621
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579848876209766e-05,
+      "loss": 0.3624,
+      "step": 23622
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957980175411907e-05,
+      "loss": 0.2588,
+      "step": 23623
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579754629442735e-05,
+      "loss": 0.3149,
+      "step": 23624
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579707502180783e-05,
+      "loss": 0.3159,
+      "step": 23625
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579660372333226e-05,
+      "loss": 0.327,
+      "step": 23626
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957961323990007e-05,
+      "loss": 0.3205,
+      "step": 23627
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579566104881332e-05,
+      "loss": 0.3279,
+      "step": 23628
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579518967277023e-05,
+      "loss": 0.3346,
+      "step": 23629
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957947182708716e-05,
+      "loss": 0.3327,
+      "step": 23630
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957942468431175e-05,
+      "loss": 0.3075,
+      "step": 23631
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957937753895081e-05,
+      "loss": 0.3087,
+      "step": 23632
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579330391004353e-05,
+      "loss": 0.3284,
+      "step": 23633
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957928324047239e-05,
+      "loss": 0.3017,
+      "step": 23634
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957923608735493e-05,
+      "loss": 0.3095,
+      "step": 23635
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957918893165199e-05,
+      "loss": 0.3156,
+      "step": 23636
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579141773363584e-05,
+      "loss": 0.3036,
+      "step": 23637
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579094612489725e-05,
+      "loss": 0.2796,
+      "step": 23638
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957904744903042e-05,
+      "loss": 0.3469,
+      "step": 23639
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9579000282985685e-05,
+      "loss": 0.2975,
+      "step": 23640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578953114355534e-05,
+      "loss": 0.2667,
+      "step": 23641
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578905943139982e-05,
+      "loss": 0.3349,
+      "step": 23642
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957885876933903e-05,
+      "loss": 0.3205,
+      "step": 23643
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957881159295271e-05,
+      "loss": 0.2712,
+      "step": 23644
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578764413981015e-05,
+      "loss": 0.3101,
+      "step": 23645
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578717232423973e-05,
+      "loss": 0.297,
+      "step": 23646
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578670048281587e-05,
+      "loss": 0.3537,
+      "step": 23647
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578622861553876e-05,
+      "loss": 0.321,
+      "step": 23648
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578575672240845e-05,
+      "loss": 0.283,
+      "step": 23649
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578528480342517e-05,
+      "loss": 0.3573,
+      "step": 23650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578481285858895e-05,
+      "loss": 0.3059,
+      "step": 23651
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578434088789997e-05,
+      "loss": 0.2839,
+      "step": 23652
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957838688913584e-05,
+      "loss": 0.3069,
+      "step": 23653
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578339686896424e-05,
+      "loss": 0.2557,
+      "step": 23654
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578292482071773e-05,
+      "loss": 0.3169,
+      "step": 23655
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578245274661897e-05,
+      "loss": 0.2981,
+      "step": 23656
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578198064666808e-05,
+      "loss": 0.2882,
+      "step": 23657
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578150852086513e-05,
+      "loss": 0.3006,
+      "step": 23658
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957810363692104e-05,
+      "loss": 0.3783,
+      "step": 23659
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9578056419170383e-05,
+      "loss": 0.2774,
+      "step": 23660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957800919883457e-05,
+      "loss": 0.2951,
+      "step": 23661
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577961975913606e-05,
+      "loss": 0.2701,
+      "step": 23662
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577914750407504e-05,
+      "loss": 0.2977,
+      "step": 23663
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577867522316278e-05,
+      "loss": 0.2886,
+      "step": 23664
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957782029163994e-05,
+      "loss": 0.2613,
+      "step": 23665
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577773058378504e-05,
+      "loss": 0.3247,
+      "step": 23666
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577725822531986e-05,
+      "loss": 0.3363,
+      "step": 23667
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577678584100392e-05,
+      "loss": 0.3115,
+      "step": 23668
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577631343083738e-05,
+      "loss": 0.3017,
+      "step": 23669
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577584099482038e-05,
+      "loss": 0.3221,
+      "step": 23670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.95775368532953e-05,
+      "loss": 0.3249,
+      "step": 23671
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577489604523542e-05,
+      "loss": 0.2856,
+      "step": 23672
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577442353166774e-05,
+      "loss": 0.2905,
+      "step": 23673
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957739509922501e-05,
+      "loss": 0.2709,
+      "step": 23674
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957734784269826e-05,
+      "loss": 0.2758,
+      "step": 23675
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577300583586544e-05,
+      "loss": 0.2551,
+      "step": 23676
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577253321889865e-05,
+      "loss": 0.2914,
+      "step": 23677
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577206057608245e-05,
+      "loss": 0.2571,
+      "step": 23678
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957715879074169e-05,
+      "loss": 0.2871,
+      "step": 23679
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577111521290215e-05,
+      "loss": 0.3431,
+      "step": 23680
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957706424925383e-05,
+      "loss": 0.2621,
+      "step": 23681
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9577016974632555e-05,
+      "loss": 0.3072,
+      "step": 23682
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576969697426398e-05,
+      "loss": 0.2789,
+      "step": 23683
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957692241763537e-05,
+      "loss": 0.3147,
+      "step": 23684
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576875135259487e-05,
+      "loss": 0.287,
+      "step": 23685
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957682785029876e-05,
+      "loss": 0.3214,
+      "step": 23686
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576780562753204e-05,
+      "loss": 0.2615,
+      "step": 23687
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957673327262283e-05,
+      "loss": 0.3221,
+      "step": 23688
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957668597990765e-05,
+      "loss": 0.2878,
+      "step": 23689
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576638684607677e-05,
+      "loss": 0.365,
+      "step": 23690
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576591386722925e-05,
+      "loss": 0.2971,
+      "step": 23691
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576544086253407e-05,
+      "loss": 0.2703,
+      "step": 23692
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576496783199132e-05,
+      "loss": 0.2963,
+      "step": 23693
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957644947756012e-05,
+      "loss": 0.2937,
+      "step": 23694
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576402169336377e-05,
+      "loss": 0.3048,
+      "step": 23695
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576354858527918e-05,
+      "loss": 0.2655,
+      "step": 23696
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957630754513476e-05,
+      "loss": 0.2977,
+      "step": 23697
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957626022915691e-05,
+      "loss": 0.2615,
+      "step": 23698
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576212910594378e-05,
+      "loss": 0.3041,
+      "step": 23699
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576165589447184e-05,
+      "loss": 0.3333,
+      "step": 23700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957611826571534e-05,
+      "loss": 0.3063,
+      "step": 23701
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576070939398856e-05,
+      "loss": 0.2712,
+      "step": 23702
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9576023610497746e-05,
+      "loss": 0.2816,
+      "step": 23703
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.957597627901202e-05,
+      "loss": 0.3133,
+      "step": 23704
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9575928944941696e-05,
+      "loss": 0.25,
+      "step": 23705
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9575881608286785e-05,
+      "loss": 0.2754,
+      "step": 23706
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575834269047296e-05,
+      "loss": 0.2513,
+      "step": 23707
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575786927223245e-05,
+      "loss": 0.2959,
+      "step": 23708
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575739582814646e-05,
+      "loss": 0.2798,
+      "step": 23709
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575692235821512e-05,
+      "loss": 0.325,
+      "step": 23710
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575644886243847e-05,
+      "loss": 0.3308,
+      "step": 23711
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575597534081674e-05,
+      "loss": 0.2974,
+      "step": 23712
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575550179335004e-05,
+      "loss": 0.2962,
+      "step": 23713
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957550282200385e-05,
+      "loss": 0.2776,
+      "step": 23714
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957545546208822e-05,
+      "loss": 0.2935,
+      "step": 23715
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575408099588128e-05,
+      "loss": 0.3157,
+      "step": 23716
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575360734503593e-05,
+      "loss": 0.3019,
+      "step": 23717
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575313366834622e-05,
+      "loss": 0.2884,
+      "step": 23718
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575265996581225e-05,
+      "loss": 0.2497,
+      "step": 23719
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575218623743425e-05,
+      "loss": 0.3189,
+      "step": 23720
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575171248321223e-05,
+      "loss": 0.2942,
+      "step": 23721
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575123870314642e-05,
+      "loss": 0.3116,
+      "step": 23722
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957507648972369e-05,
+      "loss": 0.2699,
+      "step": 23723
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575029106548377e-05,
+      "loss": 0.36,
+      "step": 23724
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957498172078872e-05,
+      "loss": 0.3239,
+      "step": 23725
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574934332444732e-05,
+      "loss": 0.3099,
+      "step": 23726
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574886941516426e-05,
+      "loss": 0.2759,
+      "step": 23727
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957483954800381e-05,
+      "loss": 0.2802,
+      "step": 23728
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.95747921519069e-05,
+      "loss": 0.3054,
+      "step": 23729
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574744753225713e-05,
+      "loss": 0.29,
+      "step": 23730
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957469735196025e-05,
+      "loss": 0.2847,
+      "step": 23731
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957464994811054e-05,
+      "loss": 0.3285,
+      "step": 23732
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574602541676583e-05,
+      "loss": 0.299,
+      "step": 23733
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574555132658398e-05,
+      "loss": 0.3351,
+      "step": 23734
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574507721055993e-05,
+      "loss": 0.2793,
+      "step": 23735
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574460306869385e-05,
+      "loss": 0.3522,
+      "step": 23736
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574412890098585e-05,
+      "loss": 0.3303,
+      "step": 23737
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574365470743606e-05,
+      "loss": 0.2999,
+      "step": 23738
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574318048804462e-05,
+      "loss": 0.2951,
+      "step": 23739
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574270624281163e-05,
+      "loss": 0.3142,
+      "step": 23740
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574223197173725e-05,
+      "loss": 0.2818,
+      "step": 23741
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957417576748216e-05,
+      "loss": 0.2769,
+      "step": 23742
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574128335206478e-05,
+      "loss": 0.3045,
+      "step": 23743
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574080900346697e-05,
+      "loss": 0.2924,
+      "step": 23744
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574033462902828e-05,
+      "loss": 0.3537,
+      "step": 23745
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573986022874878e-05,
+      "loss": 0.2791,
+      "step": 23746
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573938580262868e-05,
+      "loss": 0.2704,
+      "step": 23747
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573891135066804e-05,
+      "loss": 0.3242,
+      "step": 23748
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.95738436872867e-05,
+      "loss": 0.3537,
+      "step": 23749
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573796236922577e-05,
+      "loss": 0.312,
+      "step": 23750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957374878397444e-05,
+      "loss": 0.284,
+      "step": 23751
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573701328442302e-05,
+      "loss": 0.2767,
+      "step": 23752
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957365387032618e-05,
+      "loss": 0.3204,
+      "step": 23753
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957360640962608e-05,
+      "loss": 0.2868,
+      "step": 23754
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957355894634202e-05,
+      "loss": 0.3028,
+      "step": 23755
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573511480474013e-05,
+      "loss": 0.272,
+      "step": 23756
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957346401202207e-05,
+      "loss": 0.3019,
+      "step": 23757
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573416540986203e-05,
+      "loss": 0.2947,
+      "step": 23758
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957336906736643e-05,
+      "loss": 0.3062,
+      "step": 23759
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573321591162754e-05,
+      "loss": 0.3106,
+      "step": 23760
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.95732741123752e-05,
+      "loss": 0.2847,
+      "step": 23761
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957322663100377e-05,
+      "loss": 0.3078,
+      "step": 23762
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573179147048484e-05,
+      "loss": 0.3182,
+      "step": 23763
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957313166050935e-05,
+      "loss": 0.3351,
+      "step": 23764
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573084171386384e-05,
+      "loss": 0.3042,
+      "step": 23765
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.95730366796796e-05,
+      "loss": 0.3476,
+      "step": 23766
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572989185389007e-05,
+      "loss": 0.313,
+      "step": 23767
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957294168851462e-05,
+      "loss": 0.307,
+      "step": 23768
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957289418905645e-05,
+      "loss": 0.3197,
+      "step": 23769
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957284668701451e-05,
+      "loss": 0.3163,
+      "step": 23770
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572799182388817e-05,
+      "loss": 0.3305,
+      "step": 23771
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572751675179377e-05,
+      "loss": 0.3129,
+      "step": 23772
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957270416538621e-05,
+      "loss": 0.3172,
+      "step": 23773
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572656653009323e-05,
+      "loss": 0.3246,
+      "step": 23774
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572609138048736e-05,
+      "loss": 0.2855,
+      "step": 23775
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957256162050445e-05,
+      "loss": 0.2961,
+      "step": 23776
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957251410037649e-05,
+      "loss": 0.2907,
+      "step": 23777
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572466577664862e-05,
+      "loss": 0.3056,
+      "step": 23778
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572419052369582e-05,
+      "loss": 0.2871,
+      "step": 23779
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572371524490658e-05,
+      "loss": 0.3286,
+      "step": 23780
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572323994028107e-05,
+      "loss": 0.2568,
+      "step": 23781
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572276460981944e-05,
+      "loss": 0.3504,
+      "step": 23782
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572228925352177e-05,
+      "loss": 0.319,
+      "step": 23783
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957218138713882e-05,
+      "loss": 0.2724,
+      "step": 23784
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957213384634189e-05,
+      "loss": 0.284,
+      "step": 23785
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957208630296139e-05,
+      "loss": 0.3207,
+      "step": 23786
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572038756997345e-05,
+      "loss": 0.3451,
+      "step": 23787
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571991208449757e-05,
+      "loss": 0.3188,
+      "step": 23788
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571943657318647e-05,
+      "loss": 0.3868,
+      "step": 23789
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571896103604026e-05,
+      "loss": 0.2715,
+      "step": 23790
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571848547305903e-05,
+      "loss": 0.2996,
+      "step": 23791
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571800988424293e-05,
+      "loss": 0.3031,
+      "step": 23792
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957175342695921e-05,
+      "loss": 0.3469,
+      "step": 23793
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571705862910666e-05,
+      "loss": 0.3163,
+      "step": 23794
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571658296278673e-05,
+      "loss": 0.2999,
+      "step": 23795
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571610727063248e-05,
+      "loss": 0.343,
+      "step": 23796
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571563155264398e-05,
+      "loss": 0.3246,
+      "step": 23797
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571515580882136e-05,
+      "loss": 0.3087,
+      "step": 23798
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957146800391648e-05,
+      "loss": 0.3281,
+      "step": 23799
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571420424367442e-05,
+      "loss": 0.3027,
+      "step": 23800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957137284223503e-05,
+      "loss": 0.2949,
+      "step": 23801
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957132525751926e-05,
+      "loss": 0.3126,
+      "step": 23802
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9571277670220146e-05,
+      "loss": 0.3114,
+      "step": 23803
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.95712300803377e-05,
+      "loss": 0.3484,
+      "step": 23804
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.957118248787193e-05,
+      "loss": 0.303,
+      "step": 23805
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9571134892822857e-05,
+      "loss": 0.3191,
+      "step": 23806
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9571087295190488e-05,
+      "loss": 0.3306,
+      "step": 23807
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.957103969497484e-05,
+      "loss": 0.2823,
+      "step": 23808
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.957099209217592e-05,
+      "loss": 0.2897,
+      "step": 23809
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570944486793748e-05,
+      "loss": 0.34,
+      "step": 23810
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.957089687882833e-05,
+      "loss": 0.3314,
+      "step": 23811
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570849268279685e-05,
+      "loss": 0.3116,
+      "step": 23812
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570801655147823e-05,
+      "loss": 0.3336,
+      "step": 23813
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570754039432755e-05,
+      "loss": 0.2819,
+      "step": 23814
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570706421134497e-05,
+      "loss": 0.2613,
+      "step": 23815
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570658800253063e-05,
+      "loss": 0.3128,
+      "step": 23816
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.957061117678846e-05,
+      "loss": 0.2941,
+      "step": 23817
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570563550740704e-05,
+      "loss": 0.3428,
+      "step": 23818
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.957051592210981e-05,
+      "loss": 0.3036,
+      "step": 23819
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570468290895787e-05,
+      "loss": 0.3277,
+      "step": 23820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570420657098653e-05,
+      "loss": 0.3049,
+      "step": 23821
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570373020718417e-05,
+      "loss": 0.3595,
+      "step": 23822
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.957032538175509e-05,
+      "loss": 0.2986,
+      "step": 23823
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.957027774020869e-05,
+      "loss": 0.2876,
+      "step": 23824
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570230096079225e-05,
+      "loss": 0.2993,
+      "step": 23825
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.957018244936671e-05,
+      "loss": 0.342,
+      "step": 23826
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.957013480007116e-05,
+      "loss": 0.2897,
+      "step": 23827
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570087148192587e-05,
+      "loss": 0.3184,
+      "step": 23828
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9570039493731003e-05,
+      "loss": 0.2815,
+      "step": 23829
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569991836686415e-05,
+      "loss": 0.3337,
+      "step": 23830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569944177058845e-05,
+      "loss": 0.2928,
+      "step": 23831
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569896514848302e-05,
+      "loss": 0.2687,
+      "step": 23832
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569848850054803e-05,
+      "loss": 0.2918,
+      "step": 23833
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569801182678355e-05,
+      "loss": 0.3362,
+      "step": 23834
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956975351271897e-05,
+      "loss": 0.3449,
+      "step": 23835
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569705840176667e-05,
+      "loss": 0.2942,
+      "step": 23836
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569658165051453e-05,
+      "loss": 0.3084,
+      "step": 23837
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956961048734334e-05,
+      "loss": 0.2794,
+      "step": 23838
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956956280705235e-05,
+      "loss": 0.3039,
+      "step": 23839
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569515124178493e-05,
+      "loss": 0.3855,
+      "step": 23840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569467438721775e-05,
+      "loss": 0.2652,
+      "step": 23841
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569419750682216e-05,
+      "loss": 0.285,
+      "step": 23842
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569372060059823e-05,
+      "loss": 0.2877,
+      "step": 23843
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956932436685461e-05,
+      "loss": 0.3083,
+      "step": 23844
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569276671066593e-05,
+      "loss": 0.2732,
+      "step": 23845
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569228972695783e-05,
+      "loss": 0.2673,
+      "step": 23846
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569181271742197e-05,
+      "loss": 0.2939,
+      "step": 23847
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956913356820584e-05,
+      "loss": 0.3023,
+      "step": 23848
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569085862086733e-05,
+      "loss": 0.2974,
+      "step": 23849
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9569038153384882e-05,
+      "loss": 0.3524,
+      "step": 23850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568990442100303e-05,
+      "loss": 0.3891,
+      "step": 23851
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568942728233012e-05,
+      "loss": 0.2691,
+      "step": 23852
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568895011783016e-05,
+      "loss": 0.2721,
+      "step": 23853
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956884729275033e-05,
+      "loss": 0.3475,
+      "step": 23854
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956879957113497e-05,
+      "loss": 0.2736,
+      "step": 23855
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568751846936942e-05,
+      "loss": 0.2615,
+      "step": 23856
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568704120156264e-05,
+      "loss": 0.3447,
+      "step": 23857
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956865639079295e-05,
+      "loss": 0.3084,
+      "step": 23858
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568608658847013e-05,
+      "loss": 0.325,
+      "step": 23859
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956856092431846e-05,
+      "loss": 0.3226,
+      "step": 23860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956851318720731e-05,
+      "loss": 0.2858,
+      "step": 23861
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568465447513573e-05,
+      "loss": 0.2982,
+      "step": 23862
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956841770523726e-05,
+      "loss": 0.3162,
+      "step": 23863
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956836996037839e-05,
+      "loss": 0.2992,
+      "step": 23864
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956832221293697e-05,
+      "loss": 0.2963,
+      "step": 23865
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568274462913015e-05,
+      "loss": 0.3317,
+      "step": 23866
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568226710306537e-05,
+      "loss": 0.3011,
+      "step": 23867
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568178955117553e-05,
+      "loss": 0.3256,
+      "step": 23868
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956813119734607e-05,
+      "loss": 0.3009,
+      "step": 23869
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9568083436992107e-05,
+      "loss": 0.3084,
+      "step": 23870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956803567405567e-05,
+      "loss": 0.3301,
+      "step": 23871
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956798790853678e-05,
+      "loss": 0.3467,
+      "step": 23872
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956794014043544e-05,
+      "loss": 0.3205,
+      "step": 23873
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956789236975167e-05,
+      "loss": 0.2991,
+      "step": 23874
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956784459648548e-05,
+      "loss": 0.2969,
+      "step": 23875
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567796820636888e-05,
+      "loss": 0.3086,
+      "step": 23876
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.95677490422059e-05,
+      "loss": 0.3072,
+      "step": 23877
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956770126119253e-05,
+      "loss": 0.2538,
+      "step": 23878
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567653477596797e-05,
+      "loss": 0.3052,
+      "step": 23879
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567605691418706e-05,
+      "loss": 0.2973,
+      "step": 23880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956755790265828e-05,
+      "loss": 0.279,
+      "step": 23881
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956751011131552e-05,
+      "loss": 0.2717,
+      "step": 23882
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567462317390443e-05,
+      "loss": 0.2668,
+      "step": 23883
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567414520883067e-05,
+      "loss": 0.2926,
+      "step": 23884
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.95673667217934e-05,
+      "loss": 0.3187,
+      "step": 23885
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567318920121453e-05,
+      "loss": 0.337,
+      "step": 23886
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567271115867245e-05,
+      "loss": 0.3374,
+      "step": 23887
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567223309030786e-05,
+      "loss": 0.3222,
+      "step": 23888
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567175499612087e-05,
+      "loss": 0.3162,
+      "step": 23889
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567127687611163e-05,
+      "loss": 0.2846,
+      "step": 23890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9567079873028026e-05,
+      "loss": 0.2748,
+      "step": 23891
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956703205586269e-05,
+      "loss": 0.3202,
+      "step": 23892
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9566984236115164e-05,
+      "loss": 0.3121,
+      "step": 23893
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.956693641378547e-05,
+      "loss": 0.3148,
+      "step": 23894
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9566888588873613e-05,
+      "loss": 0.3271,
+      "step": 23895
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9566840761379607e-05,
+      "loss": 0.3369,
+      "step": 23896
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9566792931303468e-05,
+      "loss": 0.3382,
+      "step": 23897
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9566745098645206e-05,
+      "loss": 0.2821,
+      "step": 23898
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9566697263404833e-05,
+      "loss": 0.2971,
+      "step": 23899
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9566649425582364e-05,
+      "loss": 0.3155,
+      "step": 23900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9566601585177817e-05,
+      "loss": 0.2891,
+      "step": 23901
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9566553742191195e-05,
+      "loss": 0.36,
+      "step": 23902
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9566505896622515e-05,
+      "loss": 0.3115,
+      "step": 23903
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956645804847179e-05,
+      "loss": 0.3028,
+      "step": 23904
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9566410197739035e-05,
+      "loss": 0.3757,
+      "step": 23905
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956636234442426e-05,
+      "loss": 0.3639,
+      "step": 23906
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9566314488527476e-05,
+      "loss": 0.3387,
+      "step": 23907
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9566266630048707e-05,
+      "loss": 0.3179,
+      "step": 23908
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9566218768987953e-05,
+      "loss": 0.2907,
+      "step": 23909
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9566170905345233e-05,
+      "loss": 0.3279,
+      "step": 23910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9566123039120557e-05,
+      "loss": 0.3456,
+      "step": 23911
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9566075170313942e-05,
+      "loss": 0.2802,
+      "step": 23912
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9566027298925397e-05,
+      "loss": 0.2981,
+      "step": 23913
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565979424954937e-05,
+      "loss": 0.2893,
+      "step": 23914
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565931548402574e-05,
+      "loss": 0.2464,
+      "step": 23915
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565883669268323e-05,
+      "loss": 0.3221,
+      "step": 23916
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956583578755219e-05,
+      "loss": 0.2619,
+      "step": 23917
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.95657879032542e-05,
+      "loss": 0.3183,
+      "step": 23918
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565740016374354e-05,
+      "loss": 0.3333,
+      "step": 23919
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565692126912674e-05,
+      "loss": 0.3701,
+      "step": 23920
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565644234869166e-05,
+      "loss": 0.2838,
+      "step": 23921
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565596340243845e-05,
+      "loss": 0.3739,
+      "step": 23922
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956554844303673e-05,
+      "loss": 0.3341,
+      "step": 23923
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565500543247828e-05,
+      "loss": 0.3342,
+      "step": 23924
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956545264087715e-05,
+      "loss": 0.347,
+      "step": 23925
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956540473592471e-05,
+      "loss": 0.3063,
+      "step": 23926
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565356828390525e-05,
+      "loss": 0.2982,
+      "step": 23927
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565308918274607e-05,
+      "loss": 0.3038,
+      "step": 23928
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565261005576964e-05,
+      "loss": 0.3427,
+      "step": 23929
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565213090297614e-05,
+      "loss": 0.274,
+      "step": 23930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565165172436566e-05,
+      "loss": 0.2729,
+      "step": 23931
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956511725199384e-05,
+      "loss": 0.322,
+      "step": 23932
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956506932896944e-05,
+      "loss": 0.2765,
+      "step": 23933
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9565021403363385e-05,
+      "loss": 0.2952,
+      "step": 23934
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564973475175684e-05,
+      "loss": 0.2864,
+      "step": 23935
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564925544406356e-05,
+      "loss": 0.2428,
+      "step": 23936
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564877611055406e-05,
+      "loss": 0.3205,
+      "step": 23937
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564829675122852e-05,
+      "loss": 0.2684,
+      "step": 23938
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956478173660871e-05,
+      "loss": 0.3107,
+      "step": 23939
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956473379551298e-05,
+      "loss": 0.2911,
+      "step": 23940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956468585183569e-05,
+      "loss": 0.2787,
+      "step": 23941
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564637905576846e-05,
+      "loss": 0.3046,
+      "step": 23942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956458995673646e-05,
+      "loss": 0.2835,
+      "step": 23943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564542005314546e-05,
+      "loss": 0.2971,
+      "step": 23944
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956449405131112e-05,
+      "loss": 0.291,
+      "step": 23945
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956444609472619e-05,
+      "loss": 0.2655,
+      "step": 23946
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564398135559773e-05,
+      "loss": 0.2807,
+      "step": 23947
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564350173811878e-05,
+      "loss": 0.3088,
+      "step": 23948
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956430220948252e-05,
+      "loss": 0.3142,
+      "step": 23949
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564254242571716e-05,
+      "loss": 0.3072,
+      "step": 23950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564206273079473e-05,
+      "loss": 0.3129,
+      "step": 23951
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564158301005806e-05,
+      "loss": 0.3072,
+      "step": 23952
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9564110326350728e-05,
+      "loss": 0.3438,
+      "step": 23953
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956406234911425e-05,
+      "loss": 0.2722,
+      "step": 23954
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956401436929639e-05,
+      "loss": 0.3277,
+      "step": 23955
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563966386897154e-05,
+      "loss": 0.2596,
+      "step": 23956
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956391840191656e-05,
+      "loss": 0.3154,
+      "step": 23957
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563870414354622e-05,
+      "loss": 0.3128,
+      "step": 23958
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956382242421135e-05,
+      "loss": 0.3032,
+      "step": 23959
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563774431486755e-05,
+      "loss": 0.3069,
+      "step": 23960
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563726436180854e-05,
+      "loss": 0.2816,
+      "step": 23961
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563678438293658e-05,
+      "loss": 0.2571,
+      "step": 23962
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563630437825183e-05,
+      "loss": 0.3058,
+      "step": 23963
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956358243477544e-05,
+      "loss": 0.3071,
+      "step": 23964
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563534429144438e-05,
+      "loss": 0.2668,
+      "step": 23965
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563486420932195e-05,
+      "loss": 0.2667,
+      "step": 23966
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563438410138724e-05,
+      "loss": 0.3244,
+      "step": 23967
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956339039676403e-05,
+      "loss": 0.3025,
+      "step": 23968
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956334238080814e-05,
+      "loss": 0.3483,
+      "step": 23969
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563294362271055e-05,
+      "loss": 0.2963,
+      "step": 23970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956324634115279e-05,
+      "loss": 0.3662,
+      "step": 23971
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563198317453363e-05,
+      "loss": 0.3449,
+      "step": 23972
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563150291172786e-05,
+      "loss": 0.3222,
+      "step": 23973
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563102262311065e-05,
+      "loss": 0.3297,
+      "step": 23974
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9563054230868225e-05,
+      "loss": 0.3313,
+      "step": 23975
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956300619684427e-05,
+      "loss": 0.3147,
+      "step": 23976
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956295816023921e-05,
+      "loss": 0.2969,
+      "step": 23977
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562910121053066e-05,
+      "loss": 0.2962,
+      "step": 23978
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562862079285846e-05,
+      "loss": 0.3093,
+      "step": 23979
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562814034937568e-05,
+      "loss": 0.3238,
+      "step": 23980
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956276598800824e-05,
+      "loss": 0.312,
+      "step": 23981
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562717938497877e-05,
+      "loss": 0.3267,
+      "step": 23982
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562669886406495e-05,
+      "loss": 0.2507,
+      "step": 23983
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562621831734098e-05,
+      "loss": 0.3034,
+      "step": 23984
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562573774480707e-05,
+      "loss": 0.2852,
+      "step": 23985
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562525714646332e-05,
+      "loss": 0.2747,
+      "step": 23986
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562477652230987e-05,
+      "loss": 0.317,
+      "step": 23987
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562429587234685e-05,
+      "loss": 0.3155,
+      "step": 23988
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956238151965744e-05,
+      "loss": 0.2544,
+      "step": 23989
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956233344949926e-05,
+      "loss": 0.2745,
+      "step": 23990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562285376760167e-05,
+      "loss": 0.3406,
+      "step": 23991
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562237301440163e-05,
+      "loss": 0.3052,
+      "step": 23992
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956218922353927e-05,
+      "loss": 0.2712,
+      "step": 23993
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562141143057498e-05,
+      "loss": 0.33,
+      "step": 23994
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9562093059994854e-05,
+      "loss": 0.2671,
+      "step": 23995
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956204497435136e-05,
+      "loss": 0.2845,
+      "step": 23996
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9561996886127024e-05,
+      "loss": 0.2739,
+      "step": 23997
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9561948795321865e-05,
+      "loss": 0.3084,
+      "step": 23998
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9561900701935885e-05,
+      "loss": 0.2839,
+      "step": 23999
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9561852605969104e-05,
+      "loss": 0.2656,
+      "step": 24000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.956180450742154e-05,
+      "loss": 0.3331,
+      "step": 24001
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561756406293196e-05,
+      "loss": 0.3262,
+      "step": 24002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956170830258409e-05,
+      "loss": 0.2668,
+      "step": 24003
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956166019629423e-05,
+      "loss": 0.2821,
+      "step": 24004
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956161208742364e-05,
+      "loss": 0.3034,
+      "step": 24005
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561563975972326e-05,
+      "loss": 0.3404,
+      "step": 24006
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561515861940297e-05,
+      "loss": 0.3206,
+      "step": 24007
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561467745327572e-05,
+      "loss": 0.3053,
+      "step": 24008
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561419626134162e-05,
+      "loss": 0.3082,
+      "step": 24009
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956137150436008e-05,
+      "loss": 0.3145,
+      "step": 24010
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561323380005338e-05,
+      "loss": 0.3037,
+      "step": 24011
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561275253069955e-05,
+      "loss": 0.2739,
+      "step": 24012
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561227123553934e-05,
+      "loss": 0.32,
+      "step": 24013
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561178991457294e-05,
+      "loss": 0.3483,
+      "step": 24014
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561130856780047e-05,
+      "loss": 0.284,
+      "step": 24015
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956108271952221e-05,
+      "loss": 0.2948,
+      "step": 24016
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9561034579683787e-05,
+      "loss": 0.2971,
+      "step": 24017
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560986437264797e-05,
+      "loss": 0.3138,
+      "step": 24018
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560938292265254e-05,
+      "loss": 0.2899,
+      "step": 24019
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560890144685168e-05,
+      "loss": 0.3047,
+      "step": 24020
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560841994524554e-05,
+      "loss": 0.3287,
+      "step": 24021
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956079384178342e-05,
+      "loss": 0.292,
+      "step": 24022
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956074568646179e-05,
+      "loss": 0.3109,
+      "step": 24023
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560697528559664e-05,
+      "loss": 0.2861,
+      "step": 24024
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560649368077064e-05,
+      "loss": 0.3389,
+      "step": 24025
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560601205014e-05,
+      "loss": 0.2904,
+      "step": 24026
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560553039370483e-05,
+      "loss": 0.3077,
+      "step": 24027
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956050487114653e-05,
+      "loss": 0.2911,
+      "step": 24028
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956045670034215e-05,
+      "loss": 0.2456,
+      "step": 24029
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560408526957362e-05,
+      "loss": 0.3079,
+      "step": 24030
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956036035099217e-05,
+      "loss": 0.3197,
+      "step": 24031
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560312172446597e-05,
+      "loss": 0.3303,
+      "step": 24032
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560263991320648e-05,
+      "loss": 0.3244,
+      "step": 24033
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956021580761434e-05,
+      "loss": 0.3013,
+      "step": 24034
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560167621327685e-05,
+      "loss": 0.3482,
+      "step": 24035
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560119432460696e-05,
+      "loss": 0.2683,
+      "step": 24036
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9560071241013386e-05,
+      "loss": 0.2744,
+      "step": 24037
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.956002304698577e-05,
+      "loss": 0.3035,
+      "step": 24038
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559974850377858e-05,
+      "loss": 0.3414,
+      "step": 24039
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955992665118966e-05,
+      "loss": 0.2959,
+      "step": 24040
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.95598784494212e-05,
+      "loss": 0.2731,
+      "step": 24041
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559830245072477e-05,
+      "loss": 0.2921,
+      "step": 24042
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559782038143516e-05,
+      "loss": 0.31,
+      "step": 24043
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559733828634323e-05,
+      "loss": 0.3479,
+      "step": 24044
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559685616544915e-05,
+      "loss": 0.3234,
+      "step": 24045
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559637401875303e-05,
+      "loss": 0.3623,
+      "step": 24046
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.95595891846255e-05,
+      "loss": 0.3482,
+      "step": 24047
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955954096479552e-05,
+      "loss": 0.2698,
+      "step": 24048
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955949274238537e-05,
+      "loss": 0.2851,
+      "step": 24049
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955944451739507e-05,
+      "loss": 0.3553,
+      "step": 24050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559396289824636e-05,
+      "loss": 0.2743,
+      "step": 24051
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559348059674073e-05,
+      "loss": 0.3638,
+      "step": 24052
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559299826943397e-05,
+      "loss": 0.3169,
+      "step": 24053
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559251591632622e-05,
+      "loss": 0.3017,
+      "step": 24054
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955920335374176e-05,
+      "loss": 0.3165,
+      "step": 24055
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559155113270827e-05,
+      "loss": 0.2828,
+      "step": 24056
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559106870219827e-05,
+      "loss": 0.256,
+      "step": 24057
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9559058624588786e-05,
+      "loss": 0.3065,
+      "step": 24058
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955901037637771e-05,
+      "loss": 0.308,
+      "step": 24059
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558962125586607e-05,
+      "loss": 0.3022,
+      "step": 24060
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.95589138722155e-05,
+      "loss": 0.3138,
+      "step": 24061
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558865616264395e-05,
+      "loss": 0.3629,
+      "step": 24062
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955881735773331e-05,
+      "loss": 0.2957,
+      "step": 24063
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558769096622253e-05,
+      "loss": 0.2961,
+      "step": 24064
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955872083293124e-05,
+      "loss": 0.2653,
+      "step": 24065
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558672566660284e-05,
+      "loss": 0.35,
+      "step": 24066
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.95586242978094e-05,
+      "loss": 0.3028,
+      "step": 24067
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558576026378595e-05,
+      "loss": 0.2952,
+      "step": 24068
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558527752367887e-05,
+      "loss": 0.3483,
+      "step": 24069
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558479475777287e-05,
+      "loss": 0.3681,
+      "step": 24070
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955843119660681e-05,
+      "loss": 0.2939,
+      "step": 24071
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558382914856467e-05,
+      "loss": 0.2557,
+      "step": 24072
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558334630526273e-05,
+      "loss": 0.2963,
+      "step": 24073
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955828634361624e-05,
+      "loss": 0.314,
+      "step": 24074
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558238054126377e-05,
+      "loss": 0.3264,
+      "step": 24075
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558189762056706e-05,
+      "loss": 0.295,
+      "step": 24076
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955814146740723e-05,
+      "loss": 0.2711,
+      "step": 24077
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955809317017797e-05,
+      "loss": 0.2881,
+      "step": 24078
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9558044870368934e-05,
+      "loss": 0.2969,
+      "step": 24079
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955799656798014e-05,
+      "loss": 0.3504,
+      "step": 24080
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557948263011597e-05,
+      "loss": 0.2942,
+      "step": 24081
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557899955463314e-05,
+      "loss": 0.3216,
+      "step": 24082
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557851645335314e-05,
+      "loss": 0.2703,
+      "step": 24083
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557803332627608e-05,
+      "loss": 0.2922,
+      "step": 24084
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557755017340202e-05,
+      "loss": 0.3074,
+      "step": 24085
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557706699473113e-05,
+      "loss": 0.3138,
+      "step": 24086
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557658379026355e-05,
+      "loss": 0.3061,
+      "step": 24087
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955761005599994e-05,
+      "loss": 0.3153,
+      "step": 24088
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557561730393885e-05,
+      "loss": 0.2707,
+      "step": 24089
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557513402208194e-05,
+      "loss": 0.3074,
+      "step": 24090
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955746507144289e-05,
+      "loss": 0.3175,
+      "step": 24091
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557416738097982e-05,
+      "loss": 0.3325,
+      "step": 24092
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.955736840217348e-05,
+      "loss": 0.29,
+      "step": 24093
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557320063669398e-05,
+      "loss": 0.3143,
+      "step": 24094
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557271722585753e-05,
+      "loss": 0.2677,
+      "step": 24095
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557223378922553e-05,
+      "loss": 0.3107,
+      "step": 24096
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557175032679817e-05,
+      "loss": 0.3633,
+      "step": 24097
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557126683857553e-05,
+      "loss": 0.2905,
+      "step": 24098
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9557078332455776e-05,
+      "loss": 0.2949,
+      "step": 24099
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.95570299784745e-05,
+      "loss": 0.3133,
+      "step": 24100
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556981621913735e-05,
+      "loss": 0.2878,
+      "step": 24101
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556933262773496e-05,
+      "loss": 0.2763,
+      "step": 24102
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.95568849010538e-05,
+      "loss": 0.2996,
+      "step": 24103
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556836536754652e-05,
+      "loss": 0.3339,
+      "step": 24104
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955678816987607e-05,
+      "loss": 0.2643,
+      "step": 24105
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556739800418067e-05,
+      "loss": 0.2896,
+      "step": 24106
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556691428380653e-05,
+      "loss": 0.3775,
+      "step": 24107
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556643053763847e-05,
+      "loss": 0.2956,
+      "step": 24108
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556594676567657e-05,
+      "loss": 0.3086,
+      "step": 24109
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556546296792095e-05,
+      "loss": 0.2788,
+      "step": 24110
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955649791443718e-05,
+      "loss": 0.3645,
+      "step": 24111
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955644952950292e-05,
+      "loss": 0.2901,
+      "step": 24112
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955640114198933e-05,
+      "loss": 0.331,
+      "step": 24113
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955635275189642e-05,
+      "loss": 0.3216,
+      "step": 24114
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955630435922421e-05,
+      "loss": 0.2871,
+      "step": 24115
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556255963972705e-05,
+      "loss": 0.3128,
+      "step": 24116
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556207566141924e-05,
+      "loss": 0.2831,
+      "step": 24117
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556159165731875e-05,
+      "loss": 0.3002,
+      "step": 24118
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955611076274258e-05,
+      "loss": 0.3177,
+      "step": 24119
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955606235717404e-05,
+      "loss": 0.3163,
+      "step": 24120
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9556013949026278e-05,
+      "loss": 0.3079,
+      "step": 24121
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555965538299302e-05,
+      "loss": 0.2677,
+      "step": 24122
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955591712499313e-05,
+      "loss": 0.3083,
+      "step": 24123
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555868709107763e-05,
+      "loss": 0.2988,
+      "step": 24124
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955582029064323e-05,
+      "loss": 0.2843,
+      "step": 24125
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955577186959953e-05,
+      "loss": 0.2662,
+      "step": 24126
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555723445976683e-05,
+      "loss": 0.3208,
+      "step": 24127
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555675019774707e-05,
+      "loss": 0.2911,
+      "step": 24128
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555626590993607e-05,
+      "loss": 0.3101,
+      "step": 24129
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.95555781596334e-05,
+      "loss": 0.306,
+      "step": 24130
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555529725694097e-05,
+      "loss": 0.247,
+      "step": 24131
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555481289175707e-05,
+      "loss": 0.2773,
+      "step": 24132
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555432850078254e-05,
+      "loss": 0.2978,
+      "step": 24133
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555384408401742e-05,
+      "loss": 0.2549,
+      "step": 24134
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555335964146187e-05,
+      "loss": 0.3182,
+      "step": 24135
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555287517311607e-05,
+      "loss": 0.3242,
+      "step": 24136
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555239067898004e-05,
+      "loss": 0.3056,
+      "step": 24137
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555190615905403e-05,
+      "loss": 0.2772,
+      "step": 24138
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555142161333806e-05,
+      "loss": 0.3239,
+      "step": 24139
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555093704183235e-05,
+      "loss": 0.3134,
+      "step": 24140
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9555045244453696e-05,
+      "loss": 0.3119,
+      "step": 24141
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554996782145206e-05,
+      "loss": 0.2759,
+      "step": 24142
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554948317257782e-05,
+      "loss": 0.273,
+      "step": 24143
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955489984979143e-05,
+      "loss": 0.3051,
+      "step": 24144
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554851379746165e-05,
+      "loss": 0.3131,
+      "step": 24145
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554802907122003e-05,
+      "loss": 0.3452,
+      "step": 24146
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554754431918958e-05,
+      "loss": 0.3197,
+      "step": 24147
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554705954137036e-05,
+      "loss": 0.2971,
+      "step": 24148
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955465747377625e-05,
+      "loss": 0.3253,
+      "step": 24149
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554608990836623e-05,
+      "loss": 0.3445,
+      "step": 24150
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955456050531816e-05,
+      "loss": 0.2899,
+      "step": 24151
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955451201722088e-05,
+      "loss": 0.2865,
+      "step": 24152
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955446352654479e-05,
+      "loss": 0.322,
+      "step": 24153
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554415033289907e-05,
+      "loss": 0.298,
+      "step": 24154
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554366537456243e-05,
+      "loss": 0.3577,
+      "step": 24155
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554318039043807e-05,
+      "loss": 0.299,
+      "step": 24156
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554269538052616e-05,
+      "loss": 0.2646,
+      "step": 24157
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554221034482688e-05,
+      "loss": 0.2606,
+      "step": 24158
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955417252833403e-05,
+      "loss": 0.3167,
+      "step": 24159
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554124019606653e-05,
+      "loss": 0.2684,
+      "step": 24160
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554075508300573e-05,
+      "loss": 0.2746,
+      "step": 24161
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9554026994415807e-05,
+      "loss": 0.2901,
+      "step": 24162
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553978477952363e-05,
+      "loss": 0.3334,
+      "step": 24163
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553929958910254e-05,
+      "loss": 0.3118,
+      "step": 24164
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553881437289494e-05,
+      "loss": 0.3728,
+      "step": 24165
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.95538329130901e-05,
+      "loss": 0.2916,
+      "step": 24166
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955378438631208e-05,
+      "loss": 0.3635,
+      "step": 24167
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553735856955448e-05,
+      "loss": 0.2727,
+      "step": 24168
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553687325020223e-05,
+      "loss": 0.3353,
+      "step": 24169
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553638790506406e-05,
+      "loss": 0.2745,
+      "step": 24170
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955359025341402e-05,
+      "loss": 0.3577,
+      "step": 24171
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553541713743078e-05,
+      "loss": 0.3439,
+      "step": 24172
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955349317149359e-05,
+      "loss": 0.3625,
+      "step": 24173
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553444626665568e-05,
+      "loss": 0.2828,
+      "step": 24174
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553396079259025e-05,
+      "loss": 0.2791,
+      "step": 24175
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553347529273978e-05,
+      "loss": 0.3494,
+      "step": 24176
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553298976710438e-05,
+      "loss": 0.2655,
+      "step": 24177
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553250421568414e-05,
+      "loss": 0.3275,
+      "step": 24178
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553201863847927e-05,
+      "loss": 0.3559,
+      "step": 24179
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553153303548987e-05,
+      "loss": 0.2721,
+      "step": 24180
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553104740671605e-05,
+      "loss": 0.2865,
+      "step": 24181
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553056175215796e-05,
+      "loss": 0.3263,
+      "step": 24182
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9553007607181568e-05,
+      "loss": 0.3004,
+      "step": 24183
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9552959036568945e-05,
+      "loss": 0.2579,
+      "step": 24184
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955291046337793e-05,
+      "loss": 0.2875,
+      "step": 24185
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955286188760854e-05,
+      "loss": 0.3035,
+      "step": 24186
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9552813309260788e-05,
+      "loss": 0.3037,
+      "step": 24187
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955276472833469e-05,
+      "loss": 0.3004,
+      "step": 24188
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9552716144830254e-05,
+      "loss": 0.3014,
+      "step": 24189
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9552667558747492e-05,
+      "loss": 0.2547,
+      "step": 24190
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9552618970086427e-05,
+      "loss": 0.3425,
+      "step": 24191
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955257037884706e-05,
+      "loss": 0.3508,
+      "step": 24192
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955252178502941e-05,
+      "loss": 0.3166,
+      "step": 24193
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.955247318863349e-05,
+      "loss": 0.355,
+      "step": 24194
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9552424589659318e-05,
+      "loss": 0.2955,
+      "step": 24195
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9552375988106895e-05,
+      "loss": 0.2925,
+      "step": 24196
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9552327383976247e-05,
+      "loss": 0.2629,
+      "step": 24197
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9552278777267376e-05,
+      "loss": 0.3309,
+      "step": 24198
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9552230167980306e-05,
+      "loss": 0.2766,
+      "step": 24199
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9552181556115037e-05,
+      "loss": 0.3583,
+      "step": 24200
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9552132941671592e-05,
+      "loss": 0.3026,
+      "step": 24201
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9552084324649983e-05,
+      "loss": 0.2877,
+      "step": 24202
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9552035705050226e-05,
+      "loss": 0.3331,
+      "step": 24203
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551987082872324e-05,
+      "loss": 0.2978,
+      "step": 24204
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551938458116298e-05,
+      "loss": 0.2779,
+      "step": 24205
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955188983078216e-05,
+      "loss": 0.3313,
+      "step": 24206
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955184120086992e-05,
+      "loss": 0.3069,
+      "step": 24207
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551792568379595e-05,
+      "loss": 0.2892,
+      "step": 24208
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551743933311197e-05,
+      "loss": 0.3263,
+      "step": 24209
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551695295664735e-05,
+      "loss": 0.3083,
+      "step": 24210
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955164665544023e-05,
+      "loss": 0.2906,
+      "step": 24211
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955159801263769e-05,
+      "loss": 0.331,
+      "step": 24212
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955154936725713e-05,
+      "loss": 0.2876,
+      "step": 24213
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955150071929856e-05,
+      "loss": 0.281,
+      "step": 24214
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551452068762e-05,
+      "loss": 0.285,
+      "step": 24215
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955140341564745e-05,
+      "loss": 0.2941,
+      "step": 24216
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551354759954936e-05,
+      "loss": 0.3454,
+      "step": 24217
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955130610168447e-05,
+      "loss": 0.312,
+      "step": 24218
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551257440836057e-05,
+      "loss": 0.304,
+      "step": 24219
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551208777409718e-05,
+      "loss": 0.2977,
+      "step": 24220
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955116011140546e-05,
+      "loss": 0.325,
+      "step": 24221
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551111442823302e-05,
+      "loss": 0.3184,
+      "step": 24222
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551062771663254e-05,
+      "loss": 0.3018,
+      "step": 24223
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9551014097925328e-05,
+      "loss": 0.2737,
+      "step": 24224
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955096542160954e-05,
+      "loss": 0.3313,
+      "step": 24225
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550916742715903e-05,
+      "loss": 0.3149,
+      "step": 24226
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550868061244424e-05,
+      "loss": 0.2781,
+      "step": 24227
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550819377195126e-05,
+      "loss": 0.2917,
+      "step": 24228
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550770690568015e-05,
+      "loss": 0.2604,
+      "step": 24229
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550722001363106e-05,
+      "loss": 0.3154,
+      "step": 24230
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550673309580414e-05,
+      "loss": 0.2919,
+      "step": 24231
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955062461521995e-05,
+      "loss": 0.3596,
+      "step": 24232
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955057591828173e-05,
+      "loss": 0.3018,
+      "step": 24233
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955052721876576e-05,
+      "loss": 0.2856,
+      "step": 24234
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955047851667206e-05,
+      "loss": 0.3364,
+      "step": 24235
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955042981200064e-05,
+      "loss": 0.2694,
+      "step": 24236
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550381104751517e-05,
+      "loss": 0.3284,
+      "step": 24237
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550332394924703e-05,
+      "loss": 0.3385,
+      "step": 24238
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550283682520208e-05,
+      "loss": 0.3084,
+      "step": 24239
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550234967538045e-05,
+      "loss": 0.2968,
+      "step": 24240
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.955018624997823e-05,
+      "loss": 0.3563,
+      "step": 24241
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550137529840775e-05,
+      "loss": 0.2908,
+      "step": 24242
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550088807125696e-05,
+      "loss": 0.3539,
+      "step": 24243
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9550040081833e-05,
+      "loss": 0.3257,
+      "step": 24244
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.95499913539627e-05,
+      "loss": 0.289,
+      "step": 24245
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549942623514822e-05,
+      "loss": 0.3237,
+      "step": 24246
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549893890489364e-05,
+      "loss": 0.2643,
+      "step": 24247
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549845154886348e-05,
+      "loss": 0.2953,
+      "step": 24248
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954979641670578e-05,
+      "loss": 0.2816,
+      "step": 24249
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954974767594768e-05,
+      "loss": 0.3025,
+      "step": 24250
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954969893261206e-05,
+      "loss": 0.2739,
+      "step": 24251
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549650186698926e-05,
+      "loss": 0.2883,
+      "step": 24252
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549601438208304e-05,
+      "loss": 0.2748,
+      "step": 24253
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549552687140193e-05,
+      "loss": 0.3026,
+      "step": 24254
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954950393349462e-05,
+      "loss": 0.2844,
+      "step": 24255
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549455177271585e-05,
+      "loss": 0.2699,
+      "step": 24256
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549406418471108e-05,
+      "loss": 0.2833,
+      "step": 24257
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549357657093204e-05,
+      "loss": 0.2637,
+      "step": 24258
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549308893137887e-05,
+      "loss": 0.2545,
+      "step": 24259
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954926012660516e-05,
+      "loss": 0.2666,
+      "step": 24260
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954921135749505e-05,
+      "loss": 0.3775,
+      "step": 24261
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549162585807555e-05,
+      "loss": 0.3493,
+      "step": 24262
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.95491138115427e-05,
+      "loss": 0.316,
+      "step": 24263
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549065034700497e-05,
+      "loss": 0.3091,
+      "step": 24264
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9549016255280957e-05,
+      "loss": 0.316,
+      "step": 24265
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954896747328409e-05,
+      "loss": 0.2854,
+      "step": 24266
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548918688709914e-05,
+      "loss": 0.2297,
+      "step": 24267
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548869901558437e-05,
+      "loss": 0.3168,
+      "step": 24268
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954882111182968e-05,
+      "loss": 0.2888,
+      "step": 24269
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954877231952365e-05,
+      "loss": 0.3684,
+      "step": 24270
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548723524640357e-05,
+      "loss": 0.2681,
+      "step": 24271
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548674727179825e-05,
+      "loss": 0.3378,
+      "step": 24272
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548625927142057e-05,
+      "loss": 0.3087,
+      "step": 24273
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548577124527072e-05,
+      "loss": 0.2586,
+      "step": 24274
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954852831933488e-05,
+      "loss": 0.2585,
+      "step": 24275
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548479511565498e-05,
+      "loss": 0.2664,
+      "step": 24276
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548430701218935e-05,
+      "loss": 0.3344,
+      "step": 24277
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548381888295207e-05,
+      "loss": 0.3111,
+      "step": 24278
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548333072794327e-05,
+      "loss": 0.2851,
+      "step": 24279
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548284254716306e-05,
+      "loss": 0.2906,
+      "step": 24280
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954823543406116e-05,
+      "loss": 0.2974,
+      "step": 24281
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548186610828898e-05,
+      "loss": 0.3267,
+      "step": 24282
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548137785019535e-05,
+      "loss": 0.2839,
+      "step": 24283
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954808895663309e-05,
+      "loss": 0.2915,
+      "step": 24284
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9548040125669568e-05,
+      "loss": 0.3324,
+      "step": 24285
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9547991292128984e-05,
+      "loss": 0.3086,
+      "step": 24286
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9547942456011352e-05,
+      "loss": 0.312,
+      "step": 24287
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9547893617316688e-05,
+      "loss": 0.3023,
+      "step": 24288
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9547844776045003e-05,
+      "loss": 0.2661,
+      "step": 24289
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9547795932196312e-05,
+      "loss": 0.3372,
+      "step": 24290
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9547747085770622e-05,
+      "loss": 0.3173,
+      "step": 24291
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9547698236767952e-05,
+      "loss": 0.2788,
+      "step": 24292
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9547649385188314e-05,
+      "loss": 0.3462,
+      "step": 24293
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954760053103172e-05,
+      "loss": 0.3278,
+      "step": 24294
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9547551674298187e-05,
+      "loss": 0.3147,
+      "step": 24295
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9547502814987723e-05,
+      "loss": 0.2646,
+      "step": 24296
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.954745395310034e-05,
+      "loss": 0.2329,
+      "step": 24297
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9547405088636057e-05,
+      "loss": 0.315,
+      "step": 24298
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954735622159489e-05,
+      "loss": 0.3562,
+      "step": 24299
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9547307351976837e-05,
+      "loss": 0.3035,
+      "step": 24300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9547258479781928e-05,
+      "loss": 0.2854,
+      "step": 24301
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954720960501017e-05,
+      "loss": 0.2855,
+      "step": 24302
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9547160727661572e-05,
+      "loss": 0.3052,
+      "step": 24303
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9547111847736152e-05,
+      "loss": 0.2763,
+      "step": 24304
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954706296523392e-05,
+      "loss": 0.2852,
+      "step": 24305
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9547014080154897e-05,
+      "loss": 0.3463,
+      "step": 24306
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546965192499085e-05,
+      "loss": 0.297,
+      "step": 24307
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.95469163022665e-05,
+      "loss": 0.3446,
+      "step": 24308
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546867409457164e-05,
+      "loss": 0.3156,
+      "step": 24309
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954681851407108e-05,
+      "loss": 0.2583,
+      "step": 24310
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546769616108268e-05,
+      "loss": 0.3905,
+      "step": 24311
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546720715568733e-05,
+      "loss": 0.3087,
+      "step": 24312
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.95466718124525e-05,
+      "loss": 0.3197,
+      "step": 24313
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546622906759572e-05,
+      "loss": 0.3183,
+      "step": 24314
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954657399848997e-05,
+      "loss": 0.3082,
+      "step": 24315
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546525087643697e-05,
+      "loss": 0.2698,
+      "step": 24316
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546476174220773e-05,
+      "loss": 0.2863,
+      "step": 24317
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546427258221213e-05,
+      "loss": 0.3225,
+      "step": 24318
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546378339645026e-05,
+      "loss": 0.2902,
+      "step": 24319
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954632941849223e-05,
+      "loss": 0.3341,
+      "step": 24320
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546280494762832e-05,
+      "loss": 0.3018,
+      "step": 24321
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954623156845685e-05,
+      "loss": 0.2915,
+      "step": 24322
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546182639574294e-05,
+      "loss": 0.2809,
+      "step": 24323
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954613370811518e-05,
+      "loss": 0.2791,
+      "step": 24324
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954608477407952e-05,
+      "loss": 0.3059,
+      "step": 24325
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9546035837467324e-05,
+      "loss": 0.3263,
+      "step": 24326
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954598689827861e-05,
+      "loss": 0.3025,
+      "step": 24327
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954593795651339e-05,
+      "loss": 0.2704,
+      "step": 24328
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545889012171676e-05,
+      "loss": 0.2656,
+      "step": 24329
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954584006525348e-05,
+      "loss": 0.2842,
+      "step": 24330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545791115758823e-05,
+      "loss": 0.3519,
+      "step": 24331
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954574216368771e-05,
+      "loss": 0.3384,
+      "step": 24332
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545693209040155e-05,
+      "loss": 0.3346,
+      "step": 24333
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545644251816175e-05,
+      "loss": 0.3109,
+      "step": 24334
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954559529201578e-05,
+      "loss": 0.2816,
+      "step": 24335
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545546329638982e-05,
+      "loss": 0.2924,
+      "step": 24336
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545497364685796e-05,
+      "loss": 0.2979,
+      "step": 24337
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954544839715624e-05,
+      "loss": 0.2744,
+      "step": 24338
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954539942705032e-05,
+      "loss": 0.2989,
+      "step": 24339
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545350454368054e-05,
+      "loss": 0.2846,
+      "step": 24340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545301479109452e-05,
+      "loss": 0.3498,
+      "step": 24341
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545252501274528e-05,
+      "loss": 0.3019,
+      "step": 24342
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545203520863296e-05,
+      "loss": 0.3426,
+      "step": 24343
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954515453787577e-05,
+      "loss": 0.3098,
+      "step": 24344
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954510555231196e-05,
+      "loss": 0.3163,
+      "step": 24345
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9545056564171885e-05,
+      "loss": 0.2795,
+      "step": 24346
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954500757345555e-05,
+      "loss": 0.2931,
+      "step": 24347
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544958580162974e-05,
+      "loss": 0.3331,
+      "step": 24348
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954490958429417e-05,
+      "loss": 0.2899,
+      "step": 24349
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544860585849153e-05,
+      "loss": 0.29,
+      "step": 24350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954481158482793e-05,
+      "loss": 0.3357,
+      "step": 24351
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954476258123052e-05,
+      "loss": 0.251,
+      "step": 24352
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544713575056934e-05,
+      "loss": 0.3978,
+      "step": 24353
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544664566307185e-05,
+      "loss": 0.3467,
+      "step": 24354
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544615554981283e-05,
+      "loss": 0.303,
+      "step": 24355
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954456654107925e-05,
+      "loss": 0.3031,
+      "step": 24356
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544517524601086e-05,
+      "loss": 0.3001,
+      "step": 24357
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954446850554682e-05,
+      "loss": 0.2996,
+      "step": 24358
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544419483916456e-05,
+      "loss": 0.2864,
+      "step": 24359
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544370459710005e-05,
+      "loss": 0.3194,
+      "step": 24360
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544321432927487e-05,
+      "loss": 0.2668,
+      "step": 24361
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954427240356891e-05,
+      "loss": 0.2932,
+      "step": 24362
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954422337163429e-05,
+      "loss": 0.3092,
+      "step": 24363
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544174337123637e-05,
+      "loss": 0.2846,
+      "step": 24364
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544125300036973e-05,
+      "loss": 0.2784,
+      "step": 24365
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.95440762603743e-05,
+      "loss": 0.2722,
+      "step": 24366
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9544027218135638e-05,
+      "loss": 0.3133,
+      "step": 24367
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543978173320997e-05,
+      "loss": 0.3967,
+      "step": 24368
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954392912593039e-05,
+      "loss": 0.3121,
+      "step": 24369
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543880075963835e-05,
+      "loss": 0.3273,
+      "step": 24370
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954383102342134e-05,
+      "loss": 0.3139,
+      "step": 24371
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543781968302924e-05,
+      "loss": 0.2909,
+      "step": 24372
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543732910608596e-05,
+      "loss": 0.3104,
+      "step": 24373
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543683850338365e-05,
+      "loss": 0.2957,
+      "step": 24374
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543634787492257e-05,
+      "loss": 0.3267,
+      "step": 24375
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543585722070274e-05,
+      "loss": 0.274,
+      "step": 24376
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954353665407243e-05,
+      "loss": 0.2681,
+      "step": 24377
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543487583498742e-05,
+      "loss": 0.2379,
+      "step": 24378
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543438510349223e-05,
+      "loss": 0.3172,
+      "step": 24379
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543389434623884e-05,
+      "loss": 0.2694,
+      "step": 24380
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543340356322738e-05,
+      "loss": 0.2765,
+      "step": 24381
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543291275445802e-05,
+      "loss": 0.3382,
+      "step": 24382
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954324219199309e-05,
+      "loss": 0.2677,
+      "step": 24383
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954319310596461e-05,
+      "loss": 0.2871,
+      "step": 24384
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543144017360377e-05,
+      "loss": 0.3129,
+      "step": 24385
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543094926180407e-05,
+      "loss": 0.3028,
+      "step": 24386
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9543045832424708e-05,
+      "loss": 0.3056,
+      "step": 24387
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9542996736093296e-05,
+      "loss": 0.2866,
+      "step": 24388
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9542947637186187e-05,
+      "loss": 0.2671,
+      "step": 24389
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9542898535703392e-05,
+      "loss": 0.3075,
+      "step": 24390
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9542849431644923e-05,
+      "loss": 0.3424,
+      "step": 24391
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9542800325010793e-05,
+      "loss": 0.3549,
+      "step": 24392
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9542751215801018e-05,
+      "loss": 0.2946,
+      "step": 24393
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954270210401561e-05,
+      "loss": 0.2622,
+      "step": 24394
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954265298965458e-05,
+      "loss": 0.2531,
+      "step": 24395
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9542603872717944e-05,
+      "loss": 0.3825,
+      "step": 24396
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9542554753205718e-05,
+      "loss": 0.3095,
+      "step": 24397
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9542505631117908e-05,
+      "loss": 0.3596,
+      "step": 24398
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9542456506454536e-05,
+      "loss": 0.2883,
+      "step": 24399
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9542407379215604e-05,
+      "loss": 0.3005,
+      "step": 24400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9542358249401133e-05,
+      "loss": 0.2901,
+      "step": 24401
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954230911701114e-05,
+      "loss": 0.2829,
+      "step": 24402
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9542259982045627e-05,
+      "loss": 0.3426,
+      "step": 24403
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954221084450462e-05,
+      "loss": 0.269,
+      "step": 24404
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954216170438812e-05,
+      "loss": 0.3019,
+      "step": 24405
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954211256169615e-05,
+      "loss": 0.304,
+      "step": 24406
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9542063416428715e-05,
+      "loss": 0.3617,
+      "step": 24407
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9542014268585835e-05,
+      "loss": 0.2884,
+      "step": 24408
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954196511816752e-05,
+      "loss": 0.303,
+      "step": 24409
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541915965173787e-05,
+      "loss": 0.3467,
+      "step": 24410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541866809604642e-05,
+      "loss": 0.3152,
+      "step": 24411
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541817651460104e-05,
+      "loss": 0.2852,
+      "step": 24412
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541768490740182e-05,
+      "loss": 0.3259,
+      "step": 24413
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541719327444897e-05,
+      "loss": 0.2973,
+      "step": 24414
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541670161574256e-05,
+      "loss": 0.2891,
+      "step": 24415
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541620993128273e-05,
+      "loss": 0.2634,
+      "step": 24416
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541571822106963e-05,
+      "loss": 0.3699,
+      "step": 24417
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541522648510335e-05,
+      "loss": 0.3149,
+      "step": 24418
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541473472338408e-05,
+      "loss": 0.2631,
+      "step": 24419
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541424293591192e-05,
+      "loss": 0.3163,
+      "step": 24420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541375112268702e-05,
+      "loss": 0.2783,
+      "step": 24421
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954132592837095e-05,
+      "loss": 0.304,
+      "step": 24422
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541276741897947e-05,
+      "loss": 0.2841,
+      "step": 24423
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954122755284971e-05,
+      "loss": 0.2985,
+      "step": 24424
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541178361226255e-05,
+      "loss": 0.3107,
+      "step": 24425
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541129167027587e-05,
+      "loss": 0.2739,
+      "step": 24426
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541079970253724e-05,
+      "loss": 0.2809,
+      "step": 24427
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9541030770904682e-05,
+      "loss": 0.2746,
+      "step": 24428
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540981568980467e-05,
+      "loss": 0.3081,
+      "step": 24429
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.95409323644811e-05,
+      "loss": 0.313,
+      "step": 24430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540883157406587e-05,
+      "loss": 0.3645,
+      "step": 24431
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540833947756947e-05,
+      "loss": 0.3236,
+      "step": 24432
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540784735532194e-05,
+      "loss": 0.3625,
+      "step": 24433
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540735520732335e-05,
+      "loss": 0.321,
+      "step": 24434
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954068630335739e-05,
+      "loss": 0.358,
+      "step": 24435
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540637083407367e-05,
+      "loss": 0.2635,
+      "step": 24436
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954058786088228e-05,
+      "loss": 0.2829,
+      "step": 24437
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540538635782145e-05,
+      "loss": 0.291,
+      "step": 24438
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540489408106976e-05,
+      "loss": 0.3092,
+      "step": 24439
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954044017785678e-05,
+      "loss": 0.2928,
+      "step": 24440
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954039094503158e-05,
+      "loss": 0.25,
+      "step": 24441
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954034170963138e-05,
+      "loss": 0.3186,
+      "step": 24442
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.95402924716562e-05,
+      "loss": 0.2672,
+      "step": 24443
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540243231106045e-05,
+      "loss": 0.2741,
+      "step": 24444
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954019398798094e-05,
+      "loss": 0.3113,
+      "step": 24445
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540144742280887e-05,
+      "loss": 0.2628,
+      "step": 24446
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.954009549400591e-05,
+      "loss": 0.2757,
+      "step": 24447
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9540046243156015e-05,
+      "loss": 0.3156,
+      "step": 24448
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539996989731214e-05,
+      "loss": 0.3171,
+      "step": 24449
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539947733731527e-05,
+      "loss": 0.3459,
+      "step": 24450
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953989847515696e-05,
+      "loss": 0.3309,
+      "step": 24451
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953984921400753e-05,
+      "loss": 0.3271,
+      "step": 24452
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953979995028325e-05,
+      "loss": 0.3077,
+      "step": 24453
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539750683984135e-05,
+      "loss": 0.2924,
+      "step": 24454
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539701415110198e-05,
+      "loss": 0.3082,
+      "step": 24455
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539652143661447e-05,
+      "loss": 0.305,
+      "step": 24456
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539602869637902e-05,
+      "loss": 0.2668,
+      "step": 24457
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539553593039574e-05,
+      "loss": 0.3586,
+      "step": 24458
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539504313866475e-05,
+      "loss": 0.2958,
+      "step": 24459
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953945503211862e-05,
+      "loss": 0.2778,
+      "step": 24460
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539405747796022e-05,
+      "loss": 0.3422,
+      "step": 24461
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953935646089869e-05,
+      "loss": 0.3436,
+      "step": 24462
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539307171426645e-05,
+      "loss": 0.3012,
+      "step": 24463
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539257879379896e-05,
+      "loss": 0.2774,
+      "step": 24464
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539208584758454e-05,
+      "loss": 0.371,
+      "step": 24465
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539159287562338e-05,
+      "loss": 0.2668,
+      "step": 24466
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539109987791556e-05,
+      "loss": 0.3012,
+      "step": 24467
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539060685446127e-05,
+      "loss": 0.3052,
+      "step": 24468
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9539011380526056e-05,
+      "loss": 0.2942,
+      "step": 24469
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538962073031364e-05,
+      "loss": 0.3435,
+      "step": 24470
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953891276296206e-05,
+      "loss": 0.3122,
+      "step": 24471
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538863450318164e-05,
+      "loss": 0.2774,
+      "step": 24472
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953881413509968e-05,
+      "loss": 0.2869,
+      "step": 24473
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538764817306626e-05,
+      "loss": 0.4172,
+      "step": 24474
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538715496939015e-05,
+      "loss": 0.2831,
+      "step": 24475
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538666173996858e-05,
+      "loss": 0.2673,
+      "step": 24476
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538616848480174e-05,
+      "loss": 0.2955,
+      "step": 24477
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953856752038897e-05,
+      "loss": 0.3107,
+      "step": 24478
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538518189723263e-05,
+      "loss": 0.3466,
+      "step": 24479
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538468856483065e-05,
+      "loss": 0.3045,
+      "step": 24480
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953841952066839e-05,
+      "loss": 0.2841,
+      "step": 24481
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953837018227925e-05,
+      "loss": 0.3093,
+      "step": 24482
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953832084131566e-05,
+      "loss": 0.2613,
+      "step": 24483
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538271497777634e-05,
+      "loss": 0.3324,
+      "step": 24484
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538222151665182e-05,
+      "loss": 0.3687,
+      "step": 24485
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538172802978322e-05,
+      "loss": 0.3363,
+      "step": 24486
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953812345171706e-05,
+      "loss": 0.2538,
+      "step": 24487
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538074097881417e-05,
+      "loss": 0.2742,
+      "step": 24488
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9538024741471406e-05,
+      "loss": 0.2839,
+      "step": 24489
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9537975382487034e-05,
+      "loss": 0.3092,
+      "step": 24490
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9537926020928316e-05,
+      "loss": 0.3057,
+      "step": 24491
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953787665679527e-05,
+      "loss": 0.3201,
+      "step": 24492
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953782729008791e-05,
+      "loss": 0.2887,
+      "step": 24493
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.953777792080624e-05,
+      "loss": 0.3097,
+      "step": 24494
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953772854895028e-05,
+      "loss": 0.278,
+      "step": 24495
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9537679174520043e-05,
+      "loss": 0.3467,
+      "step": 24496
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953762979751554e-05,
+      "loss": 0.2931,
+      "step": 24497
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953758041793679e-05,
+      "loss": 0.3376,
+      "step": 24498
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.95375310357838e-05,
+      "loss": 0.3075,
+      "step": 24499
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9537481651056586e-05,
+      "loss": 0.2927,
+      "step": 24500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953743226375516e-05,
+      "loss": 0.2753,
+      "step": 24501
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953738287387954e-05,
+      "loss": 0.3177,
+      "step": 24502
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9537333481429733e-05,
+      "loss": 0.3089,
+      "step": 24503
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9537284086405756e-05,
+      "loss": 0.3143,
+      "step": 24504
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9537234688807626e-05,
+      "loss": 0.2773,
+      "step": 24505
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9537185288635342e-05,
+      "loss": 0.3148,
+      "step": 24506
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9537135885888932e-05,
+      "loss": 0.306,
+      "step": 24507
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9537086480568406e-05,
+      "loss": 0.3142,
+      "step": 24508
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9537037072673778e-05,
+      "loss": 0.3064,
+      "step": 24509
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536987662205053e-05,
+      "loss": 0.2688,
+      "step": 24510
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536938249162253e-05,
+      "loss": 0.2919,
+      "step": 24511
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953688883354539e-05,
+      "loss": 0.2912,
+      "step": 24512
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536839415354473e-05,
+      "loss": 0.3923,
+      "step": 24513
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953678999458952e-05,
+      "loss": 0.3416,
+      "step": 24514
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536740571250544e-05,
+      "loss": 0.2875,
+      "step": 24515
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953669114533756e-05,
+      "loss": 0.3082,
+      "step": 24516
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536641716850574e-05,
+      "loss": 0.3678,
+      "step": 24517
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536592285789604e-05,
+      "loss": 0.2887,
+      "step": 24518
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536542852154664e-05,
+      "loss": 0.3118,
+      "step": 24519
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536493415945767e-05,
+      "loss": 0.3245,
+      "step": 24520
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536443977162923e-05,
+      "loss": 0.3185,
+      "step": 24521
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536394535806153e-05,
+      "loss": 0.3134,
+      "step": 24522
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536345091875463e-05,
+      "loss": 0.3018,
+      "step": 24523
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953629564537087e-05,
+      "loss": 0.3737,
+      "step": 24524
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536246196292384e-05,
+      "loss": 0.2892,
+      "step": 24525
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953619674464002e-05,
+      "loss": 0.3009,
+      "step": 24526
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536147290413794e-05,
+      "loss": 0.3503,
+      "step": 24527
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536097833613715e-05,
+      "loss": 0.2902,
+      "step": 24528
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9536048374239802e-05,
+      "loss": 0.3664,
+      "step": 24529
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535998912292064e-05,
+      "loss": 0.3094,
+      "step": 24530
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535949447770514e-05,
+      "loss": 0.3882,
+      "step": 24531
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535899980675167e-05,
+      "loss": 0.3273,
+      "step": 24532
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953585051100604e-05,
+      "loss": 0.2695,
+      "step": 24533
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535801038763134e-05,
+      "loss": 0.3226,
+      "step": 24534
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953575156394648e-05,
+      "loss": 0.3305,
+      "step": 24535
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535702086556076e-05,
+      "loss": 0.2834,
+      "step": 24536
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953565260659194e-05,
+      "loss": 0.2848,
+      "step": 24537
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535603124054092e-05,
+      "loss": 0.3025,
+      "step": 24538
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535553638942537e-05,
+      "loss": 0.304,
+      "step": 24539
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535504151257293e-05,
+      "loss": 0.2732,
+      "step": 24540
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535454660998373e-05,
+      "loss": 0.3742,
+      "step": 24541
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535405168165788e-05,
+      "loss": 0.3086,
+      "step": 24542
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953535567275955e-05,
+      "loss": 0.3597,
+      "step": 24543
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535306174779677e-05,
+      "loss": 0.3096,
+      "step": 24544
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535256674226183e-05,
+      "loss": 0.2797,
+      "step": 24545
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535207171099075e-05,
+      "loss": 0.3063,
+      "step": 24546
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953515766539837e-05,
+      "loss": 0.2902,
+      "step": 24547
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535108157124084e-05,
+      "loss": 0.3369,
+      "step": 24548
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9535058646276222e-05,
+      "loss": 0.2716,
+      "step": 24549
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953500913285481e-05,
+      "loss": 0.2784,
+      "step": 24550
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953495961685985e-05,
+      "loss": 0.3175,
+      "step": 24551
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534910098291366e-05,
+      "loss": 0.2621,
+      "step": 24552
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953486057714936e-05,
+      "loss": 0.2653,
+      "step": 24553
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953481105343385e-05,
+      "loss": 0.2974,
+      "step": 24554
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953476152714485e-05,
+      "loss": 0.2835,
+      "step": 24555
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534711998282377e-05,
+      "loss": 0.2909,
+      "step": 24556
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534662466846438e-05,
+      "loss": 0.2851,
+      "step": 24557
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534612932837046e-05,
+      "loss": 0.3175,
+      "step": 24558
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534563396254223e-05,
+      "loss": 0.3278,
+      "step": 24559
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534513857097975e-05,
+      "loss": 0.3425,
+      "step": 24560
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534464315368312e-05,
+      "loss": 0.2817,
+      "step": 24561
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953441477106526e-05,
+      "loss": 0.3107,
+      "step": 24562
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953436522418882e-05,
+      "loss": 0.2573,
+      "step": 24563
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534315674739014e-05,
+      "loss": 0.3469,
+      "step": 24564
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953426612271585e-05,
+      "loss": 0.3626,
+      "step": 24565
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534216568119344e-05,
+      "loss": 0.2774,
+      "step": 24566
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534167010949507e-05,
+      "loss": 0.3595,
+      "step": 24567
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534117451206353e-05,
+      "loss": 0.3164,
+      "step": 24568
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534067888889896e-05,
+      "loss": 0.3341,
+      "step": 24569
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9534018324000154e-05,
+      "loss": 0.3462,
+      "step": 24570
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953396875653713e-05,
+      "loss": 0.3785,
+      "step": 24571
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533919186500844e-05,
+      "loss": 0.3539,
+      "step": 24572
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953386961389131e-05,
+      "loss": 0.2644,
+      "step": 24573
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533820038708542e-05,
+      "loss": 0.2861,
+      "step": 24574
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953377046095255e-05,
+      "loss": 0.3248,
+      "step": 24575
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533720880623347e-05,
+      "loss": 0.3033,
+      "step": 24576
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953367129772095e-05,
+      "loss": 0.2888,
+      "step": 24577
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533621712245373e-05,
+      "loss": 0.3293,
+      "step": 24578
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953357212419662e-05,
+      "loss": 0.2941,
+      "step": 24579
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953352253357472e-05,
+      "loss": 0.291,
+      "step": 24580
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533472940379673e-05,
+      "loss": 0.3403,
+      "step": 24581
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533423344611497e-05,
+      "loss": 0.3461,
+      "step": 24582
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533373746270207e-05,
+      "loss": 0.2746,
+      "step": 24583
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953332414535581e-05,
+      "loss": 0.3324,
+      "step": 24584
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953327454186833e-05,
+      "loss": 0.3108,
+      "step": 24585
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533224935807774e-05,
+      "loss": 0.3047,
+      "step": 24586
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533175327174153e-05,
+      "loss": 0.3129,
+      "step": 24587
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533125715967483e-05,
+      "loss": 0.257,
+      "step": 24588
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.953307610218778e-05,
+      "loss": 0.3215,
+      "step": 24589
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9533026485835054e-05,
+      "loss": 0.2806,
+      "step": 24590
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9532976866909323e-05,
+      "loss": 0.2999,
+      "step": 24591
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9532927245410596e-05,
+      "loss": 0.2695,
+      "step": 24592
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9532877621338885e-05,
+      "loss": 0.2992,
+      "step": 24593
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532827994694206e-05,
+      "loss": 0.2823,
+      "step": 24594
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532778365476574e-05,
+      "loss": 0.3285,
+      "step": 24595
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532728733686e-05,
+      "loss": 0.3083,
+      "step": 24596
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532679099322497e-05,
+      "loss": 0.2741,
+      "step": 24597
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953262946238608e-05,
+      "loss": 0.3307,
+      "step": 24598
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953257982287676e-05,
+      "loss": 0.2746,
+      "step": 24599
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532530180794556e-05,
+      "loss": 0.2896,
+      "step": 24600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532480536139474e-05,
+      "loss": 0.3165,
+      "step": 24601
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532430888911532e-05,
+      "loss": 0.2972,
+      "step": 24602
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532381239110744e-05,
+      "loss": 0.3171,
+      "step": 24603
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532331586737118e-05,
+      "loss": 0.3705,
+      "step": 24604
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532281931790673e-05,
+      "loss": 0.265,
+      "step": 24605
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953223227427142e-05,
+      "loss": 0.2869,
+      "step": 24606
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532182614179375e-05,
+      "loss": 0.3856,
+      "step": 24607
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953213295151455e-05,
+      "loss": 0.3276,
+      "step": 24608
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532083286276956e-05,
+      "loss": 0.2883,
+      "step": 24609
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9532033618466604e-05,
+      "loss": 0.3051,
+      "step": 24610
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953198394808352e-05,
+      "loss": 0.3208,
+      "step": 24611
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531934275127704e-05,
+      "loss": 0.2958,
+      "step": 24612
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531884599599175e-05,
+      "loss": 0.3531,
+      "step": 24613
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531834921497946e-05,
+      "loss": 0.336,
+      "step": 24614
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531785240824027e-05,
+      "loss": 0.3304,
+      "step": 24615
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953173555757744e-05,
+      "loss": 0.3304,
+      "step": 24616
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531685871758192e-05,
+      "loss": 0.3037,
+      "step": 24617
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531636183366295e-05,
+      "loss": 0.3556,
+      "step": 24618
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531586492401766e-05,
+      "loss": 0.3058,
+      "step": 24619
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531536798864616e-05,
+      "loss": 0.2795,
+      "step": 24620
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531487102754863e-05,
+      "loss": 0.3108,
+      "step": 24621
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531437404072512e-05,
+      "loss": 0.3027,
+      "step": 24622
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531387702817587e-05,
+      "loss": 0.284,
+      "step": 24623
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953133799899009e-05,
+      "loss": 0.3015,
+      "step": 24624
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531288292590047e-05,
+      "loss": 0.328,
+      "step": 24625
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953123858361746e-05,
+      "loss": 0.2848,
+      "step": 24626
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953118887207235e-05,
+      "loss": 0.32,
+      "step": 24627
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531139157954724e-05,
+      "loss": 0.312,
+      "step": 24628
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531089441264602e-05,
+      "loss": 0.3915,
+      "step": 24629
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9531039722001994e-05,
+      "loss": 0.3385,
+      "step": 24630
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530990000166912e-05,
+      "loss": 0.3475,
+      "step": 24631
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530940275759375e-05,
+      "loss": 0.3254,
+      "step": 24632
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530890548779388e-05,
+      "loss": 0.2768,
+      "step": 24633
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530840819226972e-05,
+      "loss": 0.3116,
+      "step": 24634
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530791087102138e-05,
+      "loss": 0.2633,
+      "step": 24635
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530741352404895e-05,
+      "loss": 0.273,
+      "step": 24636
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530691615135263e-05,
+      "loss": 0.2919,
+      "step": 24637
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530641875293254e-05,
+      "loss": 0.319,
+      "step": 24638
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953059213287888e-05,
+      "loss": 0.3121,
+      "step": 24639
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530542387892155e-05,
+      "loss": 0.3234,
+      "step": 24640
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953049264033309e-05,
+      "loss": 0.3013,
+      "step": 24641
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.95304428902017e-05,
+      "loss": 0.3032,
+      "step": 24642
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530393137498e-05,
+      "loss": 0.3147,
+      "step": 24643
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530343382222004e-05,
+      "loss": 0.3153,
+      "step": 24644
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953029362437372e-05,
+      "loss": 0.2688,
+      "step": 24645
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530243863953168e-05,
+      "loss": 0.3197,
+      "step": 24646
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953019410096036e-05,
+      "loss": 0.3044,
+      "step": 24647
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9530144335395308e-05,
+      "loss": 0.3675,
+      "step": 24648
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953009456725802e-05,
+      "loss": 0.301,
+      "step": 24649
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.953004479654852e-05,
+      "loss": 0.3129,
+      "step": 24650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529995023266813e-05,
+      "loss": 0.3003,
+      "step": 24651
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.952994524741292e-05,
+      "loss": 0.3222,
+      "step": 24652
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.952989546898685e-05,
+      "loss": 0.2604,
+      "step": 24653
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529845687988614e-05,
+      "loss": 0.2887,
+      "step": 24654
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529795904418227e-05,
+      "loss": 0.2935,
+      "step": 24655
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529746118275706e-05,
+      "loss": 0.2794,
+      "step": 24656
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.952969632956106e-05,
+      "loss": 0.3766,
+      "step": 24657
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529646538274306e-05,
+      "loss": 0.3464,
+      "step": 24658
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529596744415457e-05,
+      "loss": 0.3359,
+      "step": 24659
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529546947984522e-05,
+      "loss": 0.3002,
+      "step": 24660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529497148981523e-05,
+      "loss": 0.3417,
+      "step": 24661
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529447347406462e-05,
+      "loss": 0.3381,
+      "step": 24662
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.952939754325936e-05,
+      "loss": 0.2662,
+      "step": 24663
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.952934773654023e-05,
+      "loss": 0.2978,
+      "step": 24664
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529297927249086e-05,
+      "loss": 0.2887,
+      "step": 24665
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529248115385942e-05,
+      "loss": 0.3418,
+      "step": 24666
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529198300950803e-05,
+      "loss": 0.3071,
+      "step": 24667
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529148483943693e-05,
+      "loss": 0.3357,
+      "step": 24668
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9529098664364624e-05,
+      "loss": 0.3545,
+      "step": 24669
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.95290488422136e-05,
+      "loss": 0.2955,
+      "step": 24670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528999017490644e-05,
+      "loss": 0.2703,
+      "step": 24671
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528949190195766e-05,
+      "loss": 0.3624,
+      "step": 24672
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528899360328984e-05,
+      "loss": 0.3561,
+      "step": 24673
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528849527890302e-05,
+      "loss": 0.2904,
+      "step": 24674
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528799692879743e-05,
+      "loss": 0.3263,
+      "step": 24675
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528749855297315e-05,
+      "loss": 0.2714,
+      "step": 24676
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528700015143033e-05,
+      "loss": 0.3132,
+      "step": 24677
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.952865017241691e-05,
+      "loss": 0.2942,
+      "step": 24678
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.952860032711896e-05,
+      "loss": 0.3068,
+      "step": 24679
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528550479249196e-05,
+      "loss": 0.3497,
+      "step": 24680
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528500628807632e-05,
+      "loss": 0.2953,
+      "step": 24681
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528450775794282e-05,
+      "loss": 0.3235,
+      "step": 24682
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528400920209158e-05,
+      "loss": 0.3263,
+      "step": 24683
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528351062052272e-05,
+      "loss": 0.3268,
+      "step": 24684
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.952830120132364e-05,
+      "loss": 0.3417,
+      "step": 24685
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.952825133802328e-05,
+      "loss": 0.3729,
+      "step": 24686
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528201472151194e-05,
+      "loss": 0.3354,
+      "step": 24687
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528151603707405e-05,
+      "loss": 0.2975,
+      "step": 24688
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528101732691922e-05,
+      "loss": 0.3387,
+      "step": 24689
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528051859104762e-05,
+      "loss": 0.3062,
+      "step": 24690
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9528001982945932e-05,
+      "loss": 0.2885,
+      "step": 24691
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527952104215453e-05,
+      "loss": 0.3019,
+      "step": 24692
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527902222913334e-05,
+      "loss": 0.2732,
+      "step": 24693
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952785233903959e-05,
+      "loss": 0.2914,
+      "step": 24694
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527802452594236e-05,
+      "loss": 0.2671,
+      "step": 24695
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527752563577283e-05,
+      "loss": 0.3073,
+      "step": 24696
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527702671988745e-05,
+      "loss": 0.3009,
+      "step": 24697
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527652777828632e-05,
+      "loss": 0.2491,
+      "step": 24698
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527602881096965e-05,
+      "loss": 0.285,
+      "step": 24699
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527552981793752e-05,
+      "loss": 0.2862,
+      "step": 24700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527503079919006e-05,
+      "loss": 0.3047,
+      "step": 24701
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527453175472745e-05,
+      "loss": 0.3021,
+      "step": 24702
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527403268454977e-05,
+      "loss": 0.3039,
+      "step": 24703
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527353358865725e-05,
+      "loss": 0.2913,
+      "step": 24704
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527303446704987e-05,
+      "loss": 0.3015,
+      "step": 24705
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952725353197279e-05,
+      "loss": 0.2802,
+      "step": 24706
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527203614669146e-05,
+      "loss": 0.3695,
+      "step": 24707
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952715369479406e-05,
+      "loss": 0.3228,
+      "step": 24708
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952710377234755e-05,
+      "loss": 0.2918,
+      "step": 24709
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9527053847329632e-05,
+      "loss": 0.2727,
+      "step": 24710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952700391974032e-05,
+      "loss": 0.3441,
+      "step": 24711
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526953989579623e-05,
+      "loss": 0.3109,
+      "step": 24712
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526904056847558e-05,
+      "loss": 0.3501,
+      "step": 24713
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526854121544137e-05,
+      "loss": 0.2993,
+      "step": 24714
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952680418366937e-05,
+      "loss": 0.3305,
+      "step": 24715
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952675424322328e-05,
+      "loss": 0.304,
+      "step": 24716
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526704300205866e-05,
+      "loss": 0.33,
+      "step": 24717
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526654354617155e-05,
+      "loss": 0.3481,
+      "step": 24718
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526604406457157e-05,
+      "loss": 0.2628,
+      "step": 24719
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526554455725882e-05,
+      "loss": 0.2889,
+      "step": 24720
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526504502423343e-05,
+      "loss": 0.3687,
+      "step": 24721
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526454546549562e-05,
+      "loss": 0.347,
+      "step": 24722
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952640458810454e-05,
+      "loss": 0.3203,
+      "step": 24723
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.95263546270883e-05,
+      "loss": 0.3059,
+      "step": 24724
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526304663500854e-05,
+      "loss": 0.2943,
+      "step": 24725
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526254697342212e-05,
+      "loss": 0.2967,
+      "step": 24726
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952620472861239e-05,
+      "loss": 0.3216,
+      "step": 24727
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526154757311396e-05,
+      "loss": 0.3247,
+      "step": 24728
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9526104783439255e-05,
+      "loss": 0.3729,
+      "step": 24729
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952605480699597e-05,
+      "loss": 0.2854,
+      "step": 24730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952600482798156e-05,
+      "loss": 0.3251,
+      "step": 24731
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952595484639604e-05,
+      "loss": 0.2643,
+      "step": 24732
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525904862239414e-05,
+      "loss": 0.3304,
+      "step": 24733
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525854875511702e-05,
+      "loss": 0.3486,
+      "step": 24734
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525804886212923e-05,
+      "loss": 0.2642,
+      "step": 24735
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525754894343078e-05,
+      "loss": 0.2391,
+      "step": 24736
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525704899902192e-05,
+      "loss": 0.3133,
+      "step": 24737
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525654902890273e-05,
+      "loss": 0.2975,
+      "step": 24738
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525604903307332e-05,
+      "loss": 0.308,
+      "step": 24739
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525554901153388e-05,
+      "loss": 0.311,
+      "step": 24740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525504896428452e-05,
+      "loss": 0.2992,
+      "step": 24741
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525454889132537e-05,
+      "loss": 0.3291,
+      "step": 24742
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952540487926566e-05,
+      "loss": 0.2822,
+      "step": 24743
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952535486682783e-05,
+      "loss": 0.2743,
+      "step": 24744
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952530485181906e-05,
+      "loss": 0.3036,
+      "step": 24745
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952525483423937e-05,
+      "loss": 0.2752,
+      "step": 24746
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525204814088768e-05,
+      "loss": 0.2874,
+      "step": 24747
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525154791367268e-05,
+      "loss": 0.3034,
+      "step": 24748
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952510476607488e-05,
+      "loss": 0.2524,
+      "step": 24749
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525054738211627e-05,
+      "loss": 0.3466,
+      "step": 24750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9525004707777517e-05,
+      "loss": 0.3017,
+      "step": 24751
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524954674772563e-05,
+      "loss": 0.33,
+      "step": 24752
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952490463919678e-05,
+      "loss": 0.3158,
+      "step": 24753
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524854601050176e-05,
+      "loss": 0.3227,
+      "step": 24754
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524804560332777e-05,
+      "loss": 0.3531,
+      "step": 24755
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524754517044582e-05,
+      "loss": 0.2986,
+      "step": 24756
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524704471185615e-05,
+      "loss": 0.3408,
+      "step": 24757
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524654422755883e-05,
+      "loss": 0.3012,
+      "step": 24758
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524604371755402e-05,
+      "loss": 0.309,
+      "step": 24759
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524554318184186e-05,
+      "loss": 0.3393,
+      "step": 24760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524504262042253e-05,
+      "loss": 0.3002,
+      "step": 24761
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524454203329605e-05,
+      "loss": 0.2985,
+      "step": 24762
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524404142046266e-05,
+      "loss": 0.2676,
+      "step": 24763
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524354078192247e-05,
+      "loss": 0.3106,
+      "step": 24764
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524304011767557e-05,
+      "loss": 0.3108,
+      "step": 24765
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524253942772214e-05,
+      "loss": 0.3003,
+      "step": 24766
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524203871206234e-05,
+      "loss": 0.2774,
+      "step": 24767
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952415379706962e-05,
+      "loss": 0.2475,
+      "step": 24768
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.95241037203624e-05,
+      "loss": 0.3209,
+      "step": 24769
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524053641084574e-05,
+      "loss": 0.3126,
+      "step": 24770
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9524003559236166e-05,
+      "loss": 0.2678,
+      "step": 24771
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952395347481718e-05,
+      "loss": 0.3008,
+      "step": 24772
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523903387827638e-05,
+      "loss": 0.2926,
+      "step": 24773
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952385329826755e-05,
+      "loss": 0.3243,
+      "step": 24774
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523803206136926e-05,
+      "loss": 0.3395,
+      "step": 24775
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523753111435786e-05,
+      "loss": 0.2927,
+      "step": 24776
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523703014164138e-05,
+      "loss": 0.3432,
+      "step": 24777
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523652914322003e-05,
+      "loss": 0.3467,
+      "step": 24778
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523602811909384e-05,
+      "loss": 0.3649,
+      "step": 24779
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523552706926304e-05,
+      "loss": 0.3107,
+      "step": 24780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952350259937277e-05,
+      "loss": 0.2732,
+      "step": 24781
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523452489248798e-05,
+      "loss": 0.2458,
+      "step": 24782
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523402376554405e-05,
+      "loss": 0.3034,
+      "step": 24783
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.95233522612896e-05,
+      "loss": 0.2911,
+      "step": 24784
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523302143454393e-05,
+      "loss": 0.3164,
+      "step": 24785
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523252023048805e-05,
+      "loss": 0.2947,
+      "step": 24786
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952320190007285e-05,
+      "loss": 0.3339,
+      "step": 24787
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9523151774526536e-05,
+      "loss": 0.3069,
+      "step": 24788
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952310164640988e-05,
+      "loss": 0.2842,
+      "step": 24789
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.952305151572289e-05,
+      "loss": 0.319,
+      "step": 24790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952300138246559e-05,
+      "loss": 0.3456,
+      "step": 24791
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9522951246637985e-05,
+      "loss": 0.3297,
+      "step": 24792
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952290110824009e-05,
+      "loss": 0.2712,
+      "step": 24793
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952285096727192e-05,
+      "loss": 0.2982,
+      "step": 24794
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952280082373349e-05,
+      "loss": 0.2928,
+      "step": 24795
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9522750677624805e-05,
+      "loss": 0.2721,
+      "step": 24796
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952270052894589e-05,
+      "loss": 0.3434,
+      "step": 24797
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9522650377696753e-05,
+      "loss": 0.2751,
+      "step": 24798
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952260022387741e-05,
+      "loss": 0.3131,
+      "step": 24799
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952255006748787e-05,
+      "loss": 0.3029,
+      "step": 24800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952249990852815e-05,
+      "loss": 0.32,
+      "step": 24801
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9522449746998263e-05,
+      "loss": 0.335,
+      "step": 24802
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952239958289822e-05,
+      "loss": 0.2721,
+      "step": 24803
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952234941622804e-05,
+      "loss": 0.3156,
+      "step": 24804
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952229924698773e-05,
+      "loss": 0.3955,
+      "step": 24805
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9522249075177313e-05,
+      "loss": 0.3846,
+      "step": 24806
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9522198900796794e-05,
+      "loss": 0.2872,
+      "step": 24807
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9522148723846186e-05,
+      "loss": 0.3202,
+      "step": 24808
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952209854432551e-05,
+      "loss": 0.3306,
+      "step": 24809
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952204836223477e-05,
+      "loss": 0.3239,
+      "step": 24810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521998177573988e-05,
+      "loss": 0.3442,
+      "step": 24811
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521947990343172e-05,
+      "loss": 0.2534,
+      "step": 24812
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952189780054234e-05,
+      "loss": 0.322,
+      "step": 24813
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.95218476081715e-05,
+      "loss": 0.3178,
+      "step": 24814
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521797413230673e-05,
+      "loss": 0.3564,
+      "step": 24815
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521747215719866e-05,
+      "loss": 0.2786,
+      "step": 24816
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521697015639096e-05,
+      "loss": 0.3446,
+      "step": 24817
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521646812988374e-05,
+      "loss": 0.2643,
+      "step": 24818
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521596607767717e-05,
+      "loss": 0.309,
+      "step": 24819
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521546399977135e-05,
+      "loss": 0.2902,
+      "step": 24820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952149618961664e-05,
+      "loss": 0.3077,
+      "step": 24821
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521445976686257e-05,
+      "loss": 0.237,
+      "step": 24822
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521395761185987e-05,
+      "loss": 0.2743,
+      "step": 24823
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521345543115844e-05,
+      "loss": 0.3555,
+      "step": 24824
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952129532247585e-05,
+      "loss": 0.2918,
+      "step": 24825
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521245099266012e-05,
+      "loss": 0.3348,
+      "step": 24826
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521194873486348e-05,
+      "loss": 0.2839,
+      "step": 24827
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521144645136866e-05,
+      "loss": 0.3063,
+      "step": 24828
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9521094414217582e-05,
+      "loss": 0.316,
+      "step": 24829
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952104418072851e-05,
+      "loss": 0.3548,
+      "step": 24830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952099394466967e-05,
+      "loss": 0.3231,
+      "step": 24831
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520943706041063e-05,
+      "loss": 0.2506,
+      "step": 24832
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952089346484271e-05,
+      "loss": 0.3304,
+      "step": 24833
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520843221074622e-05,
+      "loss": 0.2534,
+      "step": 24834
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520792974736818e-05,
+      "loss": 0.3148,
+      "step": 24835
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.95207427258293e-05,
+      "loss": 0.2683,
+      "step": 24836
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520692474352097e-05,
+      "loss": 0.3253,
+      "step": 24837
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952064222030521e-05,
+      "loss": 0.2977,
+      "step": 24838
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952059196368866e-05,
+      "loss": 0.3112,
+      "step": 24839
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520541704502455e-05,
+      "loss": 0.3182,
+      "step": 24840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520491442746614e-05,
+      "loss": 0.4111,
+      "step": 24841
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520441178421147e-05,
+      "loss": 0.3043,
+      "step": 24842
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.952039091152607e-05,
+      "loss": 0.2756,
+      "step": 24843
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520340642061392e-05,
+      "loss": 0.3004,
+      "step": 24844
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520290370027132e-05,
+      "loss": 0.321,
+      "step": 24845
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.95202400954233e-05,
+      "loss": 0.2784,
+      "step": 24846
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520189818249907e-05,
+      "loss": 0.3081,
+      "step": 24847
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520139538506976e-05,
+      "loss": 0.3078,
+      "step": 24848
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520089256194513e-05,
+      "loss": 0.4056,
+      "step": 24849
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9520038971312532e-05,
+      "loss": 0.3005,
+      "step": 24850
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951998868386105e-05,
+      "loss": 0.3329,
+      "step": 24851
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519938393840078e-05,
+      "loss": 0.2806,
+      "step": 24852
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951988810124963e-05,
+      "loss": 0.3221,
+      "step": 24853
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519837806089716e-05,
+      "loss": 0.3083,
+      "step": 24854
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519787508360356e-05,
+      "loss": 0.3184,
+      "step": 24855
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951973720806156e-05,
+      "loss": 0.2843,
+      "step": 24856
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519686905193347e-05,
+      "loss": 0.2965,
+      "step": 24857
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519636599755723e-05,
+      "loss": 0.325,
+      "step": 24858
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519586291748705e-05,
+      "loss": 0.2978,
+      "step": 24859
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519535981172304e-05,
+      "loss": 0.3148,
+      "step": 24860
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519485668026538e-05,
+      "loss": 0.3073,
+      "step": 24861
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951943535231142e-05,
+      "loss": 0.3229,
+      "step": 24862
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519385034026957e-05,
+      "loss": 0.321,
+      "step": 24863
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951933471317317e-05,
+      "loss": 0.255,
+      "step": 24864
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519284389750072e-05,
+      "loss": 0.2807,
+      "step": 24865
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951923406375767e-05,
+      "loss": 0.3391,
+      "step": 24866
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519183735195987e-05,
+      "loss": 0.3166,
+      "step": 24867
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951913340406503e-05,
+      "loss": 0.3096,
+      "step": 24868
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519083070364814e-05,
+      "loss": 0.2951,
+      "step": 24869
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9519032734095352e-05,
+      "loss": 0.3199,
+      "step": 24870
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951898239525666e-05,
+      "loss": 0.3207,
+      "step": 24871
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951893205384875e-05,
+      "loss": 0.3021,
+      "step": 24872
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518881709871637e-05,
+      "loss": 0.3076,
+      "step": 24873
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951883136332533e-05,
+      "loss": 0.3743,
+      "step": 24874
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951878101420985e-05,
+      "loss": 0.3333,
+      "step": 24875
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518730662525202e-05,
+      "loss": 0.3261,
+      "step": 24876
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518680308271406e-05,
+      "loss": 0.2858,
+      "step": 24877
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518629951448473e-05,
+      "loss": 0.3525,
+      "step": 24878
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518579592056422e-05,
+      "loss": 0.3154,
+      "step": 24879
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518529230095256e-05,
+      "loss": 0.2818,
+      "step": 24880
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518478865564998e-05,
+      "loss": 0.2531,
+      "step": 24881
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518428498465658e-05,
+      "loss": 0.3138,
+      "step": 24882
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518378128797247e-05,
+      "loss": 0.293,
+      "step": 24883
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518327756559782e-05,
+      "loss": 0.2917,
+      "step": 24884
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951827738175328e-05,
+      "loss": 0.2925,
+      "step": 24885
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518227004377747e-05,
+      "loss": 0.2761,
+      "step": 24886
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9518176624433197e-05,
+      "loss": 0.3968,
+      "step": 24887
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.951812624191965e-05,
+      "loss": 0.2885,
+      "step": 24888
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9518075856837117e-05,
+      "loss": 0.2919,
+      "step": 24889
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9518025469185607e-05,
+      "loss": 0.2861,
+      "step": 24890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517975078965145e-05,
+      "loss": 0.3074,
+      "step": 24891
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951792468617573e-05,
+      "loss": 0.3335,
+      "step": 24892
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517874290817386e-05,
+      "loss": 0.2871,
+      "step": 24893
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951782389289012e-05,
+      "loss": 0.2583,
+      "step": 24894
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951777349239395e-05,
+      "loss": 0.3234,
+      "step": 24895
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951772308932889e-05,
+      "loss": 0.3519,
+      "step": 24896
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517672683694954e-05,
+      "loss": 0.2759,
+      "step": 24897
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951762227549215e-05,
+      "loss": 0.3486,
+      "step": 24898
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517571864720498e-05,
+      "loss": 0.3079,
+      "step": 24899
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517521451380003e-05,
+      "loss": 0.3089,
+      "step": 24900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951747103547069e-05,
+      "loss": 0.2906,
+      "step": 24901
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517420616992567e-05,
+      "loss": 0.2715,
+      "step": 24902
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517370195945645e-05,
+      "loss": 0.3093,
+      "step": 24903
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517319772329943e-05,
+      "loss": 0.3354,
+      "step": 24904
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951726934614547e-05,
+      "loss": 0.2932,
+      "step": 24905
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517218917392243e-05,
+      "loss": 0.2688,
+      "step": 24906
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517168486070273e-05,
+      "loss": 0.289,
+      "step": 24907
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9517118052179573e-05,
+      "loss": 0.2978,
+      "step": 24908
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951706761572016e-05,
+      "loss": 0.3125,
+      "step": 24909
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951701717669205e-05,
+      "loss": 0.2869,
+      "step": 24910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951696673509525e-05,
+      "loss": 0.2839,
+      "step": 24911
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516916290929773e-05,
+      "loss": 0.2641,
+      "step": 24912
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516865844195638e-05,
+      "loss": 0.3132,
+      "step": 24913
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516815394892856e-05,
+      "loss": 0.2776,
+      "step": 24914
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516764943021442e-05,
+      "loss": 0.276,
+      "step": 24915
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951671448858141e-05,
+      "loss": 0.3164,
+      "step": 24916
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951666403157277e-05,
+      "loss": 0.3339,
+      "step": 24917
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516613571995536e-05,
+      "loss": 0.2881,
+      "step": 24918
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516563109849725e-05,
+      "loss": 0.313,
+      "step": 24919
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951651264513535e-05,
+      "loss": 0.2872,
+      "step": 24920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516462177852424e-05,
+      "loss": 0.3203,
+      "step": 24921
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951641170800096e-05,
+      "loss": 0.2644,
+      "step": 24922
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516361235580972e-05,
+      "loss": 0.2583,
+      "step": 24923
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516310760592475e-05,
+      "loss": 0.2809,
+      "step": 24924
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951626028303548e-05,
+      "loss": 0.2552,
+      "step": 24925
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951620980291e-05,
+      "loss": 0.2629,
+      "step": 24926
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951615932021605e-05,
+      "loss": 0.2696,
+      "step": 24927
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516108834953647e-05,
+      "loss": 0.3492,
+      "step": 24928
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9516058347122802e-05,
+      "loss": 0.2757,
+      "step": 24929
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951600785672353e-05,
+      "loss": 0.3416,
+      "step": 24930
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951595736375584e-05,
+      "loss": 0.313,
+      "step": 24931
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951590686821975e-05,
+      "loss": 0.3091,
+      "step": 24932
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515856370115268e-05,
+      "loss": 0.2809,
+      "step": 24933
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515805869442418e-05,
+      "loss": 0.2866,
+      "step": 24934
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515755366201204e-05,
+      "loss": 0.3307,
+      "step": 24935
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515704860391644e-05,
+      "loss": 0.3371,
+      "step": 24936
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951565435201375e-05,
+      "loss": 0.2846,
+      "step": 24937
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515603841067536e-05,
+      "loss": 0.2692,
+      "step": 24938
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515553327553016e-05,
+      "loss": 0.3076,
+      "step": 24939
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515502811470207e-05,
+      "loss": 0.2948,
+      "step": 24940
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515452292819115e-05,
+      "loss": 0.3147,
+      "step": 24941
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515401771599763e-05,
+      "loss": 0.3014,
+      "step": 24942
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515351247812156e-05,
+      "loss": 0.2778,
+      "step": 24943
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951530072145631e-05,
+      "loss": 0.3063,
+      "step": 24944
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951525019253224e-05,
+      "loss": 0.3061,
+      "step": 24945
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951519966103996e-05,
+      "loss": 0.3777,
+      "step": 24946
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515149126979488e-05,
+      "loss": 0.3088,
+      "step": 24947
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951509859035083e-05,
+      "loss": 0.3003,
+      "step": 24948
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9515048051154004e-05,
+      "loss": 0.2748,
+      "step": 24949
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514997509389017e-05,
+      "loss": 0.3196,
+      "step": 24950
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951494696505589e-05,
+      "loss": 0.354,
+      "step": 24951
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514896418154634e-05,
+      "loss": 0.3077,
+      "step": 24952
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514845868685262e-05,
+      "loss": 0.2637,
+      "step": 24953
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951479531664779e-05,
+      "loss": 0.2761,
+      "step": 24954
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514744762042232e-05,
+      "loss": 0.3202,
+      "step": 24955
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514694204868598e-05,
+      "loss": 0.3187,
+      "step": 24956
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514643645126905e-05,
+      "loss": 0.3341,
+      "step": 24957
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514593082817164e-05,
+      "loss": 0.2939,
+      "step": 24958
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951454251793939e-05,
+      "loss": 0.3324,
+      "step": 24959
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514491950493597e-05,
+      "loss": 0.2759,
+      "step": 24960
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.95144413804798e-05,
+      "loss": 0.2905,
+      "step": 24961
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514390807898006e-05,
+      "loss": 0.3103,
+      "step": 24962
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514340232748235e-05,
+      "loss": 0.2552,
+      "step": 24963
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.95142896550305e-05,
+      "loss": 0.347,
+      "step": 24964
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514239074744813e-05,
+      "loss": 0.3494,
+      "step": 24965
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951418849189119e-05,
+      "loss": 0.2937,
+      "step": 24966
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951413790646964e-05,
+      "loss": 0.2934,
+      "step": 24967
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9514087318480185e-05,
+      "loss": 0.3518,
+      "step": 24968
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951403672792283e-05,
+      "loss": 0.3271,
+      "step": 24969
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951398613479759e-05,
+      "loss": 0.2893,
+      "step": 24970
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513935539104482e-05,
+      "loss": 0.2964,
+      "step": 24971
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513884940843522e-05,
+      "loss": 0.2896,
+      "step": 24972
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513834340014718e-05,
+      "loss": 0.3257,
+      "step": 24973
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513783736618082e-05,
+      "loss": 0.318,
+      "step": 24974
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513733130653636e-05,
+      "loss": 0.2832,
+      "step": 24975
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513682522121388e-05,
+      "loss": 0.2756,
+      "step": 24976
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951363191102135e-05,
+      "loss": 0.284,
+      "step": 24977
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513581297353542e-05,
+      "loss": 0.3061,
+      "step": 24978
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.951353068111797e-05,
+      "loss": 0.27,
+      "step": 24979
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513480062314653e-05,
+      "loss": 0.2739,
+      "step": 24980
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513429440943605e-05,
+      "loss": 0.3533,
+      "step": 24981
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513378817004835e-05,
+      "loss": 0.2686,
+      "step": 24982
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513328190498365e-05,
+      "loss": 0.3418,
+      "step": 24983
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513277561424197e-05,
+      "loss": 0.3012,
+      "step": 24984
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513226929782353e-05,
+      "loss": 0.2899,
+      "step": 24985
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513176295572847e-05,
+      "loss": 0.3429,
+      "step": 24986
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9513125658795687e-05,
+      "loss": 0.3253,
+      "step": 24987
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951307501945089e-05,
+      "loss": 0.3266,
+      "step": 24988
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951302437753847e-05,
+      "loss": 0.3524,
+      "step": 24989
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951297373305844e-05,
+      "loss": 0.2726,
+      "step": 24990
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951292308601082e-05,
+      "loss": 0.3049,
+      "step": 24991
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951287243639561e-05,
+      "loss": 0.3116,
+      "step": 24992
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9512821784212835e-05,
+      "loss": 0.2852,
+      "step": 24993
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9512771129462506e-05,
+      "loss": 0.304,
+      "step": 24994
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951272047214463e-05,
+      "loss": 0.3178,
+      "step": 24995
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9512669812259228e-05,
+      "loss": 0.3166,
+      "step": 24996
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9512619149806317e-05,
+      "loss": 0.2966,
+      "step": 24997
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.95125684847859e-05,
+      "loss": 0.2532,
+      "step": 24998
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9512517817197997e-05,
+      "loss": 0.2876,
+      "step": 24999
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9512467147042626e-05,
+      "loss": 0.3582,
+      "step": 25000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9512416474319792e-05,
+      "loss": 0.3014,
+      "step": 25001
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951236579902951e-05,
+      "loss": 0.2597,
+      "step": 25002
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9512315121171797e-05,
+      "loss": 0.298,
+      "step": 25003
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951226444074667e-05,
+      "loss": 0.3463,
+      "step": 25004
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9512213757754135e-05,
+      "loss": 0.2707,
+      "step": 25005
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951216307219421e-05,
+      "loss": 0.2855,
+      "step": 25006
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951211238406691e-05,
+      "loss": 0.3091,
+      "step": 25007
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951206169337224e-05,
+      "loss": 0.3359,
+      "step": 25008
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9512011000110225e-05,
+      "loss": 0.2897,
+      "step": 25009
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951196030428087e-05,
+      "loss": 0.3057,
+      "step": 25010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511909605884195e-05,
+      "loss": 0.2924,
+      "step": 25011
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511858904920213e-05,
+      "loss": 0.3957,
+      "step": 25012
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511808201388934e-05,
+      "loss": 0.3206,
+      "step": 25013
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511757495290374e-05,
+      "loss": 0.249,
+      "step": 25014
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511706786624544e-05,
+      "loss": 0.2787,
+      "step": 25015
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511656075391463e-05,
+      "loss": 0.2448,
+      "step": 25016
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951160536159114e-05,
+      "loss": 0.2831,
+      "step": 25017
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951155464522359e-05,
+      "loss": 0.3109,
+      "step": 25018
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511503926288827e-05,
+      "loss": 0.3316,
+      "step": 25019
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511453204786867e-05,
+      "loss": 0.3184,
+      "step": 25020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951140248071772e-05,
+      "loss": 0.3209,
+      "step": 25021
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.95113517540814e-05,
+      "loss": 0.2825,
+      "step": 25022
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951130102487792e-05,
+      "loss": 0.2924,
+      "step": 25023
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.95112502931073e-05,
+      "loss": 0.3108,
+      "step": 25024
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511199558769544e-05,
+      "loss": 0.2791,
+      "step": 25025
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511148821864673e-05,
+      "loss": 0.3006,
+      "step": 25026
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511098082392695e-05,
+      "loss": 0.3228,
+      "step": 25027
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9511047340353632e-05,
+      "loss": 0.2972,
+      "step": 25028
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510996595747494e-05,
+      "loss": 0.3148,
+      "step": 25029
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510945848574293e-05,
+      "loss": 0.3402,
+      "step": 25030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951089509883404e-05,
+      "loss": 0.2728,
+      "step": 25031
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510844346526752e-05,
+      "loss": 0.2782,
+      "step": 25032
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510793591652443e-05,
+      "loss": 0.3045,
+      "step": 25033
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951074283421113e-05,
+      "loss": 0.2848,
+      "step": 25034
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510692074202818e-05,
+      "loss": 0.2892,
+      "step": 25035
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951064131162753e-05,
+      "loss": 0.293,
+      "step": 25036
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510590546485276e-05,
+      "loss": 0.2962,
+      "step": 25037
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510539778776067e-05,
+      "loss": 0.3035,
+      "step": 25038
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510489008499917e-05,
+      "loss": 0.2813,
+      "step": 25039
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951043823565684e-05,
+      "loss": 0.3146,
+      "step": 25040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510387460246858e-05,
+      "loss": 0.2569,
+      "step": 25041
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510336682269972e-05,
+      "loss": 0.2802,
+      "step": 25042
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510285901726208e-05,
+      "loss": 0.2816,
+      "step": 25043
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510235118615567e-05,
+      "loss": 0.3327,
+      "step": 25044
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.951018433293807e-05,
+      "loss": 0.3122,
+      "step": 25045
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510133544693734e-05,
+      "loss": 0.3164,
+      "step": 25046
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510082753882568e-05,
+      "loss": 0.3027,
+      "step": 25047
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9510031960504584e-05,
+      "loss": 0.3299,
+      "step": 25048
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509981164559795e-05,
+      "loss": 0.2579,
+      "step": 25049
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.950993036604822e-05,
+      "loss": 0.3154,
+      "step": 25050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.950987956496987e-05,
+      "loss": 0.2693,
+      "step": 25051
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509828761324764e-05,
+      "loss": 0.3405,
+      "step": 25052
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509777955112906e-05,
+      "loss": 0.2858,
+      "step": 25053
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509727146334314e-05,
+      "loss": 0.2771,
+      "step": 25054
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509676334989006e-05,
+      "loss": 0.27,
+      "step": 25055
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.950962552107699e-05,
+      "loss": 0.3284,
+      "step": 25056
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509574704598278e-05,
+      "loss": 0.3139,
+      "step": 25057
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509523885552892e-05,
+      "loss": 0.3175,
+      "step": 25058
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.950947306394084e-05,
+      "loss": 0.3026,
+      "step": 25059
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509422239762135e-05,
+      "loss": 0.324,
+      "step": 25060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509371413016793e-05,
+      "loss": 0.2828,
+      "step": 25061
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.950932058370483e-05,
+      "loss": 0.2899,
+      "step": 25062
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509269751826252e-05,
+      "loss": 0.2666,
+      "step": 25063
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509218917381082e-05,
+      "loss": 0.2991,
+      "step": 25064
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509168080369327e-05,
+      "loss": 0.306,
+      "step": 25065
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509117240791005e-05,
+      "loss": 0.3585,
+      "step": 25066
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509066398646126e-05,
+      "loss": 0.2991,
+      "step": 25067
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9509015553934706e-05,
+      "loss": 0.3029,
+      "step": 25068
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.950896470665676e-05,
+      "loss": 0.2791,
+      "step": 25069
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508913856812296e-05,
+      "loss": 0.2878,
+      "step": 25070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508863004401337e-05,
+      "loss": 0.2943,
+      "step": 25071
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508812149423888e-05,
+      "loss": 0.3062,
+      "step": 25072
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508761291879966e-05,
+      "loss": 0.2758,
+      "step": 25073
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508710431769586e-05,
+      "loss": 0.2433,
+      "step": 25074
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.950865956909276e-05,
+      "loss": 0.3078,
+      "step": 25075
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508608703849502e-05,
+      "loss": 0.3107,
+      "step": 25076
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508557836039826e-05,
+      "loss": 0.3376,
+      "step": 25077
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508506965663748e-05,
+      "loss": 0.3052,
+      "step": 25078
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508456092721276e-05,
+      "loss": 0.3598,
+      "step": 25079
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508405217212433e-05,
+      "loss": 0.277,
+      "step": 25080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508354339137222e-05,
+      "loss": 0.3121,
+      "step": 25081
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.950830345849566e-05,
+      "loss": 0.3545,
+      "step": 25082
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508252575287766e-05,
+      "loss": 0.2996,
+      "step": 25083
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.950820168951355e-05,
+      "loss": 0.328,
+      "step": 25084
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508150801173026e-05,
+      "loss": 0.3668,
+      "step": 25085
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9508099910266208e-05,
+      "loss": 0.2651,
+      "step": 25086
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9508049016793107e-05,
+      "loss": 0.3018,
+      "step": 25087
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950799812075374e-05,
+      "loss": 0.3089,
+      "step": 25088
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950794722214812e-05,
+      "loss": 0.2654,
+      "step": 25089
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507896320976263e-05,
+      "loss": 0.2917,
+      "step": 25090
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507845417238177e-05,
+      "loss": 0.3009,
+      "step": 25091
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950779451093388e-05,
+      "loss": 0.329,
+      "step": 25092
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507743602063386e-05,
+      "loss": 0.3138,
+      "step": 25093
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950769269062671e-05,
+      "loss": 0.3004,
+      "step": 25094
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507641776623857e-05,
+      "loss": 0.278,
+      "step": 25095
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950759086005485e-05,
+      "loss": 0.2948,
+      "step": 25096
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.95075399409197e-05,
+      "loss": 0.3897,
+      "step": 25097
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950748901921842e-05,
+      "loss": 0.2831,
+      "step": 25098
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507438094951024e-05,
+      "loss": 0.3035,
+      "step": 25099
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507387168117527e-05,
+      "loss": 0.291,
+      "step": 25100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507336238717944e-05,
+      "loss": 0.2818,
+      "step": 25101
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507285306752285e-05,
+      "loss": 0.3298,
+      "step": 25102
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507234372220563e-05,
+      "loss": 0.3247,
+      "step": 25103
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507183435122796e-05,
+      "loss": 0.2893,
+      "step": 25104
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507132495458996e-05,
+      "loss": 0.2725,
+      "step": 25105
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507081553229174e-05,
+      "loss": 0.2853,
+      "step": 25106
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9507030608433348e-05,
+      "loss": 0.3112,
+      "step": 25107
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950697966107153e-05,
+      "loss": 0.2935,
+      "step": 25108
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9506928711143734e-05,
+      "loss": 0.324,
+      "step": 25109
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950687775864997e-05,
+      "loss": 0.3101,
+      "step": 25110
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9506826803590263e-05,
+      "loss": 0.2966,
+      "step": 25111
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950677584596461e-05,
+      "loss": 0.2622,
+      "step": 25112
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950672488577304e-05,
+      "loss": 0.3401,
+      "step": 25113
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950667392301556e-05,
+      "loss": 0.2923,
+      "step": 25114
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950662295769218e-05,
+      "loss": 0.328,
+      "step": 25115
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9506571989802923e-05,
+      "loss": 0.3922,
+      "step": 25116
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9506521019347796e-05,
+      "loss": 0.2736,
+      "step": 25117
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9506470046326812e-05,
+      "loss": 0.2763,
+      "step": 25118
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950641907073999e-05,
+      "loss": 0.3059,
+      "step": 25119
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950636809258734e-05,
+      "loss": 0.2556,
+      "step": 25120
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9506317111868878e-05,
+      "loss": 0.3131,
+      "step": 25121
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9506266128584616e-05,
+      "loss": 0.3172,
+      "step": 25122
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950621514273457e-05,
+      "loss": 0.2924,
+      "step": 25123
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950616415431875e-05,
+      "loss": 0.2836,
+      "step": 25124
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950611316333717e-05,
+      "loss": 0.3084,
+      "step": 25125
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950606216978985e-05,
+      "loss": 0.3355,
+      "step": 25126
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9506011173676797e-05,
+      "loss": 0.3302,
+      "step": 25127
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505960174998027e-05,
+      "loss": 0.2921,
+      "step": 25128
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505909173753556e-05,
+      "loss": 0.3379,
+      "step": 25129
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505858169943392e-05,
+      "loss": 0.2992,
+      "step": 25130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505807163567552e-05,
+      "loss": 0.3149,
+      "step": 25131
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505756154626056e-05,
+      "loss": 0.3189,
+      "step": 25132
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950570514311891e-05,
+      "loss": 0.3409,
+      "step": 25133
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505654129046126e-05,
+      "loss": 0.2626,
+      "step": 25134
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505603112407726e-05,
+      "loss": 0.2722,
+      "step": 25135
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505552093203715e-05,
+      "loss": 0.3053,
+      "step": 25136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505501071434115e-05,
+      "loss": 0.3006,
+      "step": 25137
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505450047098933e-05,
+      "loss": 0.2924,
+      "step": 25138
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505399020198187e-05,
+      "loss": 0.2871,
+      "step": 25139
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950534799073189e-05,
+      "loss": 0.2888,
+      "step": 25140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505296958700052e-05,
+      "loss": 0.3572,
+      "step": 25141
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950524592410269e-05,
+      "loss": 0.2658,
+      "step": 25142
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950519488693982e-05,
+      "loss": 0.3427,
+      "step": 25143
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505143847211456e-05,
+      "loss": 0.2661,
+      "step": 25144
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9505092804917607e-05,
+      "loss": 0.2671,
+      "step": 25145
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950504176005829e-05,
+      "loss": 0.2783,
+      "step": 25146
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504990712633514e-05,
+      "loss": 0.2803,
+      "step": 25147
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504939662643297e-05,
+      "loss": 0.4267,
+      "step": 25148
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504888610087658e-05,
+      "loss": 0.3298,
+      "step": 25149
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.95048375549666e-05,
+      "loss": 0.3028,
+      "step": 25150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504786497280143e-05,
+      "loss": 0.2722,
+      "step": 25151
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.95047354370283e-05,
+      "loss": 0.3178,
+      "step": 25152
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504684374211085e-05,
+      "loss": 0.2802,
+      "step": 25153
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950463330882851e-05,
+      "loss": 0.2796,
+      "step": 25154
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504582240880592e-05,
+      "loss": 0.3019,
+      "step": 25155
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504531170367342e-05,
+      "loss": 0.2874,
+      "step": 25156
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504480097288772e-05,
+      "loss": 0.2866,
+      "step": 25157
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504429021644902e-05,
+      "loss": 0.287,
+      "step": 25158
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504377943435742e-05,
+      "loss": 0.292,
+      "step": 25159
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.95043268626613e-05,
+      "loss": 0.297,
+      "step": 25160
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504275779321605e-05,
+      "loss": 0.3136,
+      "step": 25161
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504224693416656e-05,
+      "loss": 0.2958,
+      "step": 25162
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504173604946474e-05,
+      "loss": 0.3078,
+      "step": 25163
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950412251391107e-05,
+      "loss": 0.2825,
+      "step": 25164
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950407142031046e-05,
+      "loss": 0.2823,
+      "step": 25165
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9504020324144655e-05,
+      "loss": 0.3219,
+      "step": 25166
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503969225413672e-05,
+      "loss": 0.313,
+      "step": 25167
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503918124117524e-05,
+      "loss": 0.2896,
+      "step": 25168
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950386702025622e-05,
+      "loss": 0.3057,
+      "step": 25169
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950381591382978e-05,
+      "loss": 0.3014,
+      "step": 25170
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503764804838217e-05,
+      "loss": 0.2821,
+      "step": 25171
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503713693281546e-05,
+      "loss": 0.3274,
+      "step": 25172
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503662579159774e-05,
+      "loss": 0.3247,
+      "step": 25173
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950361146247292e-05,
+      "loss": 0.3065,
+      "step": 25174
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503560343220997e-05,
+      "loss": 0.2598,
+      "step": 25175
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950350922140402e-05,
+      "loss": 0.2727,
+      "step": 25176
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503458097022e-05,
+      "loss": 0.2827,
+      "step": 25177
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503406970074952e-05,
+      "loss": 0.2881,
+      "step": 25178
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503355840562887e-05,
+      "loss": 0.3146,
+      "step": 25179
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503304708485826e-05,
+      "loss": 0.2956,
+      "step": 25180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950325357384378e-05,
+      "loss": 0.3213,
+      "step": 25181
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950320243663676e-05,
+      "loss": 0.3619,
+      "step": 25182
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950315129686478e-05,
+      "loss": 0.3724,
+      "step": 25183
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9503100154527856e-05,
+      "loss": 0.3531,
+      "step": 25184
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9503049009625998e-05,
+      "loss": 0.3024,
+      "step": 25185
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502997862159226e-05,
+      "loss": 0.3178,
+      "step": 25186
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950294671212755e-05,
+      "loss": 0.3496,
+      "step": 25187
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502895559530984e-05,
+      "loss": 0.3038,
+      "step": 25188
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950284440436954e-05,
+      "loss": 0.3439,
+      "step": 25189
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502793246643238e-05,
+      "loss": 0.3233,
+      "step": 25190
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502742086352085e-05,
+      "loss": 0.3094,
+      "step": 25191
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502690923496097e-05,
+      "loss": 0.3023,
+      "step": 25192
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950263975807529e-05,
+      "loss": 0.2859,
+      "step": 25193
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502588590089676e-05,
+      "loss": 0.3206,
+      "step": 25194
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502537419539267e-05,
+      "loss": 0.302,
+      "step": 25195
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950248624642408e-05,
+      "loss": 0.368,
+      "step": 25196
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502435070744128e-05,
+      "loss": 0.3211,
+      "step": 25197
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502383892499424e-05,
+      "loss": 0.2523,
+      "step": 25198
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950233271168998e-05,
+      "loss": 0.3168,
+      "step": 25199
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502281528315814e-05,
+      "loss": 0.2782,
+      "step": 25200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950223034237694e-05,
+      "loss": 0.3072,
+      "step": 25201
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502179153873368e-05,
+      "loss": 0.2966,
+      "step": 25202
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950212796280511e-05,
+      "loss": 0.288,
+      "step": 25203
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502076769172186e-05,
+      "loss": 0.3455,
+      "step": 25204
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9502025572974606e-05,
+      "loss": 0.2913,
+      "step": 25205
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950197437421239e-05,
+      "loss": 0.3294,
+      "step": 25206
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501923172885544e-05,
+      "loss": 0.2978,
+      "step": 25207
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950187196899408e-05,
+      "loss": 0.3038,
+      "step": 25208
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950182076253802e-05,
+      "loss": 0.2978,
+      "step": 25209
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501769553517375e-05,
+      "loss": 0.3172,
+      "step": 25210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501718341932157e-05,
+      "loss": 0.3252,
+      "step": 25211
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501667127782377e-05,
+      "loss": 0.331,
+      "step": 25212
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950161591106806e-05,
+      "loss": 0.3813,
+      "step": 25213
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501564691789204e-05,
+      "loss": 0.3267,
+      "step": 25214
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501513469945838e-05,
+      "loss": 0.2502,
+      "step": 25215
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501462245537967e-05,
+      "loss": 0.3014,
+      "step": 25216
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501411018565606e-05,
+      "loss": 0.3331,
+      "step": 25217
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501359789028772e-05,
+      "loss": 0.341,
+      "step": 25218
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501308556927477e-05,
+      "loss": 0.2486,
+      "step": 25219
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950125732226173e-05,
+      "loss": 0.3281,
+      "step": 25220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501206085031553e-05,
+      "loss": 0.3212,
+      "step": 25221
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501154845236953e-05,
+      "loss": 0.3005,
+      "step": 25222
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950110360287795e-05,
+      "loss": 0.3346,
+      "step": 25223
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9501052357954556e-05,
+      "loss": 0.3075,
+      "step": 25224
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950100111046678e-05,
+      "loss": 0.3284,
+      "step": 25225
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950094986041464e-05,
+      "loss": 0.3392,
+      "step": 25226
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950089860779815e-05,
+      "loss": 0.3217,
+      "step": 25227
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9500847352617322e-05,
+      "loss": 0.2777,
+      "step": 25228
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9500796094872173e-05,
+      "loss": 0.2909,
+      "step": 25229
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9500744834562713e-05,
+      "loss": 0.265,
+      "step": 25230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950069357168896e-05,
+      "loss": 0.3604,
+      "step": 25231
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950064230625092e-05,
+      "loss": 0.3074,
+      "step": 25232
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9500591038248616e-05,
+      "loss": 0.3403,
+      "step": 25233
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9500539767682058e-05,
+      "loss": 0.3107,
+      "step": 25234
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950048849455126e-05,
+      "loss": 0.3675,
+      "step": 25235
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9500437218856236e-05,
+      "loss": 0.3246,
+      "step": 25236
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9500385940597e-05,
+      "loss": 0.3144,
+      "step": 25237
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9500334659773563e-05,
+      "loss": 0.3305,
+      "step": 25238
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950028337638594e-05,
+      "loss": 0.306,
+      "step": 25239
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950023209043415e-05,
+      "loss": 0.3097,
+      "step": 25240
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9500180801918203e-05,
+      "loss": 0.3068,
+      "step": 25241
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950012951083811e-05,
+      "loss": 0.3536,
+      "step": 25242
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950007821719389e-05,
+      "loss": 0.329,
+      "step": 25243
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.950002692098555e-05,
+      "loss": 0.2595,
+      "step": 25244
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499975622213118e-05,
+      "loss": 0.2851,
+      "step": 25245
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499924320876587e-05,
+      "loss": 0.2904,
+      "step": 25246
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949987301697599e-05,
+      "loss": 0.2719,
+      "step": 25247
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499821710511328e-05,
+      "loss": 0.3127,
+      "step": 25248
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499770401482623e-05,
+      "loss": 0.3146,
+      "step": 25249
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499719089889885e-05,
+      "loss": 0.3164,
+      "step": 25250
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499667775733128e-05,
+      "loss": 0.365,
+      "step": 25251
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499616459012364e-05,
+      "loss": 0.3224,
+      "step": 25252
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499565139727612e-05,
+      "loss": 0.2743,
+      "step": 25253
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949951381787888e-05,
+      "loss": 0.3348,
+      "step": 25254
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499462493466187e-05,
+      "loss": 0.3205,
+      "step": 25255
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499411166489545e-05,
+      "loss": 0.2874,
+      "step": 25256
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949935983694897e-05,
+      "loss": 0.363,
+      "step": 25257
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499308504844467e-05,
+      "loss": 0.3234,
+      "step": 25258
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499257170176062e-05,
+      "loss": 0.2898,
+      "step": 25259
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949920583294376e-05,
+      "loss": 0.3482,
+      "step": 25260
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499154493147578e-05,
+      "loss": 0.3106,
+      "step": 25261
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949910315078753e-05,
+      "loss": 0.2502,
+      "step": 25262
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499051805863633e-05,
+      "loss": 0.3426,
+      "step": 25263
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9499000458375892e-05,
+      "loss": 0.306,
+      "step": 25264
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498949108324328e-05,
+      "loss": 0.2751,
+      "step": 25265
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498897755708954e-05,
+      "loss": 0.2831,
+      "step": 25266
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949884640052978e-05,
+      "loss": 0.285,
+      "step": 25267
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498795042786833e-05,
+      "loss": 0.3216,
+      "step": 25268
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949874368248011e-05,
+      "loss": 0.2765,
+      "step": 25269
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949869231960963e-05,
+      "loss": 0.265,
+      "step": 25270
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498640954175408e-05,
+      "loss": 0.2756,
+      "step": 25271
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498589586177462e-05,
+      "loss": 0.2607,
+      "step": 25272
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.94985382156158e-05,
+      "loss": 0.2805,
+      "step": 25273
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949848684249044e-05,
+      "loss": 0.2967,
+      "step": 25274
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498435466801394e-05,
+      "loss": 0.3101,
+      "step": 25275
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498384088548674e-05,
+      "loss": 0.4121,
+      "step": 25276
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498332707732298e-05,
+      "loss": 0.3558,
+      "step": 25277
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498281324352274e-05,
+      "loss": 0.3157,
+      "step": 25278
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498229938408624e-05,
+      "loss": 0.2866,
+      "step": 25279
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949817854990135e-05,
+      "loss": 0.3119,
+      "step": 25280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.949812715883048e-05,
+      "loss": 0.2755,
+      "step": 25281
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498075765196023e-05,
+      "loss": 0.3355,
+      "step": 25282
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9498024368997985e-05,
+      "loss": 0.3329,
+      "step": 25283
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949797297023639e-05,
+      "loss": 0.3116,
+      "step": 25284
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497921568911244e-05,
+      "loss": 0.3328,
+      "step": 25285
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497870165022567e-05,
+      "loss": 0.2779,
+      "step": 25286
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497818758570372e-05,
+      "loss": 0.3306,
+      "step": 25287
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949776734955467e-05,
+      "loss": 0.2448,
+      "step": 25288
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497715937975472e-05,
+      "loss": 0.3314,
+      "step": 25289
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.94976645238328e-05,
+      "loss": 0.2783,
+      "step": 25290
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497613107126662e-05,
+      "loss": 0.3089,
+      "step": 25291
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497561687857077e-05,
+      "loss": 0.2963,
+      "step": 25292
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497510266024055e-05,
+      "loss": 0.3224,
+      "step": 25293
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949745884162761e-05,
+      "loss": 0.2675,
+      "step": 25294
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497407414667754e-05,
+      "loss": 0.3423,
+      "step": 25295
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497355985144506e-05,
+      "loss": 0.3337,
+      "step": 25296
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497304553057876e-05,
+      "loss": 0.2998,
+      "step": 25297
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949725311840788e-05,
+      "loss": 0.2927,
+      "step": 25298
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497201681194528e-05,
+      "loss": 0.3121,
+      "step": 25299
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949715024141784e-05,
+      "loss": 0.2795,
+      "step": 25300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497098799077826e-05,
+      "loss": 0.2656,
+      "step": 25301
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9497047354174503e-05,
+      "loss": 0.3002,
+      "step": 25302
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496995906707878e-05,
+      "loss": 0.2832,
+      "step": 25303
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496944456677976e-05,
+      "loss": 0.3238,
+      "step": 25304
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496893004084797e-05,
+      "loss": 0.3131,
+      "step": 25305
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496841548928363e-05,
+      "loss": 0.3304,
+      "step": 25306
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496790091208693e-05,
+      "loss": 0.3558,
+      "step": 25307
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496738630925792e-05,
+      "loss": 0.2887,
+      "step": 25308
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496687168079675e-05,
+      "loss": 0.343,
+      "step": 25309
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949663570267036e-05,
+      "loss": 0.3385,
+      "step": 25310
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496584234697856e-05,
+      "loss": 0.2619,
+      "step": 25311
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949653276416218e-05,
+      "loss": 0.3275,
+      "step": 25312
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949648129106335e-05,
+      "loss": 0.3538,
+      "step": 25313
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949642981540137e-05,
+      "loss": 0.3625,
+      "step": 25314
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949637833717626e-05,
+      "loss": 0.283,
+      "step": 25315
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496326856388036e-05,
+      "loss": 0.3071,
+      "step": 25316
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496275373036705e-05,
+      "loss": 0.2924,
+      "step": 25317
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949622388712229e-05,
+      "loss": 0.2823,
+      "step": 25318
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496172398644796e-05,
+      "loss": 0.2566,
+      "step": 25319
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949612090760424e-05,
+      "loss": 0.2794,
+      "step": 25320
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949606941400064e-05,
+      "loss": 0.2965,
+      "step": 25321
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9496017917834003e-05,
+      "loss": 0.27,
+      "step": 25322
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495966419104348e-05,
+      "loss": 0.3415,
+      "step": 25323
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495914917811687e-05,
+      "loss": 0.3282,
+      "step": 25324
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495863413956036e-05,
+      "loss": 0.2997,
+      "step": 25325
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495811907537402e-05,
+      "loss": 0.3205,
+      "step": 25326
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495760398555806e-05,
+      "loss": 0.3206,
+      "step": 25327
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495708887011265e-05,
+      "loss": 0.3335,
+      "step": 25328
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495657372903784e-05,
+      "loss": 0.3265,
+      "step": 25329
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949560585623338e-05,
+      "loss": 0.2724,
+      "step": 25330
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949555433700007e-05,
+      "loss": 0.3685,
+      "step": 25331
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949550281520386e-05,
+      "loss": 0.3004,
+      "step": 25332
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495451290844777e-05,
+      "loss": 0.2996,
+      "step": 25333
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495399763922824e-05,
+      "loss": 0.2841,
+      "step": 25334
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495348234438016e-05,
+      "loss": 0.2819,
+      "step": 25335
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949529670239037e-05,
+      "loss": 0.2981,
+      "step": 25336
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495245167779902e-05,
+      "loss": 0.29,
+      "step": 25337
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949519363060662e-05,
+      "loss": 0.3517,
+      "step": 25338
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9495142090870542e-05,
+      "loss": 0.3178,
+      "step": 25339
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949509054857168e-05,
+      "loss": 0.3225,
+      "step": 25340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949503900371005e-05,
+      "loss": 0.283,
+      "step": 25341
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494987456285664e-05,
+      "loss": 0.3106,
+      "step": 25342
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494935906298536e-05,
+      "loss": 0.2442,
+      "step": 25343
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949488435374868e-05,
+      "loss": 0.3721,
+      "step": 25344
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949483279863611e-05,
+      "loss": 0.2976,
+      "step": 25345
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494781240960843e-05,
+      "loss": 0.3344,
+      "step": 25346
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494729680722888e-05,
+      "loss": 0.2783,
+      "step": 25347
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949467811792226e-05,
+      "loss": 0.3391,
+      "step": 25348
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494626552558978e-05,
+      "loss": 0.3,
+      "step": 25349
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494574984633047e-05,
+      "loss": 0.3316,
+      "step": 25350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949452341414449e-05,
+      "loss": 0.275,
+      "step": 25351
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494471841093313e-05,
+      "loss": 0.2804,
+      "step": 25352
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494420265479535e-05,
+      "loss": 0.2649,
+      "step": 25353
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949436868730317e-05,
+      "loss": 0.3044,
+      "step": 25354
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949431710656423e-05,
+      "loss": 0.3421,
+      "step": 25355
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949426552326273e-05,
+      "loss": 0.3354,
+      "step": 25356
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949421393739868e-05,
+      "loss": 0.2752,
+      "step": 25357
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494162348972102e-05,
+      "loss": 0.2837,
+      "step": 25358
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494110757983002e-05,
+      "loss": 0.2661,
+      "step": 25359
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494059164431397e-05,
+      "loss": 0.3094,
+      "step": 25360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9494007568317304e-05,
+      "loss": 0.3304,
+      "step": 25361
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493955969640733e-05,
+      "loss": 0.2811,
+      "step": 25362
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493904368401697e-05,
+      "loss": 0.271,
+      "step": 25363
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949385276460021e-05,
+      "loss": 0.3213,
+      "step": 25364
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493801158236294e-05,
+      "loss": 0.3142,
+      "step": 25365
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493749549309953e-05,
+      "loss": 0.2567,
+      "step": 25366
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493697937821206e-05,
+      "loss": 0.3687,
+      "step": 25367
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493646323770062e-05,
+      "loss": 0.2646,
+      "step": 25368
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949359470715654e-05,
+      "loss": 0.3273,
+      "step": 25369
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493543087980655e-05,
+      "loss": 0.2808,
+      "step": 25370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493491466242416e-05,
+      "loss": 0.3056,
+      "step": 25371
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949343984194184e-05,
+      "loss": 0.2957,
+      "step": 25372
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493388215078944e-05,
+      "loss": 0.2928,
+      "step": 25373
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493336585653732e-05,
+      "loss": 0.2956,
+      "step": 25374
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493284953666225e-05,
+      "loss": 0.3314,
+      "step": 25375
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949323331911644e-05,
+      "loss": 0.2798,
+      "step": 25376
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493181682004383e-05,
+      "loss": 0.2859,
+      "step": 25377
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493130042330075e-05,
+      "loss": 0.2969,
+      "step": 25378
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493078400093524e-05,
+      "loss": 0.2616,
+      "step": 25379
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9493026755294745e-05,
+      "loss": 0.3216,
+      "step": 25380
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.949297510793376e-05,
+      "loss": 0.3477,
+      "step": 25381
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949292345801057e-05,
+      "loss": 0.2607,
+      "step": 25382
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.94928718055252e-05,
+      "loss": 0.2703,
+      "step": 25383
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9492820150477662e-05,
+      "loss": 0.2928,
+      "step": 25384
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949276849286796e-05,
+      "loss": 0.2713,
+      "step": 25385
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9492716832696123e-05,
+      "loss": 0.27,
+      "step": 25386
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949266516996215e-05,
+      "loss": 0.3677,
+      "step": 25387
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9492613504666068e-05,
+      "loss": 0.2889,
+      "step": 25388
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949256183680788e-05,
+      "loss": 0.3176,
+      "step": 25389
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949251016638761e-05,
+      "loss": 0.279,
+      "step": 25390
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9492458493405266e-05,
+      "loss": 0.3093,
+      "step": 25391
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949240681786086e-05,
+      "loss": 0.3006,
+      "step": 25392
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9492355139754412e-05,
+      "loss": 0.3616,
+      "step": 25393
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9492303459085936e-05,
+      "loss": 0.2902,
+      "step": 25394
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9492251775855436e-05,
+      "loss": 0.2973,
+      "step": 25395
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949220009006294e-05,
+      "loss": 0.3348,
+      "step": 25396
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9492148401708452e-05,
+      "loss": 0.2676,
+      "step": 25397
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9492096710791985e-05,
+      "loss": 0.3419,
+      "step": 25398
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9492045017313562e-05,
+      "loss": 0.3344,
+      "step": 25399
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949199332127319e-05,
+      "loss": 0.2783,
+      "step": 25400
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949194162267088e-05,
+      "loss": 0.3075,
+      "step": 25401
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491889921506658e-05,
+      "loss": 0.2833,
+      "step": 25402
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491838217780525e-05,
+      "loss": 0.3498,
+      "step": 25403
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.94917865114925e-05,
+      "loss": 0.3207,
+      "step": 25404
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.94917348026426e-05,
+      "loss": 0.3609,
+      "step": 25405
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949168309123084e-05,
+      "loss": 0.3273,
+      "step": 25406
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491631377257226e-05,
+      "loss": 0.2866,
+      "step": 25407
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491579660721776e-05,
+      "loss": 0.3373,
+      "step": 25408
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491527941624506e-05,
+      "loss": 0.2952,
+      "step": 25409
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949147621996543e-05,
+      "loss": 0.3025,
+      "step": 25410
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491424495744555e-05,
+      "loss": 0.2572,
+      "step": 25411
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491372768961905e-05,
+      "loss": 0.3182,
+      "step": 25412
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491321039617484e-05,
+      "loss": 0.3016,
+      "step": 25413
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491269307711314e-05,
+      "loss": 0.3599,
+      "step": 25414
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949121757324341e-05,
+      "loss": 0.3367,
+      "step": 25415
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491165836213777e-05,
+      "loss": 0.335,
+      "step": 25416
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491114096622433e-05,
+      "loss": 0.2717,
+      "step": 25417
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491062354469397e-05,
+      "loss": 0.3509,
+      "step": 25418
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9491010609754676e-05,
+      "loss": 0.2723,
+      "step": 25419
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949095886247829e-05,
+      "loss": 0.3102,
+      "step": 25420
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490907112640246e-05,
+      "loss": 0.2818,
+      "step": 25421
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490855360240565e-05,
+      "loss": 0.2879,
+      "step": 25422
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490803605279256e-05,
+      "loss": 0.3256,
+      "step": 25423
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490751847756333e-05,
+      "loss": 0.2662,
+      "step": 25424
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490700087671816e-05,
+      "loss": 0.3137,
+      "step": 25425
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490648325025713e-05,
+      "loss": 0.2806,
+      "step": 25426
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949059655981804e-05,
+      "loss": 0.2731,
+      "step": 25427
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490544792048806e-05,
+      "loss": 0.3138,
+      "step": 25428
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490493021718037e-05,
+      "loss": 0.2577,
+      "step": 25429
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490441248825735e-05,
+      "loss": 0.3168,
+      "step": 25430
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490389473371917e-05,
+      "loss": 0.3197,
+      "step": 25431
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490337695356604e-05,
+      "loss": 0.3084,
+      "step": 25432
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490285914779802e-05,
+      "loss": 0.2982,
+      "step": 25433
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490234131641525e-05,
+      "loss": 0.3238,
+      "step": 25434
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949018234594179e-05,
+      "loss": 0.3495,
+      "step": 25435
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490130557680614e-05,
+      "loss": 0.2714,
+      "step": 25436
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9490078766858004e-05,
+      "loss": 0.3173,
+      "step": 25437
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.949002697347398e-05,
+      "loss": 0.2599,
+      "step": 25438
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489975177528554e-05,
+      "loss": 0.2674,
+      "step": 25439
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489923379021735e-05,
+      "loss": 0.3537,
+      "step": 25440
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489871577953543e-05,
+      "loss": 0.4013,
+      "step": 25441
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948981977432399e-05,
+      "loss": 0.2938,
+      "step": 25442
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489767968133093e-05,
+      "loss": 0.3489,
+      "step": 25443
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948971615938086e-05,
+      "loss": 0.2777,
+      "step": 25444
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948966434806731e-05,
+      "loss": 0.3137,
+      "step": 25445
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489612534192455e-05,
+      "loss": 0.3178,
+      "step": 25446
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948956071775631e-05,
+      "loss": 0.2901,
+      "step": 25447
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489508898758887e-05,
+      "loss": 0.2924,
+      "step": 25448
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489457077200202e-05,
+      "loss": 0.3524,
+      "step": 25449
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948940525308027e-05,
+      "loss": 0.3501,
+      "step": 25450
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489353426399098e-05,
+      "loss": 0.302,
+      "step": 25451
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948930159715671e-05,
+      "loss": 0.357,
+      "step": 25452
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489249765353113e-05,
+      "loss": 0.3006,
+      "step": 25453
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948919793098832e-05,
+      "loss": 0.3611,
+      "step": 25454
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489146094062353e-05,
+      "loss": 0.2766,
+      "step": 25455
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948909425457522e-05,
+      "loss": 0.3263,
+      "step": 25456
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9489042412526937e-05,
+      "loss": 0.3297,
+      "step": 25457
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488990567917514e-05,
+      "loss": 0.2736,
+      "step": 25458
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488938720746972e-05,
+      "loss": 0.3511,
+      "step": 25459
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488886871015317e-05,
+      "loss": 0.314,
+      "step": 25460
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948883501872257e-05,
+      "loss": 0.3041,
+      "step": 25461
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948878316386874e-05,
+      "loss": 0.3387,
+      "step": 25462
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488731306453844e-05,
+      "loss": 0.313,
+      "step": 25463
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488679446477895e-05,
+      "loss": 0.3199,
+      "step": 25464
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488627583940908e-05,
+      "loss": 0.2896,
+      "step": 25465
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488575718842896e-05,
+      "loss": 0.3222,
+      "step": 25466
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948852385118387e-05,
+      "loss": 0.2968,
+      "step": 25467
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948847198096385e-05,
+      "loss": 0.3105,
+      "step": 25468
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488420108182847e-05,
+      "loss": 0.2955,
+      "step": 25469
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488368232840874e-05,
+      "loss": 0.3377,
+      "step": 25470
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488316354937947e-05,
+      "loss": 0.3149,
+      "step": 25471
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948826447447408e-05,
+      "loss": 0.3683,
+      "step": 25472
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488212591449284e-05,
+      "loss": 0.2895,
+      "step": 25473
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488160705863578e-05,
+      "loss": 0.3471,
+      "step": 25474
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948810881771697e-05,
+      "loss": 0.2808,
+      "step": 25475
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488056927009477e-05,
+      "loss": 0.3182,
+      "step": 25476
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9488005033741114e-05,
+      "loss": 0.311,
+      "step": 25477
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9487953137911894e-05,
+      "loss": 0.2549,
+      "step": 25478
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9487901239521832e-05,
+      "loss": 0.3203,
+      "step": 25479
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.948784933857094e-05,
+      "loss": 0.2693,
+      "step": 25480
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9487797435059234e-05,
+      "loss": 0.3311,
+      "step": 25481
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9487745528986725e-05,
+      "loss": 0.3214,
+      "step": 25482
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9487693620353433e-05,
+      "loss": 0.2982,
+      "step": 25483
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9487641709159363e-05,
+      "loss": 0.3469,
+      "step": 25484
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948758979540454e-05,
+      "loss": 0.3111,
+      "step": 25485
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948753787908897e-05,
+      "loss": 0.2799,
+      "step": 25486
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9487485960212668e-05,
+      "loss": 0.3081,
+      "step": 25487
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948743403877565e-05,
+      "loss": 0.387,
+      "step": 25488
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9487382114777928e-05,
+      "loss": 0.2923,
+      "step": 25489
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9487330188219517e-05,
+      "loss": 0.2995,
+      "step": 25490
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948727825910043e-05,
+      "loss": 0.3746,
+      "step": 25491
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9487226327420687e-05,
+      "loss": 0.2588,
+      "step": 25492
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9487174393180294e-05,
+      "loss": 0.3199,
+      "step": 25493
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948712245637927e-05,
+      "loss": 0.2894,
+      "step": 25494
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9487070517017628e-05,
+      "loss": 0.2793,
+      "step": 25495
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948701857509538e-05,
+      "loss": 0.2797,
+      "step": 25496
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486966630612538e-05,
+      "loss": 0.3119,
+      "step": 25497
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486914683569124e-05,
+      "loss": 0.2913,
+      "step": 25498
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486862733965142e-05,
+      "loss": 0.3348,
+      "step": 25499
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486810781800618e-05,
+      "loss": 0.265,
+      "step": 25500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948675882707556e-05,
+      "loss": 0.2818,
+      "step": 25501
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486706869789972e-05,
+      "loss": 0.2795,
+      "step": 25502
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486654909943885e-05,
+      "loss": 0.3581,
+      "step": 25503
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486602947537306e-05,
+      "loss": 0.3459,
+      "step": 25504
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486550982570244e-05,
+      "loss": 0.2851,
+      "step": 25505
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948649901504272e-05,
+      "loss": 0.3281,
+      "step": 25506
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486447044954748e-05,
+      "loss": 0.2951,
+      "step": 25507
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486395072306337e-05,
+      "loss": 0.3409,
+      "step": 25508
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486343097097502e-05,
+      "loss": 0.2986,
+      "step": 25509
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486291119328262e-05,
+      "loss": 0.3114,
+      "step": 25510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486239138998625e-05,
+      "loss": 0.3045,
+      "step": 25511
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948618715610861e-05,
+      "loss": 0.2891,
+      "step": 25512
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948613517065823e-05,
+      "loss": 0.2918,
+      "step": 25513
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9486083182647493e-05,
+      "loss": 0.3037,
+      "step": 25514
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948603119207642e-05,
+      "loss": 0.3268,
+      "step": 25515
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485979198945026e-05,
+      "loss": 0.313,
+      "step": 25516
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485927203253318e-05,
+      "loss": 0.2825,
+      "step": 25517
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485875205001314e-05,
+      "loss": 0.3073,
+      "step": 25518
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485823204189027e-05,
+      "loss": 0.2713,
+      "step": 25519
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485771200816475e-05,
+      "loss": 0.2782,
+      "step": 25520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485719194883668e-05,
+      "loss": 0.2975,
+      "step": 25521
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948566718639062e-05,
+      "loss": 0.3409,
+      "step": 25522
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485615175337348e-05,
+      "loss": 0.3042,
+      "step": 25523
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485563161723862e-05,
+      "loss": 0.3123,
+      "step": 25524
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948551114555018e-05,
+      "loss": 0.3747,
+      "step": 25525
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485459126816312e-05,
+      "loss": 0.2891,
+      "step": 25526
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485407105522278e-05,
+      "loss": 0.2888,
+      "step": 25527
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485355081668086e-05,
+      "loss": 0.2894,
+      "step": 25528
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485303055253748e-05,
+      "loss": 0.3065,
+      "step": 25529
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948525102627929e-05,
+      "loss": 0.298,
+      "step": 25530
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948519899474471e-05,
+      "loss": 0.334,
+      "step": 25531
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948514696065004e-05,
+      "loss": 0.3014,
+      "step": 25532
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948509492399528e-05,
+      "loss": 0.2965,
+      "step": 25533
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9485042884780448e-05,
+      "loss": 0.28,
+      "step": 25534
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484990843005556e-05,
+      "loss": 0.2899,
+      "step": 25535
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484938798670626e-05,
+      "loss": 0.287,
+      "step": 25536
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484886751775665e-05,
+      "loss": 0.2969,
+      "step": 25537
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948483470232069e-05,
+      "loss": 0.2613,
+      "step": 25538
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484782650305707e-05,
+      "loss": 0.3169,
+      "step": 25539
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484730595730746e-05,
+      "loss": 0.2423,
+      "step": 25540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484678538595806e-05,
+      "loss": 0.3117,
+      "step": 25541
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484626478900906e-05,
+      "loss": 0.2879,
+      "step": 25542
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484574416646066e-05,
+      "loss": 0.3031,
+      "step": 25543
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484522351831292e-05,
+      "loss": 0.335,
+      "step": 25544
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.94844702844566e-05,
+      "loss": 0.2684,
+      "step": 25545
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948441821452201e-05,
+      "loss": 0.3507,
+      "step": 25546
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484366142027526e-05,
+      "loss": 0.3218,
+      "step": 25547
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948431406697317e-05,
+      "loss": 0.2568,
+      "step": 25548
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484261989358954e-05,
+      "loss": 0.3002,
+      "step": 25549
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484209909184888e-05,
+      "loss": 0.3079,
+      "step": 25550
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948415782645099e-05,
+      "loss": 0.3148,
+      "step": 25551
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484105741157277e-05,
+      "loss": 0.2979,
+      "step": 25552
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484053653303756e-05,
+      "loss": 0.2809,
+      "step": 25553
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9484001562890444e-05,
+      "loss": 0.3297,
+      "step": 25554
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483949469917358e-05,
+      "loss": 0.3515,
+      "step": 25555
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948389737438451e-05,
+      "loss": 0.2769,
+      "step": 25556
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483845276291916e-05,
+      "loss": 0.2839,
+      "step": 25557
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483793175639582e-05,
+      "loss": 0.3298,
+      "step": 25558
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948374107242753e-05,
+      "loss": 0.2745,
+      "step": 25559
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483688966655773e-05,
+      "loss": 0.3357,
+      "step": 25560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483636858324323e-05,
+      "loss": 0.2653,
+      "step": 25561
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483584747433196e-05,
+      "loss": 0.3171,
+      "step": 25562
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483532633982407e-05,
+      "loss": 0.292,
+      "step": 25563
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483480517971963e-05,
+      "loss": 0.3119,
+      "step": 25564
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948342839940189e-05,
+      "loss": 0.2845,
+      "step": 25565
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483376278272192e-05,
+      "loss": 0.2995,
+      "step": 25566
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483324154582885e-05,
+      "loss": 0.2923,
+      "step": 25567
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483272028333988e-05,
+      "loss": 0.3229,
+      "step": 25568
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948321989952551e-05,
+      "loss": 0.3069,
+      "step": 25569
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483167768157463e-05,
+      "loss": 0.2893,
+      "step": 25570
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483115634229866e-05,
+      "loss": 0.2778,
+      "step": 25571
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483063497742735e-05,
+      "loss": 0.2867,
+      "step": 25572
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9483011358696074e-05,
+      "loss": 0.2582,
+      "step": 25573
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948295921708991e-05,
+      "loss": 0.3026,
+      "step": 25574
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9482907072924253e-05,
+      "loss": 0.3055,
+      "step": 25575
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948285492619911e-05,
+      "loss": 0.3002,
+      "step": 25576
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9482802776914504e-05,
+      "loss": 0.2853,
+      "step": 25577
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9482750625070443e-05,
+      "loss": 0.292,
+      "step": 25578
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482698470666942e-05,
+      "loss": 0.3756,
+      "step": 25579
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948264631370402e-05,
+      "loss": 0.3183,
+      "step": 25580
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482594154181683e-05,
+      "loss": 0.2922,
+      "step": 25581
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482541992099954e-05,
+      "loss": 0.336,
+      "step": 25582
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482489827458838e-05,
+      "loss": 0.3346,
+      "step": 25583
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482437660258357e-05,
+      "loss": 0.2753,
+      "step": 25584
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482385490498522e-05,
+      "loss": 0.2882,
+      "step": 25585
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482333318179346e-05,
+      "loss": 0.3592,
+      "step": 25586
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482281143300844e-05,
+      "loss": 0.3129,
+      "step": 25587
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948222896586303e-05,
+      "loss": 0.2941,
+      "step": 25588
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482176785865915e-05,
+      "loss": 0.2703,
+      "step": 25589
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948212460330952e-05,
+      "loss": 0.3017,
+      "step": 25590
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482072418193855e-05,
+      "loss": 0.2611,
+      "step": 25591
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948202023051893e-05,
+      "loss": 0.2906,
+      "step": 25592
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481968040284772e-05,
+      "loss": 0.3207,
+      "step": 25593
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948191584749138e-05,
+      "loss": 0.3085,
+      "step": 25594
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481863652138773e-05,
+      "loss": 0.2917,
+      "step": 25595
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948181145422697e-05,
+      "loss": 0.3582,
+      "step": 25596
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481759253755982e-05,
+      "loss": 0.3426,
+      "step": 25597
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481707050725824e-05,
+      "loss": 0.2895,
+      "step": 25598
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481654845136506e-05,
+      "loss": 0.3004,
+      "step": 25599
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481602636988047e-05,
+      "loss": 0.334,
+      "step": 25600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481550426280457e-05,
+      "loss": 0.3396,
+      "step": 25601
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481498213013754e-05,
+      "loss": 0.3318,
+      "step": 25602
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948144599718795e-05,
+      "loss": 0.2843,
+      "step": 25603
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948139377880306e-05,
+      "loss": 0.3022,
+      "step": 25604
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481341557859096e-05,
+      "loss": 0.328,
+      "step": 25605
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481289334356074e-05,
+      "loss": 0.3007,
+      "step": 25606
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948123710829401e-05,
+      "loss": 0.3045,
+      "step": 25607
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481184879672913e-05,
+      "loss": 0.2709,
+      "step": 25608
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.94811326484928e-05,
+      "loss": 0.2764,
+      "step": 25609
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481080414753685e-05,
+      "loss": 0.2894,
+      "step": 25610
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9481028178455587e-05,
+      "loss": 0.2904,
+      "step": 25611
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948097593959851e-05,
+      "loss": 0.3117,
+      "step": 25612
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480923698182473e-05,
+      "loss": 0.2454,
+      "step": 25613
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480871454207492e-05,
+      "loss": 0.3502,
+      "step": 25614
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480819207673576e-05,
+      "loss": 0.3144,
+      "step": 25615
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480766958580747e-05,
+      "loss": 0.3475,
+      "step": 25616
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480714706929015e-05,
+      "loss": 0.3136,
+      "step": 25617
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480662452718393e-05,
+      "loss": 0.2844,
+      "step": 25618
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480610195948894e-05,
+      "loss": 0.2776,
+      "step": 25619
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480557936620536e-05,
+      "loss": 0.3533,
+      "step": 25620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480505674733328e-05,
+      "loss": 0.3481,
+      "step": 25621
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948045341028729e-05,
+      "loss": 0.2812,
+      "step": 25622
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480401143282433e-05,
+      "loss": 0.341,
+      "step": 25623
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948034887371877e-05,
+      "loss": 0.2414,
+      "step": 25624
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480296601596317e-05,
+      "loss": 0.345,
+      "step": 25625
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948024432691509e-05,
+      "loss": 0.28,
+      "step": 25626
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480192049675096e-05,
+      "loss": 0.2857,
+      "step": 25627
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480139769876357e-05,
+      "loss": 0.3132,
+      "step": 25628
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480087487518887e-05,
+      "loss": 0.3704,
+      "step": 25629
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948003520260269e-05,
+      "loss": 0.3301,
+      "step": 25630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479982915127797e-05,
+      "loss": 0.2874,
+      "step": 25631
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479930625094203e-05,
+      "loss": 0.2988,
+      "step": 25632
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479878332501932e-05,
+      "loss": 0.3294,
+      "step": 25633
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479826037351003e-05,
+      "loss": 0.3056,
+      "step": 25634
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479773739641423e-05,
+      "loss": 0.38,
+      "step": 25635
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479721439373205e-05,
+      "loss": 0.3164,
+      "step": 25636
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479669136546367e-05,
+      "loss": 0.3157,
+      "step": 25637
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479616831160925e-05,
+      "loss": 0.3293,
+      "step": 25638
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479564523216887e-05,
+      "loss": 0.3088,
+      "step": 25639
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479512212714268e-05,
+      "loss": 0.3163,
+      "step": 25640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479459899653087e-05,
+      "loss": 0.2865,
+      "step": 25641
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.947940758403336e-05,
+      "loss": 0.329,
+      "step": 25642
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.947935526585509e-05,
+      "loss": 0.3092,
+      "step": 25643
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479302945118303e-05,
+      "loss": 0.372,
+      "step": 25644
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479250621823e-05,
+      "loss": 0.3117,
+      "step": 25645
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479198295969208e-05,
+      "loss": 0.3151,
+      "step": 25646
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479145967556938e-05,
+      "loss": 0.2719,
+      "step": 25647
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.94790936365862e-05,
+      "loss": 0.26,
+      "step": 25648
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9479041303057008e-05,
+      "loss": 0.3538,
+      "step": 25649
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478988966969382e-05,
+      "loss": 0.2749,
+      "step": 25650
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478936628323334e-05,
+      "loss": 0.2772,
+      "step": 25651
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.947888428711887e-05,
+      "loss": 0.3369,
+      "step": 25652
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478831943356015e-05,
+      "loss": 0.312,
+      "step": 25653
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478779597034778e-05,
+      "loss": 0.3031,
+      "step": 25654
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478727248155176e-05,
+      "loss": 0.2927,
+      "step": 25655
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.947867489671722e-05,
+      "loss": 0.2859,
+      "step": 25656
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478622542720925e-05,
+      "loss": 0.3062,
+      "step": 25657
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478570186166307e-05,
+      "loss": 0.3083,
+      "step": 25658
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478517827053374e-05,
+      "loss": 0.3158,
+      "step": 25659
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478465465382152e-05,
+      "loss": 0.3205,
+      "step": 25660
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478413101152643e-05,
+      "loss": 0.2744,
+      "step": 25661
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478360734364867e-05,
+      "loss": 0.2907,
+      "step": 25662
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478308365018838e-05,
+      "loss": 0.3431,
+      "step": 25663
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478255993114567e-05,
+      "loss": 0.2836,
+      "step": 25664
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478203618652073e-05,
+      "loss": 0.3299,
+      "step": 25665
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478151241631366e-05,
+      "loss": 0.3637,
+      "step": 25666
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.947809886205246e-05,
+      "loss": 0.2707,
+      "step": 25667
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9478046479915374e-05,
+      "loss": 0.3311,
+      "step": 25668
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.947799409522012e-05,
+      "loss": 0.2844,
+      "step": 25669
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9477941707966706e-05,
+      "loss": 0.2704,
+      "step": 25670
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9477889318155155e-05,
+      "loss": 0.3105,
+      "step": 25671
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9477836925785477e-05,
+      "loss": 0.3403,
+      "step": 25672
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9477784530857687e-05,
+      "loss": 0.3688,
+      "step": 25673
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9477732133371798e-05,
+      "loss": 0.3285,
+      "step": 25674
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9477679733327825e-05,
+      "loss": 0.3223,
+      "step": 25675
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9477627330725778e-05,
+      "loss": 0.3183,
+      "step": 25676
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.947757492556568e-05,
+      "loss": 0.2658,
+      "step": 25677
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947752251784754e-05,
+      "loss": 0.3211,
+      "step": 25678
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947747010757137e-05,
+      "loss": 0.3069,
+      "step": 25679
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947741769473719e-05,
+      "loss": 0.3214,
+      "step": 25680
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9477365279345007e-05,
+      "loss": 0.2814,
+      "step": 25681
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9477312861394838e-05,
+      "loss": 0.307,
+      "step": 25682
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.94772604408867e-05,
+      "loss": 0.287,
+      "step": 25683
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9477208017820607e-05,
+      "loss": 0.3019,
+      "step": 25684
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9477155592196567e-05,
+      "loss": 0.3707,
+      "step": 25685
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.94771031640146e-05,
+      "loss": 0.2956,
+      "step": 25686
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9477050733274722e-05,
+      "loss": 0.2618,
+      "step": 25687
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947699829997694e-05,
+      "loss": 0.2679,
+      "step": 25688
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476945864121274e-05,
+      "loss": 0.3801,
+      "step": 25689
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947689342570773e-05,
+      "loss": 0.324,
+      "step": 25690
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476840984736336e-05,
+      "loss": 0.269,
+      "step": 25691
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476788541207097e-05,
+      "loss": 0.3377,
+      "step": 25692
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476736095120026e-05,
+      "loss": 0.3137,
+      "step": 25693
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476683646475142e-05,
+      "loss": 0.3225,
+      "step": 25694
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476631195272456e-05,
+      "loss": 0.3302,
+      "step": 25695
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947657874151198e-05,
+      "loss": 0.274,
+      "step": 25696
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476526285193734e-05,
+      "loss": 0.3348,
+      "step": 25697
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947647382631773e-05,
+      "loss": 0.3023,
+      "step": 25698
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947642136488398e-05,
+      "loss": 0.3115,
+      "step": 25699
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476368900892497e-05,
+      "loss": 0.3056,
+      "step": 25700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476316434343302e-05,
+      "loss": 0.2872,
+      "step": 25701
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476263965236402e-05,
+      "loss": 0.2932,
+      "step": 25702
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476211493571816e-05,
+      "loss": 0.3388,
+      "step": 25703
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476159019349556e-05,
+      "loss": 0.2936,
+      "step": 25704
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476106542569636e-05,
+      "loss": 0.3338,
+      "step": 25705
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947605406323207e-05,
+      "loss": 0.2676,
+      "step": 25706
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9476001581336873e-05,
+      "loss": 0.295,
+      "step": 25707
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475949096884057e-05,
+      "loss": 0.3044,
+      "step": 25708
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475896609873643e-05,
+      "loss": 0.3108,
+      "step": 25709
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475844120305636e-05,
+      "loss": 0.3088,
+      "step": 25710
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475791628180057e-05,
+      "loss": 0.276,
+      "step": 25711
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475739133496913e-05,
+      "loss": 0.3153,
+      "step": 25712
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475686636256227e-05,
+      "loss": 0.3832,
+      "step": 25713
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475634136458007e-05,
+      "loss": 0.3002,
+      "step": 25714
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475581634102273e-05,
+      "loss": 0.3515,
+      "step": 25715
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475529129189028e-05,
+      "loss": 0.3038,
+      "step": 25716
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475476621718298e-05,
+      "loss": 0.2705,
+      "step": 25717
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475424111690092e-05,
+      "loss": 0.3322,
+      "step": 25718
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475371599104426e-05,
+      "loss": 0.2744,
+      "step": 25719
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947531908396131e-05,
+      "loss": 0.3037,
+      "step": 25720
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475266566260763e-05,
+      "loss": 0.3278,
+      "step": 25721
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475214046002798e-05,
+      "loss": 0.3577,
+      "step": 25722
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475161523187426e-05,
+      "loss": 0.2746,
+      "step": 25723
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475108997814666e-05,
+      "loss": 0.2811,
+      "step": 25724
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9475056469884527e-05,
+      "loss": 0.3011,
+      "step": 25725
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947500393939703e-05,
+      "loss": 0.3074,
+      "step": 25726
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474951406352183e-05,
+      "loss": 0.3153,
+      "step": 25727
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474898870750002e-05,
+      "loss": 0.3712,
+      "step": 25728
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474846332590503e-05,
+      "loss": 0.2742,
+      "step": 25729
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474793791873697e-05,
+      "loss": 0.2848,
+      "step": 25730
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.94747412485996e-05,
+      "loss": 0.2968,
+      "step": 25731
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474688702768226e-05,
+      "loss": 0.3224,
+      "step": 25732
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474636154379588e-05,
+      "loss": 0.3286,
+      "step": 25733
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474583603433707e-05,
+      "loss": 0.3105,
+      "step": 25734
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474531049930586e-05,
+      "loss": 0.3077,
+      "step": 25735
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474478493870245e-05,
+      "loss": 0.3036,
+      "step": 25736
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474425935252702e-05,
+      "loss": 0.3487,
+      "step": 25737
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474373374077963e-05,
+      "loss": 0.3004,
+      "step": 25738
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947432081034605e-05,
+      "loss": 0.3465,
+      "step": 25739
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947426824405697e-05,
+      "loss": 0.324,
+      "step": 25740
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947421567521074e-05,
+      "loss": 0.2772,
+      "step": 25741
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474163103807377e-05,
+      "loss": 0.3809,
+      "step": 25742
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474110529846896e-05,
+      "loss": 0.3018,
+      "step": 25743
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9474057953329303e-05,
+      "loss": 0.2879,
+      "step": 25744
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947400537425462e-05,
+      "loss": 0.3321,
+      "step": 25745
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947395279262286e-05,
+      "loss": 0.2736,
+      "step": 25746
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473900208434032e-05,
+      "loss": 0.3839,
+      "step": 25747
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473847621688158e-05,
+      "loss": 0.3132,
+      "step": 25748
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473795032385244e-05,
+      "loss": 0.2867,
+      "step": 25749
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473742440525313e-05,
+      "loss": 0.4192,
+      "step": 25750
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473689846108373e-05,
+      "loss": 0.2779,
+      "step": 25751
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473637249134437e-05,
+      "loss": 0.3178,
+      "step": 25752
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947358464960353e-05,
+      "loss": 0.3194,
+      "step": 25753
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473532047515648e-05,
+      "loss": 0.3303,
+      "step": 25754
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473479442870822e-05,
+      "loss": 0.2952,
+      "step": 25755
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947342683566906e-05,
+      "loss": 0.3326,
+      "step": 25756
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947337422591037e-05,
+      "loss": 0.3264,
+      "step": 25757
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473321613594776e-05,
+      "loss": 0.2764,
+      "step": 25758
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947326899872229e-05,
+      "loss": 0.2989,
+      "step": 25759
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947321638129292e-05,
+      "loss": 0.3306,
+      "step": 25760
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473163761306688e-05,
+      "loss": 0.2742,
+      "step": 25761
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473111138763606e-05,
+      "loss": 0.304,
+      "step": 25762
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9473058513663683e-05,
+      "loss": 0.315,
+      "step": 25763
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947300588600694e-05,
+      "loss": 0.286,
+      "step": 25764
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472953255793385e-05,
+      "loss": 0.3114,
+      "step": 25765
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472900623023037e-05,
+      "loss": 0.2896,
+      "step": 25766
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472847987695914e-05,
+      "loss": 0.3059,
+      "step": 25767
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472795349812018e-05,
+      "loss": 0.3084,
+      "step": 25768
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472742709371376e-05,
+      "loss": 0.3015,
+      "step": 25769
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472690066373993e-05,
+      "loss": 0.3088,
+      "step": 25770
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472637420819885e-05,
+      "loss": 0.2928,
+      "step": 25771
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472584772709072e-05,
+      "loss": 0.3504,
+      "step": 25772
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472532122041562e-05,
+      "loss": 0.3228,
+      "step": 25773
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.947247946881737e-05,
+      "loss": 0.299,
+      "step": 25774
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472426813036512e-05,
+      "loss": 0.3026,
+      "step": 25775
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9472374154699003e-05,
+      "loss": 0.3107,
+      "step": 25776
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9472321493804857e-05,
+      "loss": 0.3625,
+      "step": 25777
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9472268830354085e-05,
+      "loss": 0.317,
+      "step": 25778
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9472216164346706e-05,
+      "loss": 0.2837,
+      "step": 25779
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9472163495782728e-05,
+      "loss": 0.3391,
+      "step": 25780
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947211082466217e-05,
+      "loss": 0.2856,
+      "step": 25781
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9472058150985047e-05,
+      "loss": 0.3187,
+      "step": 25782
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947200547475137e-05,
+      "loss": 0.3621,
+      "step": 25783
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471952795961155e-05,
+      "loss": 0.3698,
+      "step": 25784
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471900114614416e-05,
+      "loss": 0.3547,
+      "step": 25785
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471847430711165e-05,
+      "loss": 0.37,
+      "step": 25786
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471794744251417e-05,
+      "loss": 0.3026,
+      "step": 25787
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947174205523519e-05,
+      "loss": 0.3698,
+      "step": 25788
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471689363662495e-05,
+      "loss": 0.3311,
+      "step": 25789
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471636669533348e-05,
+      "loss": 0.3087,
+      "step": 25790
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471583972847757e-05,
+      "loss": 0.3024,
+      "step": 25791
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947153127360575e-05,
+      "loss": 0.3484,
+      "step": 25792
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471478571807326e-05,
+      "loss": 0.3151,
+      "step": 25793
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471425867452506e-05,
+      "loss": 0.3641,
+      "step": 25794
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471373160541306e-05,
+      "loss": 0.2979,
+      "step": 25795
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471320451073734e-05,
+      "loss": 0.2171,
+      "step": 25796
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471267739049813e-05,
+      "loss": 0.2938,
+      "step": 25797
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947121502446955e-05,
+      "loss": 0.3002,
+      "step": 25798
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471162307332962e-05,
+      "loss": 0.2553,
+      "step": 25799
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9471109587640066e-05,
+      "loss": 0.3206,
+      "step": 25800
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947105686539087e-05,
+      "loss": 0.2895,
+      "step": 25801
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947100414058539e-05,
+      "loss": 0.3015,
+      "step": 25802
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470951413223647e-05,
+      "loss": 0.2888,
+      "step": 25803
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470898683305647e-05,
+      "loss": 0.3021,
+      "step": 25804
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470845950831404e-05,
+      "loss": 0.3446,
+      "step": 25805
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947079321580094e-05,
+      "loss": 0.3309,
+      "step": 25806
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470740478214262e-05,
+      "loss": 0.2807,
+      "step": 25807
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947068773807139e-05,
+      "loss": 0.3025,
+      "step": 25808
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470634995372332e-05,
+      "loss": 0.2822,
+      "step": 25809
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470582250117107e-05,
+      "loss": 0.2634,
+      "step": 25810
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947052950230573e-05,
+      "loss": 0.3033,
+      "step": 25811
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470476751938205e-05,
+      "loss": 0.3478,
+      "step": 25812
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470423999014563e-05,
+      "loss": 0.3091,
+      "step": 25813
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470371243534803e-05,
+      "loss": 0.2997,
+      "step": 25814
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470318485498948e-05,
+      "loss": 0.2999,
+      "step": 25815
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947026572490701e-05,
+      "loss": 0.2993,
+      "step": 25816
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470212961759004e-05,
+      "loss": 0.3494,
+      "step": 25817
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947016019605494e-05,
+      "loss": 0.2881,
+      "step": 25818
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470107427794838e-05,
+      "loss": 0.2905,
+      "step": 25819
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947005465697871e-05,
+      "loss": 0.2988,
+      "step": 25820
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9470001883606567e-05,
+      "loss": 0.3192,
+      "step": 25821
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946994910767843e-05,
+      "loss": 0.3093,
+      "step": 25822
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469896329194307e-05,
+      "loss": 0.3195,
+      "step": 25823
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469843548154216e-05,
+      "loss": 0.3173,
+      "step": 25824
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469790764558166e-05,
+      "loss": 0.3432,
+      "step": 25825
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946973797840618e-05,
+      "loss": 0.3032,
+      "step": 25826
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946968518969827e-05,
+      "loss": 0.3007,
+      "step": 25827
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946963239843444e-05,
+      "loss": 0.3359,
+      "step": 25828
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469579604614716e-05,
+      "loss": 0.3186,
+      "step": 25829
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469526808239108e-05,
+      "loss": 0.3217,
+      "step": 25830
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946947400930763e-05,
+      "loss": 0.3193,
+      "step": 25831
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469421207820297e-05,
+      "loss": 0.4227,
+      "step": 25832
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469368403777126e-05,
+      "loss": 0.3262,
+      "step": 25833
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469315597178124e-05,
+      "loss": 0.3094,
+      "step": 25834
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469262788023312e-05,
+      "loss": 0.3142,
+      "step": 25835
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469209976312697e-05,
+      "loss": 0.2985,
+      "step": 25836
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469157162046302e-05,
+      "loss": 0.3059,
+      "step": 25837
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469104345224138e-05,
+      "loss": 0.3109,
+      "step": 25838
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9469051525846218e-05,
+      "loss": 0.3917,
+      "step": 25839
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468998703912556e-05,
+      "loss": 0.2865,
+      "step": 25840
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468945879423164e-05,
+      "loss": 0.2965,
+      "step": 25841
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468893052378065e-05,
+      "loss": 0.2996,
+      "step": 25842
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468840222777263e-05,
+      "loss": 0.3635,
+      "step": 25843
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468787390620777e-05,
+      "loss": 0.3392,
+      "step": 25844
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468734555908623e-05,
+      "loss": 0.2971,
+      "step": 25845
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946868171864081e-05,
+      "loss": 0.3364,
+      "step": 25846
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468628878817362e-05,
+      "loss": 0.3096,
+      "step": 25847
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946857603643828e-05,
+      "loss": 0.3061,
+      "step": 25848
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946852319150359e-05,
+      "loss": 0.3107,
+      "step": 25849
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.94684703440133e-05,
+      "loss": 0.3731,
+      "step": 25850
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946841749396742e-05,
+      "loss": 0.2892,
+      "step": 25851
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946836464136598e-05,
+      "loss": 0.3151,
+      "step": 25852
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468311786208976e-05,
+      "loss": 0.364,
+      "step": 25853
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468258928496432e-05,
+      "loss": 0.2929,
+      "step": 25854
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468206068228362e-05,
+      "loss": 0.3097,
+      "step": 25855
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468153205404784e-05,
+      "loss": 0.3587,
+      "step": 25856
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.94681003400257e-05,
+      "loss": 0.3641,
+      "step": 25857
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468047472091132e-05,
+      "loss": 0.3103,
+      "step": 25858
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467994601601096e-05,
+      "loss": 0.323,
+      "step": 25859
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467941728555605e-05,
+      "loss": 0.3024,
+      "step": 25860
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946788885295467e-05,
+      "loss": 0.249,
+      "step": 25861
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467835974798305e-05,
+      "loss": 0.3302,
+      "step": 25862
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467783094086533e-05,
+      "loss": 0.3035,
+      "step": 25863
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467730210819357e-05,
+      "loss": 0.2828,
+      "step": 25864
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467677324996798e-05,
+      "loss": 0.2916,
+      "step": 25865
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946762443661887e-05,
+      "loss": 0.3098,
+      "step": 25866
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467571545685584e-05,
+      "loss": 0.3153,
+      "step": 25867
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946751865219696e-05,
+      "loss": 0.2686,
+      "step": 25868
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467465756153004e-05,
+      "loss": 0.359,
+      "step": 25869
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467412857553736e-05,
+      "loss": 0.2901,
+      "step": 25870
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.946735995639917e-05,
+      "loss": 0.2871,
+      "step": 25871
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467307052689317e-05,
+      "loss": 0.2741,
+      "step": 25872
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467254146424194e-05,
+      "loss": 0.2844,
+      "step": 25873
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9467201237603816e-05,
+      "loss": 0.3156,
+      "step": 25874
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9467148326228197e-05,
+      "loss": 0.3275,
+      "step": 25875
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946709541229735e-05,
+      "loss": 0.2552,
+      "step": 25876
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9467042495811288e-05,
+      "loss": 0.317,
+      "step": 25877
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946698957677003e-05,
+      "loss": 0.3233,
+      "step": 25878
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466936655173585e-05,
+      "loss": 0.2904,
+      "step": 25879
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946688373102197e-05,
+      "loss": 0.3087,
+      "step": 25880
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.94668308043152e-05,
+      "loss": 0.2617,
+      "step": 25881
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466777875053283e-05,
+      "loss": 0.3479,
+      "step": 25882
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466724943236242e-05,
+      "loss": 0.3471,
+      "step": 25883
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466672008864093e-05,
+      "loss": 0.3041,
+      "step": 25884
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466619071936837e-05,
+      "loss": 0.2637,
+      "step": 25885
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.94665661324545e-05,
+      "loss": 0.3115,
+      "step": 25886
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466513190417092e-05,
+      "loss": 0.3066,
+      "step": 25887
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946646024582463e-05,
+      "loss": 0.2999,
+      "step": 25888
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466407298677122e-05,
+      "loss": 0.3201,
+      "step": 25889
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466354348974587e-05,
+      "loss": 0.3386,
+      "step": 25890
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946630139671704e-05,
+      "loss": 0.2968,
+      "step": 25891
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466248441904493e-05,
+      "loss": 0.3234,
+      "step": 25892
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466195484536964e-05,
+      "loss": 0.358,
+      "step": 25893
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466142524614462e-05,
+      "loss": 0.3563,
+      "step": 25894
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9466089562137007e-05,
+      "loss": 0.3371,
+      "step": 25895
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946603659710461e-05,
+      "loss": 0.2735,
+      "step": 25896
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465983629517278e-05,
+      "loss": 0.3308,
+      "step": 25897
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465930659375042e-05,
+      "loss": 0.3212,
+      "step": 25898
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.94658776866779e-05,
+      "loss": 0.2785,
+      "step": 25899
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465824711425874e-05,
+      "loss": 0.3323,
+      "step": 25900
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465771733618983e-05,
+      "loss": 0.2953,
+      "step": 25901
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465718753257233e-05,
+      "loss": 0.3089,
+      "step": 25902
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946566577034064e-05,
+      "loss": 0.3195,
+      "step": 25903
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946561278486922e-05,
+      "loss": 0.3161,
+      "step": 25904
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465559796842992e-05,
+      "loss": 0.3302,
+      "step": 25905
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465506806261957e-05,
+      "loss": 0.2427,
+      "step": 25906
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465453813126144e-05,
+      "loss": 0.302,
+      "step": 25907
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465400817435558e-05,
+      "loss": 0.3127,
+      "step": 25908
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465347819190215e-05,
+      "loss": 0.3241,
+      "step": 25909
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465294818390133e-05,
+      "loss": 0.2893,
+      "step": 25910
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465241815035322e-05,
+      "loss": 0.3212,
+      "step": 25911
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.94651888091258e-05,
+      "loss": 0.2978,
+      "step": 25912
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465135800661576e-05,
+      "loss": 0.2804,
+      "step": 25913
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465082789642672e-05,
+      "loss": 0.3023,
+      "step": 25914
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9465029776069093e-05,
+      "loss": 0.2886,
+      "step": 25915
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464976759940862e-05,
+      "loss": 0.3022,
+      "step": 25916
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464923741257987e-05,
+      "loss": 0.2587,
+      "step": 25917
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464870720020484e-05,
+      "loss": 0.3131,
+      "step": 25918
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946481769622837e-05,
+      "loss": 0.3226,
+      "step": 25919
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464764669881656e-05,
+      "loss": 0.3137,
+      "step": 25920
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464711640980362e-05,
+      "loss": 0.3001,
+      "step": 25921
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464658609524494e-05,
+      "loss": 0.3287,
+      "step": 25922
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946460557551407e-05,
+      "loss": 0.3182,
+      "step": 25923
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946455253894911e-05,
+      "loss": 0.2996,
+      "step": 25924
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464499499829616e-05,
+      "loss": 0.2908,
+      "step": 25925
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464446458155614e-05,
+      "loss": 0.3532,
+      "step": 25926
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464393413927112e-05,
+      "loss": 0.3424,
+      "step": 25927
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464340367144126e-05,
+      "loss": 0.252,
+      "step": 25928
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946428731780667e-05,
+      "loss": 0.2954,
+      "step": 25929
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946423426591476e-05,
+      "loss": 0.3241,
+      "step": 25930
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946418121146841e-05,
+      "loss": 0.3215,
+      "step": 25931
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946412815446763e-05,
+      "loss": 0.2871,
+      "step": 25932
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9464075094912438e-05,
+      "loss": 0.266,
+      "step": 25933
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946402203280285e-05,
+      "loss": 0.3055,
+      "step": 25934
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463968968138877e-05,
+      "loss": 0.3015,
+      "step": 25935
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463915900920532e-05,
+      "loss": 0.3335,
+      "step": 25936
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463862831147838e-05,
+      "loss": 0.2924,
+      "step": 25937
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.94638097588208e-05,
+      "loss": 0.3764,
+      "step": 25938
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463756683939434e-05,
+      "loss": 0.299,
+      "step": 25939
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946370360650376e-05,
+      "loss": 0.295,
+      "step": 25940
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946365052651378e-05,
+      "loss": 0.3447,
+      "step": 25941
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463597443969524e-05,
+      "loss": 0.3096,
+      "step": 25942
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463544358871e-05,
+      "loss": 0.3018,
+      "step": 25943
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463491271218215e-05,
+      "loss": 0.3227,
+      "step": 25944
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463438181011193e-05,
+      "loss": 0.2921,
+      "step": 25945
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463385088249943e-05,
+      "loss": 0.3071,
+      "step": 25946
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463331992934483e-05,
+      "loss": 0.3399,
+      "step": 25947
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946327889506482e-05,
+      "loss": 0.369,
+      "step": 25948
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946322579464098e-05,
+      "loss": 0.276,
+      "step": 25949
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946317269166297e-05,
+      "loss": 0.3149,
+      "step": 25950
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9463119586130806e-05,
+      "loss": 0.3282,
+      "step": 25951
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.94630664780445e-05,
+      "loss": 0.304,
+      "step": 25952
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946301336740407e-05,
+      "loss": 0.2803,
+      "step": 25953
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462960254209523e-05,
+      "loss": 0.3086,
+      "step": 25954
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462907138460883e-05,
+      "loss": 0.3768,
+      "step": 25955
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462854020158165e-05,
+      "loss": 0.3184,
+      "step": 25956
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946280089930137e-05,
+      "loss": 0.3778,
+      "step": 25957
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462747775890526e-05,
+      "loss": 0.3286,
+      "step": 25958
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462694649925642e-05,
+      "loss": 0.2792,
+      "step": 25959
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462641521406727e-05,
+      "loss": 0.3022,
+      "step": 25960
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946258839033381e-05,
+      "loss": 0.2998,
+      "step": 25961
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946253525670689e-05,
+      "loss": 0.3237,
+      "step": 25962
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462482120525985e-05,
+      "loss": 0.3079,
+      "step": 25963
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462428981791117e-05,
+      "loss": 0.2842,
+      "step": 25964
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462375840502293e-05,
+      "loss": 0.3056,
+      "step": 25965
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946232269665953e-05,
+      "loss": 0.3265,
+      "step": 25966
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462269550262842e-05,
+      "loss": 0.33,
+      "step": 25967
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462216401312245e-05,
+      "loss": 0.2869,
+      "step": 25968
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946216324980775e-05,
+      "loss": 0.283,
+      "step": 25969
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462110095749372e-05,
+      "loss": 0.3289,
+      "step": 25970
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462056939137125e-05,
+      "loss": 0.3019,
+      "step": 25971
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9462003779971027e-05,
+      "loss": 0.3774,
+      "step": 25972
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.946195061825109e-05,
+      "loss": 0.2744,
+      "step": 25973
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461897453977326e-05,
+      "loss": 0.3262,
+      "step": 25974
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461844287149754e-05,
+      "loss": 0.2957,
+      "step": 25975
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461791117768384e-05,
+      "loss": 0.2838,
+      "step": 25976
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461737945833237e-05,
+      "loss": 0.3067,
+      "step": 25977
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946168477134432e-05,
+      "loss": 0.2626,
+      "step": 25978
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461631594301648e-05,
+      "loss": 0.3591,
+      "step": 25979
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946157841470524e-05,
+      "loss": 0.2968,
+      "step": 25980
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461525232555106e-05,
+      "loss": 0.3189,
+      "step": 25981
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461472047851263e-05,
+      "loss": 0.301,
+      "step": 25982
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946141886059372e-05,
+      "loss": 0.329,
+      "step": 25983
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461365670782505e-05,
+      "loss": 0.3439,
+      "step": 25984
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461312478417618e-05,
+      "loss": 0.2789,
+      "step": 25985
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461259283499075e-05,
+      "loss": 0.3187,
+      "step": 25986
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461206086026898e-05,
+      "loss": 0.3476,
+      "step": 25987
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.94611528860011e-05,
+      "loss": 0.353,
+      "step": 25988
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9461099683421686e-05,
+      "loss": 0.2999,
+      "step": 25989
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946104647828868e-05,
+      "loss": 0.2891,
+      "step": 25990
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460993270602096e-05,
+      "loss": 0.3102,
+      "step": 25991
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460940060361944e-05,
+      "loss": 0.3666,
+      "step": 25992
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946088684756824e-05,
+      "loss": 0.2756,
+      "step": 25993
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460833632220996e-05,
+      "loss": 0.299,
+      "step": 25994
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460780414320234e-05,
+      "loss": 0.3085,
+      "step": 25995
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946072719386596e-05,
+      "loss": 0.3617,
+      "step": 25996
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460673970858188e-05,
+      "loss": 0.3507,
+      "step": 25997
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946062074529694e-05,
+      "loss": 0.341,
+      "step": 25998
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460567517182225e-05,
+      "loss": 0.3008,
+      "step": 25999
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946051428651406e-05,
+      "loss": 0.3217,
+      "step": 26000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460461053292458e-05,
+      "loss": 0.3335,
+      "step": 26001
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946040781751743e-05,
+      "loss": 0.308,
+      "step": 26002
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460354579188998e-05,
+      "loss": 0.2709,
+      "step": 26003
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946030133830717e-05,
+      "loss": 0.2734,
+      "step": 26004
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946024809487196e-05,
+      "loss": 0.3491,
+      "step": 26005
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.946019484888339e-05,
+      "loss": 0.3304,
+      "step": 26006
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460141600341468e-05,
+      "loss": 0.3195,
+      "step": 26007
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460088349246206e-05,
+      "loss": 0.2792,
+      "step": 26008
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9460035095597625e-05,
+      "loss": 0.3521,
+      "step": 26009
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459981839395734e-05,
+      "loss": 0.3482,
+      "step": 26010
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459928580640553e-05,
+      "loss": 0.3046,
+      "step": 26011
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945987531933209e-05,
+      "loss": 0.3053,
+      "step": 26012
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459822055470365e-05,
+      "loss": 0.3188,
+      "step": 26013
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459768789055387e-05,
+      "loss": 0.2726,
+      "step": 26014
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459715520087175e-05,
+      "loss": 0.3125,
+      "step": 26015
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459662248565744e-05,
+      "loss": 0.2801,
+      "step": 26016
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.94596089744911e-05,
+      "loss": 0.3258,
+      "step": 26017
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945955569786327e-05,
+      "loss": 0.3182,
+      "step": 26018
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459502418682254e-05,
+      "loss": 0.289,
+      "step": 26019
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945944913694808e-05,
+      "loss": 0.3389,
+      "step": 26020
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459395852660754e-05,
+      "loss": 0.2705,
+      "step": 26021
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459342565820292e-05,
+      "loss": 0.3209,
+      "step": 26022
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945928927642671e-05,
+      "loss": 0.3231,
+      "step": 26023
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459235984480025e-05,
+      "loss": 0.2689,
+      "step": 26024
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459182689980244e-05,
+      "loss": 0.3033,
+      "step": 26025
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459129392927387e-05,
+      "loss": 0.3362,
+      "step": 26026
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459076093321468e-05,
+      "loss": 0.3239,
+      "step": 26027
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9459022791162496e-05,
+      "loss": 0.4058,
+      "step": 26028
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458969486450493e-05,
+      "loss": 0.2927,
+      "step": 26029
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945891617918547e-05,
+      "loss": 0.3333,
+      "step": 26030
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945886286936744e-05,
+      "loss": 0.3269,
+      "step": 26031
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945880955699642e-05,
+      "loss": 0.3053,
+      "step": 26032
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945875624207242e-05,
+      "loss": 0.2959,
+      "step": 26033
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458702924595462e-05,
+      "loss": 0.3833,
+      "step": 26034
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458649604565552e-05,
+      "loss": 0.332,
+      "step": 26035
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945859628198271e-05,
+      "loss": 0.2946,
+      "step": 26036
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945854295684695e-05,
+      "loss": 0.3261,
+      "step": 26037
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458489629158282e-05,
+      "loss": 0.2878,
+      "step": 26038
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458436298916723e-05,
+      "loss": 0.2455,
+      "step": 26039
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945838296612229e-05,
+      "loss": 0.3093,
+      "step": 26040
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458329630774996e-05,
+      "loss": 0.3157,
+      "step": 26041
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458276292874853e-05,
+      "loss": 0.2495,
+      "step": 26042
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945822295242188e-05,
+      "loss": 0.3576,
+      "step": 26043
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458169609416083e-05,
+      "loss": 0.2975,
+      "step": 26044
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458116263857488e-05,
+      "loss": 0.2776,
+      "step": 26045
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458062915746095e-05,
+      "loss": 0.2687,
+      "step": 26046
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9458009565081934e-05,
+      "loss": 0.3796,
+      "step": 26047
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945795621186501e-05,
+      "loss": 0.2841,
+      "step": 26048
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945790285609534e-05,
+      "loss": 0.346,
+      "step": 26049
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9457849497772938e-05,
+      "loss": 0.2617,
+      "step": 26050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9457796136897815e-05,
+      "loss": 0.3194,
+      "step": 26051
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945774277346999e-05,
+      "loss": 0.2955,
+      "step": 26052
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9457689407489477e-05,
+      "loss": 0.2929,
+      "step": 26053
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945763603895629e-05,
+      "loss": 0.3007,
+      "step": 26054
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945758266787044e-05,
+      "loss": 0.3127,
+      "step": 26055
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945752929423195e-05,
+      "loss": 0.305,
+      "step": 26056
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9457475918040823e-05,
+      "loss": 0.2697,
+      "step": 26057
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945742253929708e-05,
+      "loss": 0.2748,
+      "step": 26058
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9457369158000735e-05,
+      "loss": 0.302,
+      "step": 26059
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9457315774151802e-05,
+      "loss": 0.3022,
+      "step": 26060
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9457262387750294e-05,
+      "loss": 0.2928,
+      "step": 26061
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945720899879623e-05,
+      "loss": 0.2957,
+      "step": 26062
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945715560728962e-05,
+      "loss": 0.2814,
+      "step": 26063
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9457102213230478e-05,
+      "loss": 0.3211,
+      "step": 26064
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.945704881661882e-05,
+      "loss": 0.319,
+      "step": 26065
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9456995417454662e-05,
+      "loss": 0.2777,
+      "step": 26066
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9456942015738013e-05,
+      "loss": 0.2842,
+      "step": 26067
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9456888611468897e-05,
+      "loss": 0.2864,
+      "step": 26068
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9456835204647316e-05,
+      "loss": 0.3202,
+      "step": 26069
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9456781795273296e-05,
+      "loss": 0.3627,
+      "step": 26070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9456728383346847e-05,
+      "loss": 0.2541,
+      "step": 26071
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945667496886798e-05,
+      "loss": 0.3166,
+      "step": 26072
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945662155183671e-05,
+      "loss": 0.2739,
+      "step": 26073
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9456568132253056e-05,
+      "loss": 0.2662,
+      "step": 26074
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9456514710117028e-05,
+      "loss": 0.3441,
+      "step": 26075
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945646128542865e-05,
+      "loss": 0.2683,
+      "step": 26076
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945640785818792e-05,
+      "loss": 0.3827,
+      "step": 26077
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9456354428394868e-05,
+      "loss": 0.2999,
+      "step": 26078
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9456300996049497e-05,
+      "loss": 0.3053,
+      "step": 26079
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945624756115183e-05,
+      "loss": 0.2952,
+      "step": 26080
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9456194123701873e-05,
+      "loss": 0.3345,
+      "step": 26081
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945614068369965e-05,
+      "loss": 0.2873,
+      "step": 26082
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9456087241145167e-05,
+      "loss": 0.2637,
+      "step": 26083
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945603379603844e-05,
+      "loss": 0.2885,
+      "step": 26084
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945598034837949e-05,
+      "loss": 0.368,
+      "step": 26085
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455926898168325e-05,
+      "loss": 0.3044,
+      "step": 26086
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945587344540496e-05,
+      "loss": 0.3803,
+      "step": 26087
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455819990089412e-05,
+      "loss": 0.3411,
+      "step": 26088
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455766532221695e-05,
+      "loss": 0.2811,
+      "step": 26089
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945571307180182e-05,
+      "loss": 0.3109,
+      "step": 26090
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455659608829805e-05,
+      "loss": 0.3365,
+      "step": 26091
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455606143305662e-05,
+      "loss": 0.352,
+      "step": 26092
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455552675229406e-05,
+      "loss": 0.2748,
+      "step": 26093
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455499204601057e-05,
+      "loss": 0.28,
+      "step": 26094
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945544573142062e-05,
+      "loss": 0.3209,
+      "step": 26095
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455392255688115e-05,
+      "loss": 0.3339,
+      "step": 26096
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455338777403556e-05,
+      "loss": 0.3065,
+      "step": 26097
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455285296566955e-05,
+      "loss": 0.2641,
+      "step": 26098
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945523181317833e-05,
+      "loss": 0.3454,
+      "step": 26099
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455178327237694e-05,
+      "loss": 0.3304,
+      "step": 26100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455124838745064e-05,
+      "loss": 0.3484,
+      "step": 26101
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945507134770045e-05,
+      "loss": 0.2683,
+      "step": 26102
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9455017854103862e-05,
+      "loss": 0.324,
+      "step": 26103
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454964357955325e-05,
+      "loss": 0.2781,
+      "step": 26104
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454910859254853e-05,
+      "loss": 0.2996,
+      "step": 26105
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945485735800245e-05,
+      "loss": 0.3577,
+      "step": 26106
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454803854198138e-05,
+      "loss": 0.2898,
+      "step": 26107
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454750347841932e-05,
+      "loss": 0.3331,
+      "step": 26108
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454696838933843e-05,
+      "loss": 0.2777,
+      "step": 26109
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454643327473887e-05,
+      "loss": 0.2838,
+      "step": 26110
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454589813462082e-05,
+      "loss": 0.2934,
+      "step": 26111
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454536296898438e-05,
+      "loss": 0.2629,
+      "step": 26112
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454482777782968e-05,
+      "loss": 0.3083,
+      "step": 26113
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945442925611569e-05,
+      "loss": 0.3235,
+      "step": 26114
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945437573189662e-05,
+      "loss": 0.329,
+      "step": 26115
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454322205125763e-05,
+      "loss": 0.3012,
+      "step": 26116
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454268675803147e-05,
+      "loss": 0.2831,
+      "step": 26117
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454215143928775e-05,
+      "loss": 0.3038,
+      "step": 26118
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945416160950267e-05,
+      "loss": 0.2883,
+      "step": 26119
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945410807252484e-05,
+      "loss": 0.293,
+      "step": 26120
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454054532995304e-05,
+      "loss": 0.3441,
+      "step": 26121
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9454000990914074e-05,
+      "loss": 0.2887,
+      "step": 26122
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453947446281164e-05,
+      "loss": 0.3185,
+      "step": 26123
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945389389909659e-05,
+      "loss": 0.3052,
+      "step": 26124
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453840349360365e-05,
+      "loss": 0.313,
+      "step": 26125
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453786797072506e-05,
+      "loss": 0.2981,
+      "step": 26126
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453733242233022e-05,
+      "loss": 0.386,
+      "step": 26127
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453679684841936e-05,
+      "loss": 0.354,
+      "step": 26128
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453626124899254e-05,
+      "loss": 0.2739,
+      "step": 26129
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453572562404998e-05,
+      "loss": 0.3383,
+      "step": 26130
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453518997359176e-05,
+      "loss": 0.3482,
+      "step": 26131
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453465429761804e-05,
+      "loss": 0.3074,
+      "step": 26132
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453411859612897e-05,
+      "loss": 0.3013,
+      "step": 26133
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453358286912473e-05,
+      "loss": 0.2835,
+      "step": 26134
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945330471166054e-05,
+      "loss": 0.3078,
+      "step": 26135
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945325113385712e-05,
+      "loss": 0.3168,
+      "step": 26136
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945319755350222e-05,
+      "loss": 0.2964,
+      "step": 26137
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945314397059586e-05,
+      "loss": 0.2791,
+      "step": 26138
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9453090385138052e-05,
+      "loss": 0.2669,
+      "step": 26139
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945303679712881e-05,
+      "loss": 0.2803,
+      "step": 26140
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945298320656815e-05,
+      "loss": 0.2969,
+      "step": 26141
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452929613456084e-05,
+      "loss": 0.2801,
+      "step": 26142
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945287601779263e-05,
+      "loss": 0.2877,
+      "step": 26143
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.94528224195778e-05,
+      "loss": 0.2881,
+      "step": 26144
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452768818811605e-05,
+      "loss": 0.2928,
+      "step": 26145
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452715215494067e-05,
+      "loss": 0.3157,
+      "step": 26146
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.94526616096252e-05,
+      "loss": 0.3221,
+      "step": 26147
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452608001205012e-05,
+      "loss": 0.3594,
+      "step": 26148
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452554390233522e-05,
+      "loss": 0.3486,
+      "step": 26149
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452500776710744e-05,
+      "loss": 0.357,
+      "step": 26150
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452447160636687e-05,
+      "loss": 0.2851,
+      "step": 26151
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452393542011373e-05,
+      "loss": 0.3498,
+      "step": 26152
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452339920834816e-05,
+      "loss": 0.3748,
+      "step": 26153
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452286297107027e-05,
+      "loss": 0.3088,
+      "step": 26154
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945223267082802e-05,
+      "loss": 0.3777,
+      "step": 26155
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452179041997817e-05,
+      "loss": 0.3254,
+      "step": 26156
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945212541061642e-05,
+      "loss": 0.333,
+      "step": 26157
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9452071776683853e-05,
+      "loss": 0.3118,
+      "step": 26158
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945201814020013e-05,
+      "loss": 0.2751,
+      "step": 26159
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9451964501165262e-05,
+      "loss": 0.3592,
+      "step": 26160
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945191085957926e-05,
+      "loss": 0.2937,
+      "step": 26161
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9451857215442147e-05,
+      "loss": 0.2823,
+      "step": 26162
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9451803568753934e-05,
+      "loss": 0.3407,
+      "step": 26163
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9451749919514633e-05,
+      "loss": 0.3136,
+      "step": 26164
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945169626772426e-05,
+      "loss": 0.3428,
+      "step": 26165
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9451642613382832e-05,
+      "loss": 0.2767,
+      "step": 26166
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9451588956490362e-05,
+      "loss": 0.3332,
+      "step": 26167
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9451535297046863e-05,
+      "loss": 0.2597,
+      "step": 26168
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945148163505235e-05,
+      "loss": 0.313,
+      "step": 26169
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.945142797050684e-05,
+      "loss": 0.3398,
+      "step": 26170
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9451374303410343e-05,
+      "loss": 0.3526,
+      "step": 26171
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9451320633762875e-05,
+      "loss": 0.3167,
+      "step": 26172
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.945126696156445e-05,
+      "loss": 0.2747,
+      "step": 26173
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.945121328681509e-05,
+      "loss": 0.3802,
+      "step": 26174
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.94511596095148e-05,
+      "loss": 0.3346,
+      "step": 26175
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9451105929663598e-05,
+      "loss": 0.3885,
+      "step": 26176
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.94510522472615e-05,
+      "loss": 0.3087,
+      "step": 26177
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450998562308516e-05,
+      "loss": 0.2837,
+      "step": 26178
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450944874804665e-05,
+      "loss": 0.2565,
+      "step": 26179
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.945089118474996e-05,
+      "loss": 0.304,
+      "step": 26180
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450837492144413e-05,
+      "loss": 0.3,
+      "step": 26181
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450783796988042e-05,
+      "loss": 0.2879,
+      "step": 26182
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450730099280863e-05,
+      "loss": 0.3012,
+      "step": 26183
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450676399022884e-05,
+      "loss": 0.3101,
+      "step": 26184
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450622696214126e-05,
+      "loss": 0.3021,
+      "step": 26185
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450568990854597e-05,
+      "loss": 0.2864,
+      "step": 26186
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.945051528294432e-05,
+      "loss": 0.2781,
+      "step": 26187
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.94504615724833e-05,
+      "loss": 0.3199,
+      "step": 26188
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.945040785947156e-05,
+      "loss": 0.3438,
+      "step": 26189
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.945035414390911e-05,
+      "loss": 0.3172,
+      "step": 26190
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450300425795964e-05,
+      "loss": 0.2968,
+      "step": 26191
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450246705132138e-05,
+      "loss": 0.3143,
+      "step": 26192
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450192981917645e-05,
+      "loss": 0.2966,
+      "step": 26193
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450139256152503e-05,
+      "loss": 0.3645,
+      "step": 26194
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450085527836724e-05,
+      "loss": 0.3309,
+      "step": 26195
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450031796970322e-05,
+      "loss": 0.3141,
+      "step": 26196
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449978063553314e-05,
+      "loss": 0.3586,
+      "step": 26197
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944992432758571e-05,
+      "loss": 0.3486,
+      "step": 26198
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449870589067528e-05,
+      "loss": 0.2655,
+      "step": 26199
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449816847998783e-05,
+      "loss": 0.2763,
+      "step": 26200
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449763104379487e-05,
+      "loss": 0.2796,
+      "step": 26201
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944970935820966e-05,
+      "loss": 0.298,
+      "step": 26202
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449655609489304e-05,
+      "loss": 0.3092,
+      "step": 26203
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449601858218448e-05,
+      "loss": 0.3087,
+      "step": 26204
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449548104397097e-05,
+      "loss": 0.2666,
+      "step": 26205
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449494348025273e-05,
+      "loss": 0.3877,
+      "step": 26206
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449440589102985e-05,
+      "loss": 0.2866,
+      "step": 26207
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449386827630247e-05,
+      "loss": 0.3198,
+      "step": 26208
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449333063607076e-05,
+      "loss": 0.3107,
+      "step": 26209
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449279297033485e-05,
+      "loss": 0.2929,
+      "step": 26210
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944922552790949e-05,
+      "loss": 0.2817,
+      "step": 26211
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449171756235102e-05,
+      "loss": 0.3142,
+      "step": 26212
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449117982010344e-05,
+      "loss": 0.2862,
+      "step": 26213
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944906420523522e-05,
+      "loss": 0.3661,
+      "step": 26214
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449010425909752e-05,
+      "loss": 0.3047,
+      "step": 26215
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448956644033955e-05,
+      "loss": 0.287,
+      "step": 26216
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448902859607837e-05,
+      "loss": 0.2892,
+      "step": 26217
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944884907263141e-05,
+      "loss": 0.32,
+      "step": 26218
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448795283104705e-05,
+      "loss": 0.3546,
+      "step": 26219
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944874149102772e-05,
+      "loss": 0.2904,
+      "step": 26220
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448687696400477e-05,
+      "loss": 0.3014,
+      "step": 26221
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944863389922299e-05,
+      "loss": 0.3157,
+      "step": 26222
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448580099495272e-05,
+      "loss": 0.3084,
+      "step": 26223
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448526297217335e-05,
+      "loss": 0.2988,
+      "step": 26224
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448472492389202e-05,
+      "loss": 0.2999,
+      "step": 26225
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944841868501088e-05,
+      "loss": 0.2911,
+      "step": 26226
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448364875082386e-05,
+      "loss": 0.2896,
+      "step": 26227
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944831106260373e-05,
+      "loss": 0.3459,
+      "step": 26228
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448257247574934e-05,
+      "loss": 0.3282,
+      "step": 26229
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448203429996007e-05,
+      "loss": 0.3346,
+      "step": 26230
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944814960986697e-05,
+      "loss": 0.3463,
+      "step": 26231
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944809578718783e-05,
+      "loss": 0.2843,
+      "step": 26232
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9448041961958604e-05,
+      "loss": 0.3102,
+      "step": 26233
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944798813417931e-05,
+      "loss": 0.3159,
+      "step": 26234
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944793430384996e-05,
+      "loss": 0.2804,
+      "step": 26235
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447880470970565e-05,
+      "loss": 0.2866,
+      "step": 26236
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447826635541144e-05,
+      "loss": 0.2914,
+      "step": 26237
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447772797561714e-05,
+      "loss": 0.3336,
+      "step": 26238
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447718957032284e-05,
+      "loss": 0.2461,
+      "step": 26239
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447665113952868e-05,
+      "loss": 0.2955,
+      "step": 26240
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447611268323487e-05,
+      "loss": 0.2886,
+      "step": 26241
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447557420144146e-05,
+      "loss": 0.2847,
+      "step": 26242
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944750356941487e-05,
+      "loss": 0.3272,
+      "step": 26243
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447449716135667e-05,
+      "loss": 0.3616,
+      "step": 26244
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447395860306556e-05,
+      "loss": 0.3473,
+      "step": 26245
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447342001927546e-05,
+      "loss": 0.3058,
+      "step": 26246
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447288140998653e-05,
+      "loss": 0.2666,
+      "step": 26247
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944723427751989e-05,
+      "loss": 0.266,
+      "step": 26248
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447180411491283e-05,
+      "loss": 0.297,
+      "step": 26249
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447126542912832e-05,
+      "loss": 0.336,
+      "step": 26250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944707267178456e-05,
+      "loss": 0.2998,
+      "step": 26251
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9447018798106476e-05,
+      "loss": 0.2779,
+      "step": 26252
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446964921878598e-05,
+      "loss": 0.2746,
+      "step": 26253
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944691104310094e-05,
+      "loss": 0.2527,
+      "step": 26254
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944685716177352e-05,
+      "loss": 0.2897,
+      "step": 26255
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446803277896348e-05,
+      "loss": 0.2937,
+      "step": 26256
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446749391469438e-05,
+      "loss": 0.3302,
+      "step": 26257
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446695502492803e-05,
+      "loss": 0.2889,
+      "step": 26258
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446641610966467e-05,
+      "loss": 0.2946,
+      "step": 26259
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446587716890433e-05,
+      "loss": 0.3066,
+      "step": 26260
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446533820264725e-05,
+      "loss": 0.349,
+      "step": 26261
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446479921089353e-05,
+      "loss": 0.3919,
+      "step": 26262
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446426019364327e-05,
+      "loss": 0.2655,
+      "step": 26263
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944637211508967e-05,
+      "loss": 0.3535,
+      "step": 26264
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446318208265395e-05,
+      "loss": 0.3356,
+      "step": 26265
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9446264298891512e-05,
+      "loss": 0.2787,
+      "step": 26266
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944621038696804e-05,
+      "loss": 0.2887,
+      "step": 26267
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.944615647249499e-05,
+      "loss": 0.2814,
+      "step": 26268
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9446102555472378e-05,
+      "loss": 0.3535,
+      "step": 26269
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944604863590022e-05,
+      "loss": 0.2839,
+      "step": 26270
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944599471377853e-05,
+      "loss": 0.2759,
+      "step": 26271
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9445940789107317e-05,
+      "loss": 0.302,
+      "step": 26272
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9445886861886606e-05,
+      "loss": 0.2894,
+      "step": 26273
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9445832932116404e-05,
+      "loss": 0.3302,
+      "step": 26274
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9445778999796728e-05,
+      "loss": 0.336,
+      "step": 26275
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944572506492759e-05,
+      "loss": 0.3057,
+      "step": 26276
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944567112750901e-05,
+      "loss": 0.3421,
+      "step": 26277
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9445617187540998e-05,
+      "loss": 0.3186,
+      "step": 26278
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9445563245023566e-05,
+      "loss": 0.2875,
+      "step": 26279
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944550929995674e-05,
+      "loss": 0.3047,
+      "step": 26280
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944545535234052e-05,
+      "loss": 0.3028,
+      "step": 26281
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9445401402174928e-05,
+      "loss": 0.3042,
+      "step": 26282
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9445347449459982e-05,
+      "loss": 0.2785,
+      "step": 26283
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944529349419569e-05,
+      "loss": 0.3118,
+      "step": 26284
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944523953638207e-05,
+      "loss": 0.3326,
+      "step": 26285
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9445185576019135e-05,
+      "loss": 0.2944,
+      "step": 26286
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.94451316131069e-05,
+      "loss": 0.2976,
+      "step": 26287
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944507764764538e-05,
+      "loss": 0.331,
+      "step": 26288
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944502367963459e-05,
+      "loss": 0.3202,
+      "step": 26289
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444969709074543e-05,
+      "loss": 0.2799,
+      "step": 26290
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444915735965256e-05,
+      "loss": 0.2849,
+      "step": 26291
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444861760306742e-05,
+      "loss": 0.2987,
+      "step": 26292
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444807782099015e-05,
+      "loss": 0.3789,
+      "step": 26293
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444753801342088e-05,
+      "loss": 0.2948,
+      "step": 26294
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944469981803598e-05,
+      "loss": 0.341,
+      "step": 26295
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444645832180706e-05,
+      "loss": 0.3432,
+      "step": 26296
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444591843776274e-05,
+      "loss": 0.3363,
+      "step": 26297
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444537852822703e-05,
+      "loss": 0.3082,
+      "step": 26298
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944448385932001e-05,
+      "loss": 0.3193,
+      "step": 26299
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444429863268203e-05,
+      "loss": 0.3222,
+      "step": 26300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.94443758646673e-05,
+      "loss": 0.2652,
+      "step": 26301
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944432186351732e-05,
+      "loss": 0.3481,
+      "step": 26302
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444267859818272e-05,
+      "loss": 0.321,
+      "step": 26303
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944421385357017e-05,
+      "loss": 0.3197,
+      "step": 26304
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444159844773033e-05,
+      "loss": 0.2708,
+      "step": 26305
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944410583342687e-05,
+      "loss": 0.2492,
+      "step": 26306
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444051819531703e-05,
+      "loss": 0.2843,
+      "step": 26307
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443997803087537e-05,
+      "loss": 0.3026,
+      "step": 26308
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443943784094394e-05,
+      "loss": 0.2683,
+      "step": 26309
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944388976255229e-05,
+      "loss": 0.3284,
+      "step": 26310
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944383573846123e-05,
+      "loss": 0.333,
+      "step": 26311
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443781711821237e-05,
+      "loss": 0.313,
+      "step": 26312
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944372768263232e-05,
+      "loss": 0.3173,
+      "step": 26313
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443673650894507e-05,
+      "loss": 0.2958,
+      "step": 26314
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443619616607795e-05,
+      "loss": 0.3267,
+      "step": 26315
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443565579772205e-05,
+      "loss": 0.2973,
+      "step": 26316
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443511540387753e-05,
+      "loss": 0.308,
+      "step": 26317
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443457498454454e-05,
+      "loss": 0.3158,
+      "step": 26318
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944340345397232e-05,
+      "loss": 0.2813,
+      "step": 26319
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443349406941372e-05,
+      "loss": 0.333,
+      "step": 26320
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443295357361615e-05,
+      "loss": 0.3041,
+      "step": 26321
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443241305233075e-05,
+      "loss": 0.3137,
+      "step": 26322
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443187250555752e-05,
+      "loss": 0.2729,
+      "step": 26323
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443133193329674e-05,
+      "loss": 0.3078,
+      "step": 26324
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944307913355485e-05,
+      "loss": 0.2762,
+      "step": 26325
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443025071231288e-05,
+      "loss": 0.3622,
+      "step": 26326
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9442971006359016e-05,
+      "loss": 0.3122,
+      "step": 26327
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944291693893804e-05,
+      "loss": 0.3158,
+      "step": 26328
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944286286896838e-05,
+      "loss": 0.2708,
+      "step": 26329
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9442808796450042e-05,
+      "loss": 0.3175,
+      "step": 26330
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9442754721383047e-05,
+      "loss": 0.3029,
+      "step": 26331
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9442700643767408e-05,
+      "loss": 0.2847,
+      "step": 26332
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944264656360314e-05,
+      "loss": 0.2652,
+      "step": 26333
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944259248089026e-05,
+      "loss": 0.3164,
+      "step": 26334
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9442538395628784e-05,
+      "loss": 0.2611,
+      "step": 26335
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9442484307818717e-05,
+      "loss": 0.3099,
+      "step": 26336
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944243021746008e-05,
+      "loss": 0.3107,
+      "step": 26337
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9442376124552888e-05,
+      "loss": 0.2947,
+      "step": 26338
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9442322029097153e-05,
+      "loss": 0.3018,
+      "step": 26339
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944226793109289e-05,
+      "loss": 0.3176,
+      "step": 26340
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9442213830540115e-05,
+      "loss": 0.313,
+      "step": 26341
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9442159727438844e-05,
+      "loss": 0.3634,
+      "step": 26342
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944210562178909e-05,
+      "loss": 0.3429,
+      "step": 26343
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944205151359087e-05,
+      "loss": 0.3307,
+      "step": 26344
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944199740284419e-05,
+      "loss": 0.3172,
+      "step": 26345
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441943289549077e-05,
+      "loss": 0.2912,
+      "step": 26346
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441889173705533e-05,
+      "loss": 0.3241,
+      "step": 26347
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441835055313586e-05,
+      "loss": 0.2997,
+      "step": 26348
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944178093437324e-05,
+      "loss": 0.2571,
+      "step": 26349
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441726810884514e-05,
+      "loss": 0.3016,
+      "step": 26350
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944167268484742e-05,
+      "loss": 0.3314,
+      "step": 26351
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441618556261976e-05,
+      "loss": 0.3167,
+      "step": 26352
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441564425128193e-05,
+      "loss": 0.3082,
+      "step": 26353
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944151029144609e-05,
+      "loss": 0.3007,
+      "step": 26354
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944145615521568e-05,
+      "loss": 0.3652,
+      "step": 26355
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441402016436975e-05,
+      "loss": 0.2854,
+      "step": 26356
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441347875109994e-05,
+      "loss": 0.2878,
+      "step": 26357
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944129373123474e-05,
+      "loss": 0.3131,
+      "step": 26358
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441239584811248e-05,
+      "loss": 0.317,
+      "step": 26359
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441185435839517e-05,
+      "loss": 0.263,
+      "step": 26360
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441131284319566e-05,
+      "loss": 0.3255,
+      "step": 26361
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944107713025141e-05,
+      "loss": 0.3014,
+      "step": 26362
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944102297363506e-05,
+      "loss": 0.2989,
+      "step": 26363
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944096881447054e-05,
+      "loss": 0.3026,
+      "step": 26364
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944091465275785e-05,
+      "loss": 0.2553,
+      "step": 26365
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.944086048849702e-05,
+      "loss": 0.2929,
+      "step": 26366
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9440806321688055e-05,
+      "loss": 0.3027,
+      "step": 26367
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440752152330972e-05,
+      "loss": 0.3077,
+      "step": 26368
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440697980425786e-05,
+      "loss": 0.3186,
+      "step": 26369
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440643805972513e-05,
+      "loss": 0.2804,
+      "step": 26370
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440589628971164e-05,
+      "loss": 0.3278,
+      "step": 26371
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440535449421757e-05,
+      "loss": 0.2804,
+      "step": 26372
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440481267324306e-05,
+      "loss": 0.267,
+      "step": 26373
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440427082678823e-05,
+      "loss": 0.2702,
+      "step": 26374
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440372895485327e-05,
+      "loss": 0.2619,
+      "step": 26375
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440318705743827e-05,
+      "loss": 0.2872,
+      "step": 26376
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440264513454343e-05,
+      "loss": 0.3312,
+      "step": 26377
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.944021031861689e-05,
+      "loss": 0.2861,
+      "step": 26378
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440156121231472e-05,
+      "loss": 0.3135,
+      "step": 26379
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.944010192129812e-05,
+      "loss": 0.3771,
+      "step": 26380
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9440047718816837e-05,
+      "loss": 0.3531,
+      "step": 26381
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943999351378764e-05,
+      "loss": 0.3501,
+      "step": 26382
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439939306210545e-05,
+      "loss": 0.2605,
+      "step": 26383
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943988509608557e-05,
+      "loss": 0.3341,
+      "step": 26384
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439830883412723e-05,
+      "loss": 0.3356,
+      "step": 26385
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943977666819202e-05,
+      "loss": 0.3138,
+      "step": 26386
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439722450423478e-05,
+      "loss": 0.2998,
+      "step": 26387
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439668230107114e-05,
+      "loss": 0.3715,
+      "step": 26388
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439614007242936e-05,
+      "loss": 0.2678,
+      "step": 26389
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439559781830964e-05,
+      "loss": 0.3288,
+      "step": 26390
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439505553871208e-05,
+      "loss": 0.3334,
+      "step": 26391
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943945132336369e-05,
+      "loss": 0.3314,
+      "step": 26392
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439397090308417e-05,
+      "loss": 0.284,
+      "step": 26393
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439342854705402e-05,
+      "loss": 0.3012,
+      "step": 26394
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943928861655467e-05,
+      "loss": 0.3784,
+      "step": 26395
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943923437585623e-05,
+      "loss": 0.2906,
+      "step": 26396
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439180132610096e-05,
+      "loss": 0.2987,
+      "step": 26397
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439125886816282e-05,
+      "loss": 0.3209,
+      "step": 26398
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439071638474803e-05,
+      "loss": 0.3313,
+      "step": 26399
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9439017387585676e-05,
+      "loss": 0.3182,
+      "step": 26400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9438963134148914e-05,
+      "loss": 0.3224,
+      "step": 26401
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943890887816453e-05,
+      "loss": 0.3033,
+      "step": 26402
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9438854619632545e-05,
+      "loss": 0.3109,
+      "step": 26403
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9438800358552964e-05,
+      "loss": 0.3012,
+      "step": 26404
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943874609492581e-05,
+      "loss": 0.3234,
+      "step": 26405
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9438691828751092e-05,
+      "loss": 0.311,
+      "step": 26406
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943863756002883e-05,
+      "loss": 0.2875,
+      "step": 26407
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943858328875903e-05,
+      "loss": 0.2751,
+      "step": 26408
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9438529014941715e-05,
+      "loss": 0.305,
+      "step": 26409
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.94384747385769e-05,
+      "loss": 0.3307,
+      "step": 26410
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9438420459664592e-05,
+      "loss": 0.3072,
+      "step": 26411
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9438366178204813e-05,
+      "loss": 0.3846,
+      "step": 26412
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9438311894197575e-05,
+      "loss": 0.322,
+      "step": 26413
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943825760764289e-05,
+      "loss": 0.3105,
+      "step": 26414
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9438203318540778e-05,
+      "loss": 0.2977,
+      "step": 26415
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943814902689125e-05,
+      "loss": 0.3742,
+      "step": 26416
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943809473269432e-05,
+      "loss": 0.2936,
+      "step": 26417
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9438040435950005e-05,
+      "loss": 0.3151,
+      "step": 26418
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943798613665832e-05,
+      "loss": 0.345,
+      "step": 26419
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437931834819278e-05,
+      "loss": 0.2777,
+      "step": 26420
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437877530432892e-05,
+      "loss": 0.3118,
+      "step": 26421
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437823223499182e-05,
+      "loss": 0.3034,
+      "step": 26422
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437768914018157e-05,
+      "loss": 0.3188,
+      "step": 26423
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437714601989833e-05,
+      "loss": 0.293,
+      "step": 26424
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943766028741423e-05,
+      "loss": 0.2772,
+      "step": 26425
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437605970291354e-05,
+      "loss": 0.337,
+      "step": 26426
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437551650621227e-05,
+      "loss": 0.3072,
+      "step": 26427
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943749732840386e-05,
+      "loss": 0.2567,
+      "step": 26428
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437443003639266e-05,
+      "loss": 0.3149,
+      "step": 26429
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437388676327465e-05,
+      "loss": 0.2446,
+      "step": 26430
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437334346468467e-05,
+      "loss": 0.3158,
+      "step": 26431
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437280014062288e-05,
+      "loss": 0.3483,
+      "step": 26432
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437225679108946e-05,
+      "loss": 0.3593,
+      "step": 26433
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437171341608447e-05,
+      "loss": 0.271,
+      "step": 26434
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437117001560818e-05,
+      "loss": 0.2725,
+      "step": 26435
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9437062658966063e-05,
+      "loss": 0.3454,
+      "step": 26436
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.94370083138242e-05,
+      "loss": 0.268,
+      "step": 26437
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436953966135245e-05,
+      "loss": 0.3265,
+      "step": 26438
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436899615899215e-05,
+      "loss": 0.2874,
+      "step": 26439
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943684526311612e-05,
+      "loss": 0.2746,
+      "step": 26440
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436790907785975e-05,
+      "loss": 0.2394,
+      "step": 26441
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436736549908797e-05,
+      "loss": 0.2976,
+      "step": 26442
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436682189484598e-05,
+      "loss": 0.3124,
+      "step": 26443
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436627826513398e-05,
+      "loss": 0.2685,
+      "step": 26444
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436573460995204e-05,
+      "loss": 0.2899,
+      "step": 26445
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436519092930036e-05,
+      "loss": 0.3013,
+      "step": 26446
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943646472231791e-05,
+      "loss": 0.3914,
+      "step": 26447
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436410349158836e-05,
+      "loss": 0.3068,
+      "step": 26448
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943635597345283e-05,
+      "loss": 0.3174,
+      "step": 26449
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943630159519991e-05,
+      "loss": 0.2894,
+      "step": 26450
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436247214400086e-05,
+      "loss": 0.3197,
+      "step": 26451
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436192831053376e-05,
+      "loss": 0.3354,
+      "step": 26452
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943613844515979e-05,
+      "loss": 0.2938,
+      "step": 26453
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943608405671935e-05,
+      "loss": 0.3319,
+      "step": 26454
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9436029665732065e-05,
+      "loss": 0.3258,
+      "step": 26455
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9435975272197954e-05,
+      "loss": 0.3164,
+      "step": 26456
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9435920876117026e-05,
+      "loss": 0.3549,
+      "step": 26457
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.94358664774893e-05,
+      "loss": 0.3404,
+      "step": 26458
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943581207631479e-05,
+      "loss": 0.2968,
+      "step": 26459
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.943575767259351e-05,
+      "loss": 0.2764,
+      "step": 26460
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9435703266325474e-05,
+      "loss": 0.2376,
+      "step": 26461
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.94356488575107e-05,
+      "loss": 0.3324,
+      "step": 26462
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.94355944461492e-05,
+      "loss": 0.331,
+      "step": 26463
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9435540032240984e-05,
+      "loss": 0.2911,
+      "step": 26464
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9435485615786078e-05,
+      "loss": 0.2845,
+      "step": 26465
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9435431196784485e-05,
+      "loss": 0.3253,
+      "step": 26466
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9435376775236227e-05,
+      "loss": 0.3177,
+      "step": 26467
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9435322351141317e-05,
+      "loss": 0.3032,
+      "step": 26468
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943526792449977e-05,
+      "loss": 0.3092,
+      "step": 26469
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.94352134953116e-05,
+      "loss": 0.2832,
+      "step": 26470
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943515906357682e-05,
+      "loss": 0.2933,
+      "step": 26471
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9435104629295448e-05,
+      "loss": 0.253,
+      "step": 26472
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9435050192467498e-05,
+      "loss": 0.3002,
+      "step": 26473
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943499575309298e-05,
+      "loss": 0.3135,
+      "step": 26474
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434941311171917e-05,
+      "loss": 0.3221,
+      "step": 26475
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434886866704323e-05,
+      "loss": 0.2977,
+      "step": 26476
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434832419690202e-05,
+      "loss": 0.3209,
+      "step": 26477
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943477797012958e-05,
+      "loss": 0.3164,
+      "step": 26478
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943472351802246e-05,
+      "loss": 0.2716,
+      "step": 26479
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943466906336887e-05,
+      "loss": 0.3142,
+      "step": 26480
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434614606168815e-05,
+      "loss": 0.2716,
+      "step": 26481
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434560146422316e-05,
+      "loss": 0.3218,
+      "step": 26482
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434505684129384e-05,
+      "loss": 0.3168,
+      "step": 26483
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434451219290034e-05,
+      "loss": 0.3063,
+      "step": 26484
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434396751904287e-05,
+      "loss": 0.3356,
+      "step": 26485
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434342281972148e-05,
+      "loss": 0.3389,
+      "step": 26486
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434287809493635e-05,
+      "loss": 0.3289,
+      "step": 26487
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434233334468765e-05,
+      "loss": 0.2923,
+      "step": 26488
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943417885689755e-05,
+      "loss": 0.3442,
+      "step": 26489
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943412437678001e-05,
+      "loss": 0.3042,
+      "step": 26490
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943406989411615e-05,
+      "loss": 0.3088,
+      "step": 26491
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9434015408905997e-05,
+      "loss": 0.3102,
+      "step": 26492
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433960921149554e-05,
+      "loss": 0.2725,
+      "step": 26493
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433906430846843e-05,
+      "loss": 0.2563,
+      "step": 26494
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433851937997875e-05,
+      "loss": 0.3195,
+      "step": 26495
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943379744260267e-05,
+      "loss": 0.2855,
+      "step": 26496
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433742944661234e-05,
+      "loss": 0.2705,
+      "step": 26497
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943368844417359e-05,
+      "loss": 0.3112,
+      "step": 26498
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433633941139748e-05,
+      "loss": 0.3247,
+      "step": 26499
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433579435559723e-05,
+      "loss": 0.2872,
+      "step": 26500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433524927433533e-05,
+      "loss": 0.357,
+      "step": 26501
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943347041676119e-05,
+      "loss": 0.2866,
+      "step": 26502
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943341590354271e-05,
+      "loss": 0.2996,
+      "step": 26503
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433361387778103e-05,
+      "loss": 0.3322,
+      "step": 26504
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433306869467393e-05,
+      "loss": 0.2733,
+      "step": 26505
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433252348610587e-05,
+      "loss": 0.3669,
+      "step": 26506
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.94331978252077e-05,
+      "loss": 0.2881,
+      "step": 26507
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433143299258752e-05,
+      "loss": 0.3045,
+      "step": 26508
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9433088770763752e-05,
+      "loss": 0.2942,
+      "step": 26509
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943303423972272e-05,
+      "loss": 0.3286,
+      "step": 26510
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9432979706135665e-05,
+      "loss": 0.3356,
+      "step": 26511
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943292517000261e-05,
+      "loss": 0.2624,
+      "step": 26512
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943287063132356e-05,
+      "loss": 0.3303,
+      "step": 26513
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9432816090098534e-05,
+      "loss": 0.3118,
+      "step": 26514
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9432761546327548e-05,
+      "loss": 0.2805,
+      "step": 26515
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9432707000010612e-05,
+      "loss": 0.2725,
+      "step": 26516
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9432652451147752e-05,
+      "loss": 0.3211,
+      "step": 26517
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943259789973897e-05,
+      "loss": 0.2835,
+      "step": 26518
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9432543345784285e-05,
+      "loss": 0.288,
+      "step": 26519
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9432488789283713e-05,
+      "loss": 0.2886,
+      "step": 26520
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943243423023727e-05,
+      "loss": 0.2806,
+      "step": 26521
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9432379668644966e-05,
+      "loss": 0.3249,
+      "step": 26522
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9432325104506822e-05,
+      "loss": 0.2762,
+      "step": 26523
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943227053782285e-05,
+      "loss": 0.337,
+      "step": 26524
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943221596859306e-05,
+      "loss": 0.3157,
+      "step": 26525
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9432161396817473e-05,
+      "loss": 0.3047,
+      "step": 26526
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.94321068224961e-05,
+      "loss": 0.3346,
+      "step": 26527
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943205224562896e-05,
+      "loss": 0.3169,
+      "step": 26528
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431997666216063e-05,
+      "loss": 0.3073,
+      "step": 26529
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431943084257424e-05,
+      "loss": 0.2944,
+      "step": 26530
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431888499753064e-05,
+      "loss": 0.2944,
+      "step": 26531
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943183391270299e-05,
+      "loss": 0.2942,
+      "step": 26532
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943177932310722e-05,
+      "loss": 0.3397,
+      "step": 26533
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431724730965772e-05,
+      "loss": 0.3497,
+      "step": 26534
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431670136278657e-05,
+      "loss": 0.3158,
+      "step": 26535
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431615539045886e-05,
+      "loss": 0.3424,
+      "step": 26536
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943156093926748e-05,
+      "loss": 0.2753,
+      "step": 26537
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943150633694345e-05,
+      "loss": 0.3655,
+      "step": 26538
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431451732073813e-05,
+      "loss": 0.2542,
+      "step": 26539
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431397124658584e-05,
+      "loss": 0.2982,
+      "step": 26540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431342514697778e-05,
+      "loss": 0.3187,
+      "step": 26541
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431287902191407e-05,
+      "loss": 0.2981,
+      "step": 26542
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943123328713949e-05,
+      "loss": 0.3091,
+      "step": 26543
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431178669542036e-05,
+      "loss": 0.3164,
+      "step": 26544
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431124049399063e-05,
+      "loss": 0.3583,
+      "step": 26545
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9431069426710584e-05,
+      "loss": 0.323,
+      "step": 26546
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943101480147662e-05,
+      "loss": 0.2951,
+      "step": 26547
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943096017369718e-05,
+      "loss": 0.2996,
+      "step": 26548
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430905543372277e-05,
+      "loss": 0.3453,
+      "step": 26549
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943085091050193e-05,
+      "loss": 0.314,
+      "step": 26550
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943079627508615e-05,
+      "loss": 0.3375,
+      "step": 26551
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430741637124958e-05,
+      "loss": 0.3036,
+      "step": 26552
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943068699661836e-05,
+      "loss": 0.326,
+      "step": 26553
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943063235356638e-05,
+      "loss": 0.2938,
+      "step": 26554
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430577707969027e-05,
+      "loss": 0.269,
+      "step": 26555
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430523059826316e-05,
+      "loss": 0.3899,
+      "step": 26556
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430468409138264e-05,
+      "loss": 0.3324,
+      "step": 26557
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430413755904882e-05,
+      "loss": 0.2856,
+      "step": 26558
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430359100126187e-05,
+      "loss": 0.3503,
+      "step": 26559
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430304441802197e-05,
+      "loss": 0.2904,
+      "step": 26560
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430249780932925e-05,
+      "loss": 0.3147,
+      "step": 26561
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.943019511751838e-05,
+      "loss": 0.2788,
+      "step": 26562
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430140451558585e-05,
+      "loss": 0.2914,
+      "step": 26563
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9430085783053548e-05,
+      "loss": 0.2837,
+      "step": 26564
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.943003111200329e-05,
+      "loss": 0.3138,
+      "step": 26565
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942997643840782e-05,
+      "loss": 0.3221,
+      "step": 26566
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9429921762267158e-05,
+      "loss": 0.3104,
+      "step": 26567
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9429867083581313e-05,
+      "loss": 0.3666,
+      "step": 26568
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9429812402350304e-05,
+      "loss": 0.3031,
+      "step": 26569
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9429757718574144e-05,
+      "loss": 0.3179,
+      "step": 26570
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942970303225285e-05,
+      "loss": 0.2857,
+      "step": 26571
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942964834338643e-05,
+      "loss": 0.3287,
+      "step": 26572
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942959365197491e-05,
+      "loss": 0.3534,
+      "step": 26573
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9429538958018294e-05,
+      "loss": 0.2749,
+      "step": 26574
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942948426151661e-05,
+      "loss": 0.3149,
+      "step": 26575
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9429429562469854e-05,
+      "loss": 0.3676,
+      "step": 26576
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9429374860878053e-05,
+      "loss": 0.3131,
+      "step": 26577
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942932015674122e-05,
+      "loss": 0.3473,
+      "step": 26578
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9429265450059374e-05,
+      "loss": 0.3331,
+      "step": 26579
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942921074083252e-05,
+      "loss": 0.3563,
+      "step": 26580
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942915602906068e-05,
+      "loss": 0.282,
+      "step": 26581
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9429101314743866e-05,
+      "loss": 0.3333,
+      "step": 26582
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9429046597882094e-05,
+      "loss": 0.3123,
+      "step": 26583
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428991878475378e-05,
+      "loss": 0.3113,
+      "step": 26584
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428937156523734e-05,
+      "loss": 0.3041,
+      "step": 26585
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428882432027176e-05,
+      "loss": 0.3057,
+      "step": 26586
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942882770498572e-05,
+      "loss": 0.3879,
+      "step": 26587
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428772975399374e-05,
+      "loss": 0.3418,
+      "step": 26588
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428718243268162e-05,
+      "loss": 0.3249,
+      "step": 26589
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942866350859209e-05,
+      "loss": 0.283,
+      "step": 26590
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942860877137118e-05,
+      "loss": 0.3535,
+      "step": 26591
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942855403160545e-05,
+      "loss": 0.2957,
+      "step": 26592
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428499289294904e-05,
+      "loss": 0.3198,
+      "step": 26593
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428444544439564e-05,
+      "loss": 0.3771,
+      "step": 26594
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942838979703944e-05,
+      "loss": 0.3356,
+      "step": 26595
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942833504709455e-05,
+      "loss": 0.2969,
+      "step": 26596
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428280294604914e-05,
+      "loss": 0.2897,
+      "step": 26597
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428225539570534e-05,
+      "loss": 0.2763,
+      "step": 26598
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428170781991435e-05,
+      "loss": 0.2708,
+      "step": 26599
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428116021867628e-05,
+      "loss": 0.2871,
+      "step": 26600
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942806125919913e-05,
+      "loss": 0.3506,
+      "step": 26601
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9428006493985953e-05,
+      "loss": 0.3202,
+      "step": 26602
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427951726228113e-05,
+      "loss": 0.295,
+      "step": 26603
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427896955925622e-05,
+      "loss": 0.3114,
+      "step": 26604
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427842183078503e-05,
+      "loss": 0.2908,
+      "step": 26605
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942778740768676e-05,
+      "loss": 0.3292,
+      "step": 26606
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427732629750417e-05,
+      "loss": 0.2413,
+      "step": 26607
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427677849269484e-05,
+      "loss": 0.2942,
+      "step": 26608
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427623066243977e-05,
+      "loss": 0.3278,
+      "step": 26609
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427568280673907e-05,
+      "loss": 0.3054,
+      "step": 26610
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427513492559297e-05,
+      "loss": 0.3873,
+      "step": 26611
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427458701900152e-05,
+      "loss": 0.276,
+      "step": 26612
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427403908696496e-05,
+      "loss": 0.3014,
+      "step": 26613
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427349112948337e-05,
+      "loss": 0.2872,
+      "step": 26614
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427294314655697e-05,
+      "loss": 0.325,
+      "step": 26615
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942723951381858e-05,
+      "loss": 0.2937,
+      "step": 26616
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942718471043701e-05,
+      "loss": 0.3285,
+      "step": 26617
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427129904511e-05,
+      "loss": 0.3161,
+      "step": 26618
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942707509604056e-05,
+      "loss": 0.3856,
+      "step": 26619
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9427020285025713e-05,
+      "loss": 0.2859,
+      "step": 26620
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426965471466466e-05,
+      "loss": 0.3108,
+      "step": 26621
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426910655362838e-05,
+      "loss": 0.3729,
+      "step": 26622
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942685583671484e-05,
+      "loss": 0.3182,
+      "step": 26623
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426801015522493e-05,
+      "loss": 0.3096,
+      "step": 26624
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426746191785807e-05,
+      "loss": 0.3202,
+      "step": 26625
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.94266913655048e-05,
+      "loss": 0.312,
+      "step": 26626
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426636536679486e-05,
+      "loss": 0.3058,
+      "step": 26627
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426581705309874e-05,
+      "loss": 0.2977,
+      "step": 26628
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426526871395985e-05,
+      "loss": 0.355,
+      "step": 26629
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426472034937833e-05,
+      "loss": 0.3414,
+      "step": 26630
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942641719593543e-05,
+      "loss": 0.2854,
+      "step": 26631
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.94263623543888e-05,
+      "loss": 0.2715,
+      "step": 26632
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426307510297943e-05,
+      "loss": 0.3488,
+      "step": 26633
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426252663662883e-05,
+      "loss": 0.2881,
+      "step": 26634
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9426197814483635e-05,
+      "loss": 0.3459,
+      "step": 26635
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942614296276021e-05,
+      "loss": 0.3129,
+      "step": 26636
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942608810849263e-05,
+      "loss": 0.3446,
+      "step": 26637
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.94260332516809e-05,
+      "loss": 0.3181,
+      "step": 26638
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942597839232504e-05,
+      "loss": 0.2859,
+      "step": 26639
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942592353042507e-05,
+      "loss": 0.2978,
+      "step": 26640
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425868665980992e-05,
+      "loss": 0.2928,
+      "step": 26641
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425813798992832e-05,
+      "loss": 0.3292,
+      "step": 26642
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425758929460598e-05,
+      "loss": 0.3225,
+      "step": 26643
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942570405738431e-05,
+      "loss": 0.2819,
+      "step": 26644
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942564918276398e-05,
+      "loss": 0.3423,
+      "step": 26645
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942559430559962e-05,
+      "loss": 0.356,
+      "step": 26646
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425539425891255e-05,
+      "loss": 0.3085,
+      "step": 26647
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425484543638887e-05,
+      "loss": 0.3156,
+      "step": 26648
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942542965884254e-05,
+      "loss": 0.2849,
+      "step": 26649
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942537477150222e-05,
+      "loss": 0.2674,
+      "step": 26650
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425319881617954e-05,
+      "loss": 0.3735,
+      "step": 26651
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425264989189748e-05,
+      "loss": 0.3431,
+      "step": 26652
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425210094217615e-05,
+      "loss": 0.3059,
+      "step": 26653
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425155196701578e-05,
+      "loss": 0.2815,
+      "step": 26654
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425100296641646e-05,
+      "loss": 0.3108,
+      "step": 26655
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9425045394037837e-05,
+      "loss": 0.3086,
+      "step": 26656
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9424990488890162e-05,
+      "loss": 0.3519,
+      "step": 26657
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942493558119864e-05,
+      "loss": 0.3207,
+      "step": 26658
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.942488067096328e-05,
+      "loss": 0.3077,
+      "step": 26659
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9424825758184106e-05,
+      "loss": 0.3255,
+      "step": 26660
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9424770842861122e-05,
+      "loss": 0.337,
+      "step": 26661
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9424715924994353e-05,
+      "loss": 0.3207,
+      "step": 26662
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9424661004583807e-05,
+      "loss": 0.2824,
+      "step": 26663
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.94246060816295e-05,
+      "loss": 0.3159,
+      "step": 26664
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9424551156131448e-05,
+      "loss": 0.3237,
+      "step": 26665
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9424496228089668e-05,
+      "loss": 0.311,
+      "step": 26666
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942444129750417e-05,
+      "loss": 0.26,
+      "step": 26667
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942438636437497e-05,
+      "loss": 0.3218,
+      "step": 26668
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942433142870209e-05,
+      "loss": 0.2812,
+      "step": 26669
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9424276490485535e-05,
+      "loss": 0.299,
+      "step": 26670
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9424221549725325e-05,
+      "loss": 0.2978,
+      "step": 26671
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942416660642147e-05,
+      "loss": 0.3057,
+      "step": 26672
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942411166057399e-05,
+      "loss": 0.3004,
+      "step": 26673
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9424056712182902e-05,
+      "loss": 0.3155,
+      "step": 26674
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9424001761248213e-05,
+      "loss": 0.3061,
+      "step": 26675
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423946807769942e-05,
+      "loss": 0.306,
+      "step": 26676
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423891851748106e-05,
+      "loss": 0.2637,
+      "step": 26677
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423836893182714e-05,
+      "loss": 0.3229,
+      "step": 26678
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423781932073788e-05,
+      "loss": 0.3174,
+      "step": 26679
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423726968421337e-05,
+      "loss": 0.3217,
+      "step": 26680
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942367200222538e-05,
+      "loss": 0.355,
+      "step": 26681
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423617033485928e-05,
+      "loss": 0.2989,
+      "step": 26682
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423562062202997e-05,
+      "loss": 0.3683,
+      "step": 26683
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423507088376606e-05,
+      "loss": 0.3208,
+      "step": 26684
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423452112006765e-05,
+      "loss": 0.2731,
+      "step": 26685
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942339713309349e-05,
+      "loss": 0.3523,
+      "step": 26686
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423342151636796e-05,
+      "loss": 0.3907,
+      "step": 26687
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423287167636696e-05,
+      "loss": 0.3045,
+      "step": 26688
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423232181093206e-05,
+      "loss": 0.3275,
+      "step": 26689
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423177192006348e-05,
+      "loss": 0.3637,
+      "step": 26690
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423122200376127e-05,
+      "loss": 0.324,
+      "step": 26691
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9423067206202558e-05,
+      "loss": 0.3139,
+      "step": 26692
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942301220948566e-05,
+      "loss": 0.3734,
+      "step": 26693
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422957210225452e-05,
+      "loss": 0.3715,
+      "step": 26694
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942290220842194e-05,
+      "loss": 0.2905,
+      "step": 26695
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942284720407514e-05,
+      "loss": 0.2873,
+      "step": 26696
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422792197185073e-05,
+      "loss": 0.2946,
+      "step": 26697
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422737187751752e-05,
+      "loss": 0.3108,
+      "step": 26698
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422682175775187e-05,
+      "loss": 0.3361,
+      "step": 26699
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422627161255398e-05,
+      "loss": 0.3095,
+      "step": 26700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422572144192397e-05,
+      "loss": 0.2719,
+      "step": 26701
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.94225171245862e-05,
+      "loss": 0.3393,
+      "step": 26702
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422462102436822e-05,
+      "loss": 0.3281,
+      "step": 26703
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422407077744277e-05,
+      "loss": 0.3006,
+      "step": 26704
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942235205050858e-05,
+      "loss": 0.3158,
+      "step": 26705
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422297020729743e-05,
+      "loss": 0.3337,
+      "step": 26706
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942224198840779e-05,
+      "loss": 0.3053,
+      "step": 26707
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422186953542726e-05,
+      "loss": 0.3141,
+      "step": 26708
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942213191613457e-05,
+      "loss": 0.3255,
+      "step": 26709
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422076876183338e-05,
+      "loss": 0.3448,
+      "step": 26710
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9422021833689043e-05,
+      "loss": 0.3312,
+      "step": 26711
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.94219667886517e-05,
+      "loss": 0.296,
+      "step": 26712
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421911741071322e-05,
+      "loss": 0.349,
+      "step": 26713
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942185669094793e-05,
+      "loss": 0.2981,
+      "step": 26714
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421801638281532e-05,
+      "loss": 0.324,
+      "step": 26715
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421746583072148e-05,
+      "loss": 0.314,
+      "step": 26716
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421691525319788e-05,
+      "loss": 0.2999,
+      "step": 26717
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421636465024472e-05,
+      "loss": 0.2718,
+      "step": 26718
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942158140218621e-05,
+      "loss": 0.3815,
+      "step": 26719
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421526336805018e-05,
+      "loss": 0.3271,
+      "step": 26720
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421471268880917e-05,
+      "loss": 0.3992,
+      "step": 26721
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942141619841391e-05,
+      "loss": 0.3249,
+      "step": 26722
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421361125404024e-05,
+      "loss": 0.2653,
+      "step": 26723
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421306049851267e-05,
+      "loss": 0.2986,
+      "step": 26724
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421250971755655e-05,
+      "loss": 0.2956,
+      "step": 26725
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421195891117203e-05,
+      "loss": 0.3342,
+      "step": 26726
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942114080793593e-05,
+      "loss": 0.3224,
+      "step": 26727
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9421085722211843e-05,
+      "loss": 0.2894,
+      "step": 26728
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942103063394496e-05,
+      "loss": 0.3066,
+      "step": 26729
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.94209755431353e-05,
+      "loss": 0.3355,
+      "step": 26730
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420920449782874e-05,
+      "loss": 0.3165,
+      "step": 26731
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420865353887695e-05,
+      "loss": 0.3515,
+      "step": 26732
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420810255449785e-05,
+      "loss": 0.2749,
+      "step": 26733
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420755154469152e-05,
+      "loss": 0.2663,
+      "step": 26734
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942070005094581e-05,
+      "loss": 0.3531,
+      "step": 26735
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942064494487978e-05,
+      "loss": 0.3457,
+      "step": 26736
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420589836271077e-05,
+      "loss": 0.3393,
+      "step": 26737
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942053472511971e-05,
+      "loss": 0.3565,
+      "step": 26738
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420479611425696e-05,
+      "loss": 0.3174,
+      "step": 26739
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942042449518905e-05,
+      "loss": 0.3704,
+      "step": 26740
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420369376409786e-05,
+      "loss": 0.3591,
+      "step": 26741
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420314255087925e-05,
+      "loss": 0.3276,
+      "step": 26742
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420259131223475e-05,
+      "loss": 0.2895,
+      "step": 26743
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942020400481645e-05,
+      "loss": 0.3114,
+      "step": 26744
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420148875866872e-05,
+      "loss": 0.3151,
+      "step": 26745
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.942009374437475e-05,
+      "loss": 0.2579,
+      "step": 26746
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.94200386103401e-05,
+      "loss": 0.2909,
+      "step": 26747
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.941998347376294e-05,
+      "loss": 0.3442,
+      "step": 26748
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.941992833464328e-05,
+      "loss": 0.2869,
+      "step": 26749
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.941987319298114e-05,
+      "loss": 0.297,
+      "step": 26750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9419818048776527e-05,
+      "loss": 0.3554,
+      "step": 26751
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9419762902029468e-05,
+      "loss": 0.3071,
+      "step": 26752
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9419707752739968e-05,
+      "loss": 0.3155,
+      "step": 26753
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9419652600908044e-05,
+      "loss": 0.3019,
+      "step": 26754
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9419597446533708e-05,
+      "loss": 0.3009,
+      "step": 26755
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9419542289616984e-05,
+      "loss": 0.3043,
+      "step": 26756
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.941948713015788e-05,
+      "loss": 0.3196,
+      "step": 26757
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9419431968156416e-05,
+      "loss": 0.2762,
+      "step": 26758
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.94193768036126e-05,
+      "loss": 0.279,
+      "step": 26759
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.941932163652645e-05,
+      "loss": 0.2876,
+      "step": 26760
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.941926646689798e-05,
+      "loss": 0.3014,
+      "step": 26761
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9419211294727203e-05,
+      "loss": 0.3266,
+      "step": 26762
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9419156120014145e-05,
+      "loss": 0.3286,
+      "step": 26763
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9419100942758806e-05,
+      "loss": 0.3227,
+      "step": 26764
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9419045762961212e-05,
+      "loss": 0.3139,
+      "step": 26765
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941899058062137e-05,
+      "loss": 0.3079,
+      "step": 26766
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.94189353957393e-05,
+      "loss": 0.3481,
+      "step": 26767
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9418880208315017e-05,
+      "loss": 0.3549,
+      "step": 26768
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941882501834853e-05,
+      "loss": 0.2958,
+      "step": 26769
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941876982583986e-05,
+      "loss": 0.2963,
+      "step": 26770
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9418714630789023e-05,
+      "loss": 0.3242,
+      "step": 26771
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941865943319603e-05,
+      "loss": 0.3109,
+      "step": 26772
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9418604233060894e-05,
+      "loss": 0.3002,
+      "step": 26773
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9418549030383635e-05,
+      "loss": 0.328,
+      "step": 26774
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9418493825164266e-05,
+      "loss": 0.3131,
+      "step": 26775
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9418438617402797e-05,
+      "loss": 0.3363,
+      "step": 26776
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941838340709925e-05,
+      "loss": 0.2846,
+      "step": 26777
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941832819425364e-05,
+      "loss": 0.2416,
+      "step": 26778
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941827297886598e-05,
+      "loss": 0.3711,
+      "step": 26779
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9418217760936276e-05,
+      "loss": 0.2692,
+      "step": 26780
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9418162540464557e-05,
+      "loss": 0.3228,
+      "step": 26781
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941810731745083e-05,
+      "loss": 0.3393,
+      "step": 26782
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9418052091895114e-05,
+      "loss": 0.3413,
+      "step": 26783
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417996863797422e-05,
+      "loss": 0.3834,
+      "step": 26784
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417941633157766e-05,
+      "loss": 0.2764,
+      "step": 26785
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417886399976165e-05,
+      "loss": 0.3065,
+      "step": 26786
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941783116425263e-05,
+      "loss": 0.2959,
+      "step": 26787
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417775925987183e-05,
+      "loss": 0.358,
+      "step": 26788
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417720685179832e-05,
+      "loss": 0.3185,
+      "step": 26789
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417665441830593e-05,
+      "loss": 0.3769,
+      "step": 26790
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417610195939486e-05,
+      "loss": 0.3291,
+      "step": 26791
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417554947506518e-05,
+      "loss": 0.3225,
+      "step": 26792
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941749969653171e-05,
+      "loss": 0.3666,
+      "step": 26793
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417444443015072e-05,
+      "loss": 0.3084,
+      "step": 26794
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417389186956622e-05,
+      "loss": 0.3198,
+      "step": 26795
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417333928356375e-05,
+      "loss": 0.2669,
+      "step": 26796
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941727866721435e-05,
+      "loss": 0.2713,
+      "step": 26797
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417223403530552e-05,
+      "loss": 0.23,
+      "step": 26798
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417168137305003e-05,
+      "loss": 0.3132,
+      "step": 26799
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417112868537718e-05,
+      "loss": 0.2912,
+      "step": 26800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9417057597228708e-05,
+      "loss": 0.2803,
+      "step": 26801
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941700232337799e-05,
+      "loss": 0.2965,
+      "step": 26802
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941694704698558e-05,
+      "loss": 0.2663,
+      "step": 26803
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416891768051495e-05,
+      "loss": 0.3124,
+      "step": 26804
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416836486575744e-05,
+      "loss": 0.3262,
+      "step": 26805
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416781202558342e-05,
+      "loss": 0.3215,
+      "step": 26806
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941672591599931e-05,
+      "loss": 0.3237,
+      "step": 26807
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941667062689866e-05,
+      "loss": 0.2644,
+      "step": 26808
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416615335256407e-05,
+      "loss": 0.3231,
+      "step": 26809
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416560041072564e-05,
+      "loss": 0.3308,
+      "step": 26810
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416504744347147e-05,
+      "loss": 0.3206,
+      "step": 26811
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941644944508017e-05,
+      "loss": 0.2617,
+      "step": 26812
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416394143271655e-05,
+      "loss": 0.3288,
+      "step": 26813
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416338838921604e-05,
+      "loss": 0.3361,
+      "step": 26814
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416283532030043e-05,
+      "loss": 0.2771,
+      "step": 26815
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416228222596984e-05,
+      "loss": 0.3091,
+      "step": 26816
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941617291062244e-05,
+      "loss": 0.3042,
+      "step": 26817
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416117596106425e-05,
+      "loss": 0.2923,
+      "step": 26818
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9416062279048958e-05,
+      "loss": 0.2997,
+      "step": 26819
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941600695945005e-05,
+      "loss": 0.3006,
+      "step": 26820
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415951637309722e-05,
+      "loss": 0.2568,
+      "step": 26821
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415896312627983e-05,
+      "loss": 0.3559,
+      "step": 26822
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415840985404847e-05,
+      "loss": 0.3083,
+      "step": 26823
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415785655640334e-05,
+      "loss": 0.3474,
+      "step": 26824
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415730323334455e-05,
+      "loss": 0.3,
+      "step": 26825
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415674988487226e-05,
+      "loss": 0.4614,
+      "step": 26826
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941561965109866e-05,
+      "loss": 0.3123,
+      "step": 26827
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415564311168777e-05,
+      "loss": 0.3074,
+      "step": 26828
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941550896869759e-05,
+      "loss": 0.296,
+      "step": 26829
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415453623685113e-05,
+      "loss": 0.3185,
+      "step": 26830
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941539827613136e-05,
+      "loss": 0.3617,
+      "step": 26831
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415342926036346e-05,
+      "loss": 0.3877,
+      "step": 26832
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941528757340009e-05,
+      "loss": 0.2915,
+      "step": 26833
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415232218222603e-05,
+      "loss": 0.3909,
+      "step": 26834
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.94151768605039e-05,
+      "loss": 0.2778,
+      "step": 26835
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415121500243995e-05,
+      "loss": 0.3204,
+      "step": 26836
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9415066137442907e-05,
+      "loss": 0.3302,
+      "step": 26837
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941501077210065e-05,
+      "loss": 0.3274,
+      "step": 26838
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414955404217234e-05,
+      "loss": 0.284,
+      "step": 26839
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941490003379268e-05,
+      "loss": 0.3269,
+      "step": 26840
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414844660827e-05,
+      "loss": 0.3307,
+      "step": 26841
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414789285320208e-05,
+      "loss": 0.3073,
+      "step": 26842
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414733907272324e-05,
+      "loss": 0.3472,
+      "step": 26843
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414678526683354e-05,
+      "loss": 0.3749,
+      "step": 26844
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414623143553323e-05,
+      "loss": 0.3533,
+      "step": 26845
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414567757882237e-05,
+      "loss": 0.3093,
+      "step": 26846
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414512369670116e-05,
+      "loss": 0.3592,
+      "step": 26847
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414456978916978e-05,
+      "loss": 0.3491,
+      "step": 26848
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941440158562283e-05,
+      "loss": 0.3124,
+      "step": 26849
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414346189787694e-05,
+      "loss": 0.3601,
+      "step": 26850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941429079141158e-05,
+      "loss": 0.3215,
+      "step": 26851
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414235390494505e-05,
+      "loss": 0.375,
+      "step": 26852
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414179987036486e-05,
+      "loss": 0.3859,
+      "step": 26853
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414124581037534e-05,
+      "loss": 0.2929,
+      "step": 26854
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414069172497666e-05,
+      "loss": 0.2869,
+      "step": 26855
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9414013761416893e-05,
+      "loss": 0.3109,
+      "step": 26856
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9413958347795237e-05,
+      "loss": 0.2682,
+      "step": 26857
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9413902931632714e-05,
+      "loss": 0.3341,
+      "step": 26858
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9413847512929328e-05,
+      "loss": 0.3205,
+      "step": 26859
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9413792091685108e-05,
+      "loss": 0.3425,
+      "step": 26860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413736667900053e-05,
+      "loss": 0.3364,
+      "step": 26861
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413681241574188e-05,
+      "loss": 0.3047,
+      "step": 26862
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941362581270753e-05,
+      "loss": 0.2675,
+      "step": 26863
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413570381300092e-05,
+      "loss": 0.3051,
+      "step": 26864
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413514947351882e-05,
+      "loss": 0.3124,
+      "step": 26865
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413459510862922e-05,
+      "loss": 0.3296,
+      "step": 26866
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413404071833226e-05,
+      "loss": 0.3215,
+      "step": 26867
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413348630262807e-05,
+      "loss": 0.2938,
+      "step": 26868
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413293186151682e-05,
+      "loss": 0.2919,
+      "step": 26869
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413237739499865e-05,
+      "loss": 0.3698,
+      "step": 26870
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413182290307373e-05,
+      "loss": 0.384,
+      "step": 26871
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941312683857422e-05,
+      "loss": 0.2878,
+      "step": 26872
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9413071384300416e-05,
+      "loss": 0.347,
+      "step": 26873
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941301592748598e-05,
+      "loss": 0.2781,
+      "step": 26874
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941296046813093e-05,
+      "loss": 0.2545,
+      "step": 26875
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412905006235276e-05,
+      "loss": 0.3303,
+      "step": 26876
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412849541799037e-05,
+      "loss": 0.3051,
+      "step": 26877
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412794074822227e-05,
+      "loss": 0.3176,
+      "step": 26878
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412738605304854e-05,
+      "loss": 0.3117,
+      "step": 26879
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412683133246944e-05,
+      "loss": 0.2877,
+      "step": 26880
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412627658648505e-05,
+      "loss": 0.3189,
+      "step": 26881
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412572181509553e-05,
+      "loss": 0.2796,
+      "step": 26882
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412516701830105e-05,
+      "loss": 0.2986,
+      "step": 26883
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412461219610176e-05,
+      "loss": 0.351,
+      "step": 26884
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412405734849774e-05,
+      "loss": 0.3613,
+      "step": 26885
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412350247548927e-05,
+      "loss": 0.3479,
+      "step": 26886
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941229475770764e-05,
+      "loss": 0.3178,
+      "step": 26887
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941223926532593e-05,
+      "loss": 0.3253,
+      "step": 26888
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412183770403808e-05,
+      "loss": 0.2836,
+      "step": 26889
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.94121282729413e-05,
+      "loss": 0.3159,
+      "step": 26890
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941207277293841e-05,
+      "loss": 0.2664,
+      "step": 26891
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9412017270395166e-05,
+      "loss": 0.2802,
+      "step": 26892
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411961765311567e-05,
+      "loss": 0.3314,
+      "step": 26893
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941190625768764e-05,
+      "loss": 0.3154,
+      "step": 26894
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941185074752339e-05,
+      "loss": 0.3324,
+      "step": 26895
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411795234818843e-05,
+      "loss": 0.2859,
+      "step": 26896
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411739719574007e-05,
+      "loss": 0.2946,
+      "step": 26897
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411684201788895e-05,
+      "loss": 0.3214,
+      "step": 26898
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411628681463528e-05,
+      "loss": 0.2899,
+      "step": 26899
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941157315859792e-05,
+      "loss": 0.2923,
+      "step": 26900
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411517633192082e-05,
+      "loss": 0.3151,
+      "step": 26901
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941146210524603e-05,
+      "loss": 0.3202,
+      "step": 26902
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411406574759785e-05,
+      "loss": 0.3138,
+      "step": 26903
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411351041733355e-05,
+      "loss": 0.3032,
+      "step": 26904
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411295506166755e-05,
+      "loss": 0.3007,
+      "step": 26905
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411239968060008e-05,
+      "loss": 0.3242,
+      "step": 26906
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411184427413118e-05,
+      "loss": 0.2611,
+      "step": 26907
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411128884226108e-05,
+      "loss": 0.3525,
+      "step": 26908
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9411073338498988e-05,
+      "loss": 0.3033,
+      "step": 26909
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941101779023178e-05,
+      "loss": 0.305,
+      "step": 26910
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410962239424488e-05,
+      "loss": 0.2824,
+      "step": 26911
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410906686077138e-05,
+      "loss": 0.2723,
+      "step": 26912
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941085113018974e-05,
+      "loss": 0.2991,
+      "step": 26913
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410795571762307e-05,
+      "loss": 0.322,
+      "step": 26914
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941074001079486e-05,
+      "loss": 0.301,
+      "step": 26915
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410684447287403e-05,
+      "loss": 0.331,
+      "step": 26916
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410628881239963e-05,
+      "loss": 0.3107,
+      "step": 26917
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410573312652552e-05,
+      "loss": 0.3603,
+      "step": 26918
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941051774152518e-05,
+      "loss": 0.3013,
+      "step": 26919
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941046216785787e-05,
+      "loss": 0.2903,
+      "step": 26920
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941040659165063e-05,
+      "loss": 0.3193,
+      "step": 26921
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410351012903472e-05,
+      "loss": 0.2616,
+      "step": 26922
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410295431616422e-05,
+      "loss": 0.3171,
+      "step": 26923
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410239847789486e-05,
+      "loss": 0.2993,
+      "step": 26924
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410184261422688e-05,
+      "loss": 0.3065,
+      "step": 26925
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941012867251603e-05,
+      "loss": 0.3658,
+      "step": 26926
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.941007308106954e-05,
+      "loss": 0.2892,
+      "step": 26927
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9410017487083225e-05,
+      "loss": 0.2755,
+      "step": 26928
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.94099618905571e-05,
+      "loss": 0.312,
+      "step": 26929
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9409906291491185e-05,
+      "loss": 0.4031,
+      "step": 26930
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.940985068988549e-05,
+      "loss": 0.3178,
+      "step": 26931
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9409795085740037e-05,
+      "loss": 0.3744,
+      "step": 26932
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9409739479054833e-05,
+      "loss": 0.2878,
+      "step": 26933
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9409683869829895e-05,
+      "loss": 0.302,
+      "step": 26934
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9409628258065245e-05,
+      "loss": 0.3391,
+      "step": 26935
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9409572643760888e-05,
+      "loss": 0.3203,
+      "step": 26936
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.940951702691684e-05,
+      "loss": 0.306,
+      "step": 26937
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9409461407533126e-05,
+      "loss": 0.3133,
+      "step": 26938
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.940940578560975e-05,
+      "loss": 0.3246,
+      "step": 26939
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9409350161146735e-05,
+      "loss": 0.3712,
+      "step": 26940
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.940929453414409e-05,
+      "loss": 0.3093,
+      "step": 26941
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9409238904601834e-05,
+      "loss": 0.296,
+      "step": 26942
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.940918327251998e-05,
+      "loss": 0.2928,
+      "step": 26943
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.940912763789854e-05,
+      "loss": 0.3063,
+      "step": 26944
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9409072000737536e-05,
+      "loss": 0.3065,
+      "step": 26945
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.940901636103698e-05,
+      "loss": 0.2734,
+      "step": 26946
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9408960718796886e-05,
+      "loss": 0.3492,
+      "step": 26947
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.940890507401727e-05,
+      "loss": 0.3174,
+      "step": 26948
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9408849426698143e-05,
+      "loss": 0.2711,
+      "step": 26949
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9408793776839524e-05,
+      "loss": 0.3354,
+      "step": 26950
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9408738124441432e-05,
+      "loss": 0.2937,
+      "step": 26951
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9408682469503875e-05,
+      "loss": 0.3103,
+      "step": 26952
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9408626812026868e-05,
+      "loss": 0.2778,
+      "step": 26953
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9408571152010433e-05,
+      "loss": 0.3357,
+      "step": 26954
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.940851548945458e-05,
+      "loss": 0.2847,
+      "step": 26955
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9408459824359322e-05,
+      "loss": 0.2673,
+      "step": 26956
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.940840415672468e-05,
+      "loss": 0.2893,
+      "step": 26957
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9408348486550663e-05,
+      "loss": 0.3789,
+      "step": 26958
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940829281383729e-05,
+      "loss": 0.2942,
+      "step": 26959
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9408237138584574e-05,
+      "loss": 0.3048,
+      "step": 26960
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9408181460792533e-05,
+      "loss": 0.2528,
+      "step": 26961
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9408125780461175e-05,
+      "loss": 0.3209,
+      "step": 26962
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9408070097590524e-05,
+      "loss": 0.3224,
+      "step": 26963
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940801441218059e-05,
+      "loss": 0.3401,
+      "step": 26964
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940795872423139e-05,
+      "loss": 0.3595,
+      "step": 26965
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407903033742936e-05,
+      "loss": 0.3198,
+      "step": 26966
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407847340715248e-05,
+      "loss": 0.2918,
+      "step": 26967
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407791645148336e-05,
+      "loss": 0.322,
+      "step": 26968
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407735947042215e-05,
+      "loss": 0.3081,
+      "step": 26969
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407680246396906e-05,
+      "loss": 0.3376,
+      "step": 26970
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407624543212416e-05,
+      "loss": 0.2834,
+      "step": 26971
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940756883748877e-05,
+      "loss": 0.2998,
+      "step": 26972
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940751312922597e-05,
+      "loss": 0.3044,
+      "step": 26973
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407457418424043e-05,
+      "loss": 0.3112,
+      "step": 26974
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407401705083e-05,
+      "loss": 0.3163,
+      "step": 26975
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407345989202853e-05,
+      "loss": 0.3773,
+      "step": 26976
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407290270783618e-05,
+      "loss": 0.2964,
+      "step": 26977
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9407234549825314e-05,
+      "loss": 0.3406,
+      "step": 26978
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940717882632795e-05,
+      "loss": 0.3214,
+      "step": 26979
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940712310029155e-05,
+      "loss": 0.2982,
+      "step": 26980
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940706737171612e-05,
+      "loss": 0.3199,
+      "step": 26981
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940701164060168e-05,
+      "loss": 0.325,
+      "step": 26982
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9406955906948242e-05,
+      "loss": 0.3136,
+      "step": 26983
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9406900170755824e-05,
+      "loss": 0.3096,
+      "step": 26984
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9406844432024437e-05,
+      "loss": 0.3251,
+      "step": 26985
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.94067886907541e-05,
+      "loss": 0.3129,
+      "step": 26986
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940673294694483e-05,
+      "loss": 0.3131,
+      "step": 26987
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9406677200596635e-05,
+      "loss": 0.3463,
+      "step": 26988
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9406621451709534e-05,
+      "loss": 0.279,
+      "step": 26989
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9406565700283546e-05,
+      "loss": 0.2832,
+      "step": 26990
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940650994631868e-05,
+      "loss": 0.3224,
+      "step": 26991
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940645418981495e-05,
+      "loss": 0.2886,
+      "step": 26992
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9406398430772374e-05,
+      "loss": 0.331,
+      "step": 26993
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940634266919097e-05,
+      "loss": 0.2835,
+      "step": 26994
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940628690507075e-05,
+      "loss": 0.3103,
+      "step": 26995
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9406231138411728e-05,
+      "loss": 0.2963,
+      "step": 26996
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940617536921392e-05,
+      "loss": 0.3116,
+      "step": 26997
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9406119597477345e-05,
+      "loss": 0.2783,
+      "step": 26998
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940606382320201e-05,
+      "loss": 0.3332,
+      "step": 26999
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9406008046387934e-05,
+      "loss": 0.2867,
+      "step": 27000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405952267035135e-05,
+      "loss": 0.331,
+      "step": 27001
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405896485143625e-05,
+      "loss": 0.241,
+      "step": 27002
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940584070071342e-05,
+      "loss": 0.3573,
+      "step": 27003
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405784913744533e-05,
+      "loss": 0.3021,
+      "step": 27004
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940572912423698e-05,
+      "loss": 0.3785,
+      "step": 27005
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940567333219078e-05,
+      "loss": 0.3139,
+      "step": 27006
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405617537605943e-05,
+      "loss": 0.3105,
+      "step": 27007
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405561740482487e-05,
+      "loss": 0.3163,
+      "step": 27008
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405505940820422e-05,
+      "loss": 0.3314,
+      "step": 27009
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405450138619773e-05,
+      "loss": 0.275,
+      "step": 27010
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405394333880546e-05,
+      "loss": 0.2953,
+      "step": 27011
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405338526602758e-05,
+      "loss": 0.307,
+      "step": 27012
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405282716786425e-05,
+      "loss": 0.2574,
+      "step": 27013
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9405226904431565e-05,
+      "loss": 0.308,
+      "step": 27014
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940517108953819e-05,
+      "loss": 0.2893,
+      "step": 27015
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940511527210631e-05,
+      "loss": 0.2905,
+      "step": 27016
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940505945213595e-05,
+      "loss": 0.3012,
+      "step": 27017
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940500362962712e-05,
+      "loss": 0.3334,
+      "step": 27018
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404947804579835e-05,
+      "loss": 0.3314,
+      "step": 27019
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404891976994113e-05,
+      "loss": 0.3176,
+      "step": 27020
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404836146869966e-05,
+      "loss": 0.3128,
+      "step": 27021
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404780314207407e-05,
+      "loss": 0.3163,
+      "step": 27022
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404724479006456e-05,
+      "loss": 0.3897,
+      "step": 27023
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404668641267126e-05,
+      "loss": 0.3415,
+      "step": 27024
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404612800989432e-05,
+      "loss": 0.3526,
+      "step": 27025
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404556958173387e-05,
+      "loss": 0.3271,
+      "step": 27026
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404501112819008e-05,
+      "loss": 0.276,
+      "step": 27027
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404445264926312e-05,
+      "loss": 0.347,
+      "step": 27028
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404389414495312e-05,
+      "loss": 0.2915,
+      "step": 27029
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404333561526022e-05,
+      "loss": 0.3538,
+      "step": 27030
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404277706018462e-05,
+      "loss": 0.2471,
+      "step": 27031
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404221847972643e-05,
+      "loss": 0.343,
+      "step": 27032
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404165987388577e-05,
+      "loss": 0.2737,
+      "step": 27033
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9404110124266283e-05,
+      "loss": 0.2668,
+      "step": 27034
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940405425860578e-05,
+      "loss": 0.3312,
+      "step": 27035
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403998390407074e-05,
+      "loss": 0.2669,
+      "step": 27036
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403942519670187e-05,
+      "loss": 0.3206,
+      "step": 27037
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403886646395132e-05,
+      "loss": 0.2999,
+      "step": 27038
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403830770581923e-05,
+      "loss": 0.3205,
+      "step": 27039
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940377489223058e-05,
+      "loss": 0.3153,
+      "step": 27040
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403719011341108e-05,
+      "loss": 0.3407,
+      "step": 27041
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403663127913533e-05,
+      "loss": 0.3084,
+      "step": 27042
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940360724194786e-05,
+      "loss": 0.3633,
+      "step": 27043
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403551353444116e-05,
+      "loss": 0.2669,
+      "step": 27044
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403495462402307e-05,
+      "loss": 0.313,
+      "step": 27045
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940343956882245e-05,
+      "loss": 0.289,
+      "step": 27046
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403383672704563e-05,
+      "loss": 0.3237,
+      "step": 27047
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403327774048653e-05,
+      "loss": 0.3034,
+      "step": 27048
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940327187285475e-05,
+      "loss": 0.3129,
+      "step": 27049
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403215969122852e-05,
+      "loss": 0.3025,
+      "step": 27050
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403160062852984e-05,
+      "loss": 0.3225,
+      "step": 27051
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.940310415404516e-05,
+      "loss": 0.2726,
+      "step": 27052
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9403048242699393e-05,
+      "loss": 0.2718,
+      "step": 27053
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9402992328815702e-05,
+      "loss": 0.3488,
+      "step": 27054
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9402936412394096e-05,
+      "loss": 0.2997,
+      "step": 27055
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9402880493434594e-05,
+      "loss": 0.3334,
+      "step": 27056
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9402824571937214e-05,
+      "loss": 0.2772,
+      "step": 27057
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9402768647901965e-05,
+      "loss": 0.3397,
+      "step": 27058
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9402712721328866e-05,
+      "loss": 0.304,
+      "step": 27059
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940265679221793e-05,
+      "loss": 0.3377,
+      "step": 27060
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9402600860569174e-05,
+      "loss": 0.2599,
+      "step": 27061
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940254492638261e-05,
+      "loss": 0.3743,
+      "step": 27062
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940248898965826e-05,
+      "loss": 0.3673,
+      "step": 27063
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940243305039613e-05,
+      "loss": 0.3809,
+      "step": 27064
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940237710859624e-05,
+      "loss": 0.2825,
+      "step": 27065
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.94023211642586e-05,
+      "loss": 0.3414,
+      "step": 27066
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9402265217383237e-05,
+      "loss": 0.3221,
+      "step": 27067
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9402209267970155e-05,
+      "loss": 0.2904,
+      "step": 27068
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9402153316019373e-05,
+      "loss": 0.3238,
+      "step": 27069
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9402097361530908e-05,
+      "loss": 0.3383,
+      "step": 27070
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9402041404504772e-05,
+      "loss": 0.383,
+      "step": 27071
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940198544494098e-05,
+      "loss": 0.2982,
+      "step": 27072
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9401929482839548e-05,
+      "loss": 0.2862,
+      "step": 27073
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9401873518200494e-05,
+      "loss": 0.3025,
+      "step": 27074
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9401817551023827e-05,
+      "loss": 0.3275,
+      "step": 27075
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9401761581309565e-05,
+      "loss": 0.3377,
+      "step": 27076
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9401705609057727e-05,
+      "loss": 0.3106,
+      "step": 27077
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9401649634268325e-05,
+      "loss": 0.286,
+      "step": 27078
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940159365694137e-05,
+      "loss": 0.3078,
+      "step": 27079
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9401537677076882e-05,
+      "loss": 0.3048,
+      "step": 27080
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940148169467488e-05,
+      "loss": 0.34,
+      "step": 27081
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940142570973537e-05,
+      "loss": 0.3005,
+      "step": 27082
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940136972225837e-05,
+      "loss": 0.3455,
+      "step": 27083
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.94013137322439e-05,
+      "loss": 0.3247,
+      "step": 27084
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9401257739691964e-05,
+      "loss": 0.3297,
+      "step": 27085
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940120174460259e-05,
+      "loss": 0.3615,
+      "step": 27086
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940114574697579e-05,
+      "loss": 0.3333,
+      "step": 27087
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9401089746811575e-05,
+      "loss": 0.2855,
+      "step": 27088
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940103374410996e-05,
+      "loss": 0.2776,
+      "step": 27089
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400977738870966e-05,
+      "loss": 0.3023,
+      "step": 27090
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400921731094598e-05,
+      "loss": 0.2632,
+      "step": 27091
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940086572078088e-05,
+      "loss": 0.3237,
+      "step": 27092
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400809707929826e-05,
+      "loss": 0.3137,
+      "step": 27093
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940075369254145e-05,
+      "loss": 0.3429,
+      "step": 27094
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400697674615766e-05,
+      "loss": 0.3278,
+      "step": 27095
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940064165415279e-05,
+      "loss": 0.2626,
+      "step": 27096
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400585631152534e-05,
+      "loss": 0.3021,
+      "step": 27097
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400529605615016e-05,
+      "loss": 0.3751,
+      "step": 27098
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400473577540255e-05,
+      "loss": 0.2691,
+      "step": 27099
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940041754692826e-05,
+      "loss": 0.3363,
+      "step": 27100
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940036151377905e-05,
+      "loss": 0.3101,
+      "step": 27101
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400305478092634e-05,
+      "loss": 0.3017,
+      "step": 27102
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400249439869036e-05,
+      "loss": 0.3326,
+      "step": 27103
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400193399108266e-05,
+      "loss": 0.3665,
+      "step": 27104
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940013735581034e-05,
+      "loss": 0.2746,
+      "step": 27105
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.940008130997527e-05,
+      "loss": 0.3386,
+      "step": 27106
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9400025261603075e-05,
+      "loss": 0.2791,
+      "step": 27107
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399969210693773e-05,
+      "loss": 0.3005,
+      "step": 27108
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399913157247373e-05,
+      "loss": 0.3451,
+      "step": 27109
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399857101263896e-05,
+      "loss": 0.3464,
+      "step": 27110
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939980104274335e-05,
+      "loss": 0.3445,
+      "step": 27111
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939974498168575e-05,
+      "loss": 0.2761,
+      "step": 27112
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939968891809112e-05,
+      "loss": 0.3548,
+      "step": 27113
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399632851959468e-05,
+      "loss": 0.3304,
+      "step": 27114
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399576783290812e-05,
+      "loss": 0.3721,
+      "step": 27115
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939952071208517e-05,
+      "loss": 0.3554,
+      "step": 27116
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399464638342546e-05,
+      "loss": 0.4021,
+      "step": 27117
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399408562062967e-05,
+      "loss": 0.3235,
+      "step": 27118
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399352483246443e-05,
+      "loss": 0.3012,
+      "step": 27119
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939929640189299e-05,
+      "loss": 0.3225,
+      "step": 27120
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399240318002625e-05,
+      "loss": 0.3611,
+      "step": 27121
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399184231575357e-05,
+      "loss": 0.3096,
+      "step": 27122
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399128142611206e-05,
+      "loss": 0.2861,
+      "step": 27123
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399072051110187e-05,
+      "loss": 0.3314,
+      "step": 27124
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9399015957072315e-05,
+      "loss": 0.408,
+      "step": 27125
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398959860497606e-05,
+      "loss": 0.3003,
+      "step": 27126
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939890376138607e-05,
+      "loss": 0.3406,
+      "step": 27127
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398847659737733e-05,
+      "loss": 0.2791,
+      "step": 27128
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398791555552596e-05,
+      "loss": 0.2401,
+      "step": 27129
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398735448830684e-05,
+      "loss": 0.3137,
+      "step": 27130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939867933957201e-05,
+      "loss": 0.3077,
+      "step": 27131
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398623227776586e-05,
+      "loss": 0.3343,
+      "step": 27132
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398567113444434e-05,
+      "loss": 0.3329,
+      "step": 27133
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398510996575562e-05,
+      "loss": 0.3209,
+      "step": 27134
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398454877169985e-05,
+      "loss": 0.2829,
+      "step": 27135
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398398755227727e-05,
+      "loss": 0.3282,
+      "step": 27136
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939834263074879e-05,
+      "loss": 0.3393,
+      "step": 27137
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398286503733203e-05,
+      "loss": 0.2803,
+      "step": 27138
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398230374180974e-05,
+      "loss": 0.3355,
+      "step": 27139
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398174242092115e-05,
+      "loss": 0.264,
+      "step": 27140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398118107466647e-05,
+      "loss": 0.2799,
+      "step": 27141
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398061970304583e-05,
+      "loss": 0.3121,
+      "step": 27142
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9398005830605936e-05,
+      "loss": 0.3017,
+      "step": 27143
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9397949688370727e-05,
+      "loss": 0.3005,
+      "step": 27144
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9397893543598966e-05,
+      "loss": 0.3476,
+      "step": 27145
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939783739629067e-05,
+      "loss": 0.2928,
+      "step": 27146
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939778124644585e-05,
+      "loss": 0.3827,
+      "step": 27147
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939772509406453e-05,
+      "loss": 0.3393,
+      "step": 27148
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9397668939146718e-05,
+      "loss": 0.3408,
+      "step": 27149
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939761278169243e-05,
+      "loss": 0.2702,
+      "step": 27150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9397556621701687e-05,
+      "loss": 0.3355,
+      "step": 27151
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9397500459174492e-05,
+      "loss": 0.3383,
+      "step": 27152
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9397444294110875e-05,
+      "loss": 0.3327,
+      "step": 27153
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9397388126510837e-05,
+      "loss": 0.3831,
+      "step": 27154
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9397331956374407e-05,
+      "loss": 0.2864,
+      "step": 27155
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.939727578370159e-05,
+      "loss": 0.3409,
+      "step": 27156
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9397219608492403e-05,
+      "loss": 0.3351,
+      "step": 27157
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9397163430746862e-05,
+      "loss": 0.3095,
+      "step": 27158
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9397107250464986e-05,
+      "loss": 0.2993,
+      "step": 27159
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9397051067646786e-05,
+      "loss": 0.3045,
+      "step": 27160
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396994882292274e-05,
+      "loss": 0.3634,
+      "step": 27161
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396938694401474e-05,
+      "loss": 0.3808,
+      "step": 27162
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396882503974397e-05,
+      "loss": 0.3206,
+      "step": 27163
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396826311011052e-05,
+      "loss": 0.2595,
+      "step": 27164
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396770115511464e-05,
+      "loss": 0.3574,
+      "step": 27165
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396713917475645e-05,
+      "loss": 0.3268,
+      "step": 27166
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396657716903603e-05,
+      "loss": 0.3507,
+      "step": 27167
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396601513795365e-05,
+      "loss": 0.3795,
+      "step": 27168
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396545308150938e-05,
+      "loss": 0.346,
+      "step": 27169
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396489099970338e-05,
+      "loss": 0.3008,
+      "step": 27170
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396432889253586e-05,
+      "loss": 0.3393,
+      "step": 27171
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939637667600069e-05,
+      "loss": 0.3261,
+      "step": 27172
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396320460211666e-05,
+      "loss": 0.2712,
+      "step": 27173
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396264241886532e-05,
+      "loss": 0.3204,
+      "step": 27174
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396208021025304e-05,
+      "loss": 0.2921,
+      "step": 27175
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396151797627997e-05,
+      "loss": 0.3364,
+      "step": 27176
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9396095571694624e-05,
+      "loss": 0.3507,
+      "step": 27177
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.93960393432252e-05,
+      "loss": 0.275,
+      "step": 27178
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9395983112219738e-05,
+      "loss": 0.3717,
+      "step": 27179
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939592687867826e-05,
+      "loss": 0.2431,
+      "step": 27180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939587064260078e-05,
+      "loss": 0.3057,
+      "step": 27181
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9395814403987306e-05,
+      "loss": 0.3032,
+      "step": 27182
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9395758162837856e-05,
+      "loss": 0.2922,
+      "step": 27183
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939570191915245e-05,
+      "loss": 0.3171,
+      "step": 27184
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.93956456729311e-05,
+      "loss": 0.4155,
+      "step": 27185
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9395589424173823e-05,
+      "loss": 0.3501,
+      "step": 27186
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939553317288063e-05,
+      "loss": 0.301,
+      "step": 27187
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939547691905154e-05,
+      "loss": 0.2816,
+      "step": 27188
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9395420662686566e-05,
+      "loss": 0.3467,
+      "step": 27189
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9395364403785724e-05,
+      "loss": 0.3119,
+      "step": 27190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939530814234903e-05,
+      "loss": 0.4042,
+      "step": 27191
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9395251878376497e-05,
+      "loss": 0.3294,
+      "step": 27192
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9395195611868144e-05,
+      "loss": 0.3122,
+      "step": 27193
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939513934282398e-05,
+      "loss": 0.3462,
+      "step": 27194
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9395083071244026e-05,
+      "loss": 0.2802,
+      "step": 27195
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9395026797128298e-05,
+      "loss": 0.3128,
+      "step": 27196
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394970520476804e-05,
+      "loss": 0.3305,
+      "step": 27197
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394914241289567e-05,
+      "loss": 0.3078,
+      "step": 27198
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.93948579595666e-05,
+      "loss": 0.3109,
+      "step": 27199
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939480167530791e-05,
+      "loss": 0.322,
+      "step": 27200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394745388513524e-05,
+      "loss": 0.3932,
+      "step": 27201
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939468909918345e-05,
+      "loss": 0.2732,
+      "step": 27202
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394632807317707e-05,
+      "loss": 0.2666,
+      "step": 27203
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939457651291631e-05,
+      "loss": 0.3506,
+      "step": 27204
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939452021597927e-05,
+      "loss": 0.3012,
+      "step": 27205
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394463916506607e-05,
+      "loss": 0.3088,
+      "step": 27206
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394407614498335e-05,
+      "loss": 0.2908,
+      "step": 27207
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394351309954466e-05,
+      "loss": 0.331,
+      "step": 27208
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394295002875017e-05,
+      "loss": 0.3194,
+      "step": 27209
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394238693260006e-05,
+      "loss": 0.3107,
+      "step": 27210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394182381109446e-05,
+      "loss": 0.2885,
+      "step": 27211
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394126066423357e-05,
+      "loss": 0.3786,
+      "step": 27212
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394069749201743e-05,
+      "loss": 0.2706,
+      "step": 27213
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9394013429444627e-05,
+      "loss": 0.2677,
+      "step": 27214
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393957107152023e-05,
+      "loss": 0.2961,
+      "step": 27215
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393900782323947e-05,
+      "loss": 0.2845,
+      "step": 27216
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939384445496041e-05,
+      "loss": 0.3474,
+      "step": 27217
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393788125061434e-05,
+      "loss": 0.2679,
+      "step": 27218
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939373179262703e-05,
+      "loss": 0.324,
+      "step": 27219
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393675457657216e-05,
+      "loss": 0.2788,
+      "step": 27220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393619120152002e-05,
+      "loss": 0.2994,
+      "step": 27221
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393562780111405e-05,
+      "loss": 0.3318,
+      "step": 27222
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393506437535445e-05,
+      "loss": 0.2669,
+      "step": 27223
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393450092424132e-05,
+      "loss": 0.2831,
+      "step": 27224
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393393744777484e-05,
+      "loss": 0.37,
+      "step": 27225
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393337394595514e-05,
+      "loss": 0.3099,
+      "step": 27226
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393281041878238e-05,
+      "loss": 0.3376,
+      "step": 27227
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939322468662567e-05,
+      "loss": 0.3174,
+      "step": 27228
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939316832883783e-05,
+      "loss": 0.2723,
+      "step": 27229
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393111968514727e-05,
+      "loss": 0.3322,
+      "step": 27230
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9393055605656383e-05,
+      "loss": 0.3618,
+      "step": 27231
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392999240262806e-05,
+      "loss": 0.2884,
+      "step": 27232
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392942872334018e-05,
+      "loss": 0.3117,
+      "step": 27233
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392886501870027e-05,
+      "loss": 0.3649,
+      "step": 27234
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392830128870856e-05,
+      "loss": 0.3154,
+      "step": 27235
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392773753336512e-05,
+      "loss": 0.2752,
+      "step": 27236
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939271737526702e-05,
+      "loss": 0.3117,
+      "step": 27237
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392660994662386e-05,
+      "loss": 0.3206,
+      "step": 27238
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392604611522628e-05,
+      "loss": 0.3222,
+      "step": 27239
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939254822584776e-05,
+      "loss": 0.2757,
+      "step": 27240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392491837637803e-05,
+      "loss": 0.3302,
+      "step": 27241
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392435446892767e-05,
+      "loss": 0.3239,
+      "step": 27242
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392379053612673e-05,
+      "loss": 0.2623,
+      "step": 27243
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392322657797527e-05,
+      "loss": 0.3105,
+      "step": 27244
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939226625944735e-05,
+      "loss": 0.3107,
+      "step": 27245
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939220985856216e-05,
+      "loss": 0.3126,
+      "step": 27246
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392153455141965e-05,
+      "loss": 0.3339,
+      "step": 27247
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392097049186785e-05,
+      "loss": 0.3074,
+      "step": 27248
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9392040640696632e-05,
+      "loss": 0.3224,
+      "step": 27249
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9391984229671526e-05,
+      "loss": 0.297,
+      "step": 27250
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939192781611148e-05,
+      "loss": 0.2843,
+      "step": 27251
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9391871400016507e-05,
+      "loss": 0.2963,
+      "step": 27252
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9391814981386624e-05,
+      "loss": 0.3197,
+      "step": 27253
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.939175856022185e-05,
+      "loss": 0.3054,
+      "step": 27254
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.939170213652219e-05,
+      "loss": 0.3699,
+      "step": 27255
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.939164571028767e-05,
+      "loss": 0.344,
+      "step": 27256
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.93915892815183e-05,
+      "loss": 0.294,
+      "step": 27257
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9391532850214098e-05,
+      "loss": 0.324,
+      "step": 27258
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9391476416375074e-05,
+      "loss": 0.2919,
+      "step": 27259
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9391419980001247e-05,
+      "loss": 0.3342,
+      "step": 27260
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9391363541092633e-05,
+      "loss": 0.3162,
+      "step": 27261
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.939130709964925e-05,
+      "loss": 0.3139,
+      "step": 27262
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9391250655671103e-05,
+      "loss": 0.2973,
+      "step": 27263
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9391194209158213e-05,
+      "loss": 0.2631,
+      "step": 27264
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.93911377601106e-05,
+      "loss": 0.3727,
+      "step": 27265
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9391081308528274e-05,
+      "loss": 0.2729,
+      "step": 27266
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9391024854411255e-05,
+      "loss": 0.319,
+      "step": 27267
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390968397759547e-05,
+      "loss": 0.2927,
+      "step": 27268
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390911938573178e-05,
+      "loss": 0.2627,
+      "step": 27269
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390855476852154e-05,
+      "loss": 0.2865,
+      "step": 27270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390799012596496e-05,
+      "loss": 0.3435,
+      "step": 27271
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390742545806217e-05,
+      "loss": 0.3373,
+      "step": 27272
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390686076481335e-05,
+      "loss": 0.2867,
+      "step": 27273
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.939062960462186e-05,
+      "loss": 0.3111,
+      "step": 27274
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390573130227812e-05,
+      "loss": 0.3943,
+      "step": 27275
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390516653299206e-05,
+      "loss": 0.3011,
+      "step": 27276
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.939046017383605e-05,
+      "loss": 0.3398,
+      "step": 27277
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390403691838373e-05,
+      "loss": 0.3049,
+      "step": 27278
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390347207306176e-05,
+      "loss": 0.2887,
+      "step": 27279
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390290720239483e-05,
+      "loss": 0.3483,
+      "step": 27280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390234230638306e-05,
+      "loss": 0.3808,
+      "step": 27281
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.939017773850266e-05,
+      "loss": 0.3189,
+      "step": 27282
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390121243832562e-05,
+      "loss": 0.3119,
+      "step": 27283
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390064746628024e-05,
+      "loss": 0.3019,
+      "step": 27284
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9390008246889068e-05,
+      "loss": 0.3162,
+      "step": 27285
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.93899517446157e-05,
+      "loss": 0.3242,
+      "step": 27286
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389895239807948e-05,
+      "loss": 0.3424,
+      "step": 27287
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389838732465814e-05,
+      "loss": 0.2434,
+      "step": 27288
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389782222589318e-05,
+      "loss": 0.3137,
+      "step": 27289
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389725710178474e-05,
+      "loss": 0.3425,
+      "step": 27290
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389669195233302e-05,
+      "loss": 0.313,
+      "step": 27291
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389612677753814e-05,
+      "loss": 0.3417,
+      "step": 27292
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938955615774003e-05,
+      "loss": 0.2903,
+      "step": 27293
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389499635191954e-05,
+      "loss": 0.3252,
+      "step": 27294
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389443110109613e-05,
+      "loss": 0.3067,
+      "step": 27295
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389386582493014e-05,
+      "loss": 0.3133,
+      "step": 27296
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389330052342177e-05,
+      "loss": 0.3105,
+      "step": 27297
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389273519657116e-05,
+      "loss": 0.3217,
+      "step": 27298
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389216984437847e-05,
+      "loss": 0.3056,
+      "step": 27299
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389160446684385e-05,
+      "loss": 0.3351,
+      "step": 27300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9389103906396742e-05,
+      "loss": 0.3571,
+      "step": 27301
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938904736357494e-05,
+      "loss": 0.2656,
+      "step": 27302
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388990818218983e-05,
+      "loss": 0.3073,
+      "step": 27303
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388934270328898e-05,
+      "loss": 0.2877,
+      "step": 27304
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388877719904697e-05,
+      "loss": 0.3581,
+      "step": 27305
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388821166946393e-05,
+      "loss": 0.2842,
+      "step": 27306
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388764611454004e-05,
+      "loss": 0.2805,
+      "step": 27307
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938870805342754e-05,
+      "loss": 0.2992,
+      "step": 27308
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938865149286702e-05,
+      "loss": 0.3221,
+      "step": 27309
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388594929772463e-05,
+      "loss": 0.3341,
+      "step": 27310
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388538364143875e-05,
+      "loss": 0.3517,
+      "step": 27311
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388481795981283e-05,
+      "loss": 0.2937,
+      "step": 27312
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938842522528469e-05,
+      "loss": 0.3218,
+      "step": 27313
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938836865205412e-05,
+      "loss": 0.3088,
+      "step": 27314
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388312076289582e-05,
+      "loss": 0.2924,
+      "step": 27315
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.93882554979911e-05,
+      "loss": 0.299,
+      "step": 27316
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938819891715868e-05,
+      "loss": 0.2899,
+      "step": 27317
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388142333792343e-05,
+      "loss": 0.2988,
+      "step": 27318
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388085747892105e-05,
+      "loss": 0.3087,
+      "step": 27319
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9388029159457975e-05,
+      "loss": 0.3322,
+      "step": 27320
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387972568489972e-05,
+      "loss": 0.3565,
+      "step": 27321
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938791597498811e-05,
+      "loss": 0.2687,
+      "step": 27322
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938785937895241e-05,
+      "loss": 0.2983,
+      "step": 27323
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387802780382882e-05,
+      "loss": 0.3052,
+      "step": 27324
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938774617927954e-05,
+      "loss": 0.3092,
+      "step": 27325
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387689575642402e-05,
+      "loss": 0.3667,
+      "step": 27326
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387632969471483e-05,
+      "loss": 0.3773,
+      "step": 27327
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387576360766797e-05,
+      "loss": 0.2907,
+      "step": 27328
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387519749528365e-05,
+      "loss": 0.2989,
+      "step": 27329
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387463135756192e-05,
+      "loss": 0.2746,
+      "step": 27330
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387406519450304e-05,
+      "loss": 0.3182,
+      "step": 27331
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387349900610702e-05,
+      "loss": 0.3056,
+      "step": 27332
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938729327923742e-05,
+      "loss": 0.3068,
+      "step": 27333
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938723665533046e-05,
+      "loss": 0.284,
+      "step": 27334
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938718002888984e-05,
+      "loss": 0.2825,
+      "step": 27335
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387123399915577e-05,
+      "loss": 0.2637,
+      "step": 27336
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9387066768407686e-05,
+      "loss": 0.3005,
+      "step": 27337
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938701013436618e-05,
+      "loss": 0.291,
+      "step": 27338
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386953497791076e-05,
+      "loss": 0.3077,
+      "step": 27339
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386896858682393e-05,
+      "loss": 0.3235,
+      "step": 27340
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386840217040143e-05,
+      "loss": 0.2778,
+      "step": 27341
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386783572864336e-05,
+      "loss": 0.3558,
+      "step": 27342
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386726926154996e-05,
+      "loss": 0.3135,
+      "step": 27343
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386670276912133e-05,
+      "loss": 0.3298,
+      "step": 27344
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386613625135764e-05,
+      "loss": 0.3302,
+      "step": 27345
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386556970825902e-05,
+      "loss": 0.3332,
+      "step": 27346
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386500313982565e-05,
+      "loss": 0.3132,
+      "step": 27347
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938644365460577e-05,
+      "loss": 0.3442,
+      "step": 27348
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386386992695532e-05,
+      "loss": 0.2644,
+      "step": 27349
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938633032825186e-05,
+      "loss": 0.3408,
+      "step": 27350
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9386273661274774e-05,
+      "loss": 0.3838,
+      "step": 27351
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938621699176429e-05,
+      "loss": 0.2971,
+      "step": 27352
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.938616031972042e-05,
+      "loss": 0.3709,
+      "step": 27353
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9386103645143185e-05,
+      "loss": 0.3014,
+      "step": 27354
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9386046968032596e-05,
+      "loss": 0.3069,
+      "step": 27355
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938599028838867e-05,
+      "loss": 0.3269,
+      "step": 27356
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938593360621142e-05,
+      "loss": 0.3011,
+      "step": 27357
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938587692150086e-05,
+      "loss": 0.323,
+      "step": 27358
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385820234257012e-05,
+      "loss": 0.3418,
+      "step": 27359
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385763544479882e-05,
+      "loss": 0.2881,
+      "step": 27360
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385706852169497e-05,
+      "loss": 0.3401,
+      "step": 27361
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385650157325863e-05,
+      "loss": 0.3512,
+      "step": 27362
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385593459948998e-05,
+      "loss": 0.3221,
+      "step": 27363
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385536760038917e-05,
+      "loss": 0.277,
+      "step": 27364
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385480057595636e-05,
+      "loss": 0.3104,
+      "step": 27365
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938542335261917e-05,
+      "loss": 0.3074,
+      "step": 27366
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385366645109535e-05,
+      "loss": 0.3399,
+      "step": 27367
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385309935066745e-05,
+      "loss": 0.302,
+      "step": 27368
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385253222490815e-05,
+      "loss": 0.2832,
+      "step": 27369
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385196507381762e-05,
+      "loss": 0.2621,
+      "step": 27370
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385139789739603e-05,
+      "loss": 0.3427,
+      "step": 27371
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385083069564348e-05,
+      "loss": 0.3038,
+      "step": 27372
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9385026346856017e-05,
+      "loss": 0.26,
+      "step": 27373
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9384969621614624e-05,
+      "loss": 0.283,
+      "step": 27374
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9384912893840182e-05,
+      "loss": 0.3363,
+      "step": 27375
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938485616353271e-05,
+      "loss": 0.2658,
+      "step": 27376
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9384799430692217e-05,
+      "loss": 0.3495,
+      "step": 27377
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9384742695318727e-05,
+      "loss": 0.2887,
+      "step": 27378
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938468595741225e-05,
+      "loss": 0.3646,
+      "step": 27379
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9384629216972802e-05,
+      "loss": 0.3193,
+      "step": 27380
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.93845724740004e-05,
+      "loss": 0.2936,
+      "step": 27381
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938451572849506e-05,
+      "loss": 0.2468,
+      "step": 27382
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938445898045679e-05,
+      "loss": 0.3326,
+      "step": 27383
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9384402229885612e-05,
+      "loss": 0.3061,
+      "step": 27384
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938434547678154e-05,
+      "loss": 0.3048,
+      "step": 27385
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9384288721144594e-05,
+      "loss": 0.3959,
+      "step": 27386
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938423196297478e-05,
+      "loss": 0.2843,
+      "step": 27387
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938417520227212e-05,
+      "loss": 0.3272,
+      "step": 27388
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9384118439036627e-05,
+      "loss": 0.263,
+      "step": 27389
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9384061673268312e-05,
+      "loss": 0.3592,
+      "step": 27390
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.93840049049672e-05,
+      "loss": 0.3118,
+      "step": 27391
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.93839481341333e-05,
+      "loss": 0.3312,
+      "step": 27392
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383891360766628e-05,
+      "loss": 0.3413,
+      "step": 27393
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.93838345848672e-05,
+      "loss": 0.3273,
+      "step": 27394
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383777806435034e-05,
+      "loss": 0.3174,
+      "step": 27395
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383721025470138e-05,
+      "loss": 0.2539,
+      "step": 27396
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383664241972533e-05,
+      "loss": 0.3194,
+      "step": 27397
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383607455942234e-05,
+      "loss": 0.3763,
+      "step": 27398
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383550667379257e-05,
+      "loss": 0.3232,
+      "step": 27399
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383493876283613e-05,
+      "loss": 0.2693,
+      "step": 27400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383437082655324e-05,
+      "loss": 0.348,
+      "step": 27401
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383380286494397e-05,
+      "loss": 0.2969,
+      "step": 27402
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383323487800854e-05,
+      "loss": 0.3025,
+      "step": 27403
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383266686574707e-05,
+      "loss": 0.2749,
+      "step": 27404
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938320988281597e-05,
+      "loss": 0.2702,
+      "step": 27405
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383153076524666e-05,
+      "loss": 0.3637,
+      "step": 27406
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9383096267700802e-05,
+      "loss": 0.3034,
+      "step": 27407
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.93830394563444e-05,
+      "loss": 0.326,
+      "step": 27408
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938298264245547e-05,
+      "loss": 0.3358,
+      "step": 27409
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382925826034027e-05,
+      "loss": 0.381,
+      "step": 27410
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938286900708009e-05,
+      "loss": 0.3274,
+      "step": 27411
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382812185593673e-05,
+      "loss": 0.3426,
+      "step": 27412
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938275536157479e-05,
+      "loss": 0.3136,
+      "step": 27413
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938269853502346e-05,
+      "loss": 0.3498,
+      "step": 27414
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382641705939693e-05,
+      "loss": 0.3377,
+      "step": 27415
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382584874323508e-05,
+      "loss": 0.3212,
+      "step": 27416
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938252804017492e-05,
+      "loss": 0.3111,
+      "step": 27417
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382471203493943e-05,
+      "loss": 0.3337,
+      "step": 27418
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382414364280594e-05,
+      "loss": 0.2887,
+      "step": 27419
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382357522534888e-05,
+      "loss": 0.3319,
+      "step": 27420
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382300678256837e-05,
+      "loss": 0.2842,
+      "step": 27421
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382243831446463e-05,
+      "loss": 0.3587,
+      "step": 27422
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382186982103776e-05,
+      "loss": 0.3357,
+      "step": 27423
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382130130228792e-05,
+      "loss": 0.2913,
+      "step": 27424
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938207327582153e-05,
+      "loss": 0.2993,
+      "step": 27425
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9382016418882e-05,
+      "loss": 0.3321,
+      "step": 27426
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938195955941022e-05,
+      "loss": 0.3016,
+      "step": 27427
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9381902697406207e-05,
+      "loss": 0.3386,
+      "step": 27428
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938184583286997e-05,
+      "loss": 0.2586,
+      "step": 27429
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9381788965801537e-05,
+      "loss": 0.2698,
+      "step": 27430
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938173209620091e-05,
+      "loss": 0.3013,
+      "step": 27431
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938167522406811e-05,
+      "loss": 0.3242,
+      "step": 27432
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938161834940315e-05,
+      "loss": 0.2838,
+      "step": 27433
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938156147220605e-05,
+      "loss": 0.3245,
+      "step": 27434
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938150459247682e-05,
+      "loss": 0.2955,
+      "step": 27435
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9381447710215485e-05,
+      "loss": 0.3488,
+      "step": 27436
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9381390825422047e-05,
+      "loss": 0.2951,
+      "step": 27437
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9381333938096528e-05,
+      "loss": 0.2892,
+      "step": 27438
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9381277048238947e-05,
+      "loss": 0.3169,
+      "step": 27439
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938122015584931e-05,
+      "loss": 0.3333,
+      "step": 27440
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938116326092764e-05,
+      "loss": 0.3041,
+      "step": 27441
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9381106363473953e-05,
+      "loss": 0.3284,
+      "step": 27442
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938104946348826e-05,
+      "loss": 0.3301,
+      "step": 27443
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9380992560970576e-05,
+      "loss": 0.328,
+      "step": 27444
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9380935655920924e-05,
+      "loss": 0.3118,
+      "step": 27445
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9380878748339307e-05,
+      "loss": 0.3511,
+      "step": 27446
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938082183822575e-05,
+      "loss": 0.3382,
+      "step": 27447
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9380764925580265e-05,
+      "loss": 0.3106,
+      "step": 27448
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9380708010402866e-05,
+      "loss": 0.2902,
+      "step": 27449
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.938065109269357e-05,
+      "loss": 0.3211,
+      "step": 27450
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9380594172452397e-05,
+      "loss": 0.3381,
+      "step": 27451
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9380537249679353e-05,
+      "loss": 0.3917,
+      "step": 27452
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.938048032437446e-05,
+      "loss": 0.3301,
+      "step": 27453
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.938042339653773e-05,
+      "loss": 0.2825,
+      "step": 27454
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9380366466169182e-05,
+      "loss": 0.2899,
+      "step": 27455
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.938030953326883e-05,
+      "loss": 0.3519,
+      "step": 27456
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9380252597836688e-05,
+      "loss": 0.3304,
+      "step": 27457
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.938019565987277e-05,
+      "loss": 0.3266,
+      "step": 27458
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9380138719377095e-05,
+      "loss": 0.3048,
+      "step": 27459
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9380081776349677e-05,
+      "loss": 0.294,
+      "step": 27460
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9380024830790528e-05,
+      "loss": 0.2898,
+      "step": 27461
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937996788269967e-05,
+      "loss": 0.3129,
+      "step": 27462
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379910932077113e-05,
+      "loss": 0.2889,
+      "step": 27463
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379853978922874e-05,
+      "loss": 0.3113,
+      "step": 27464
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937979702323697e-05,
+      "loss": 0.276,
+      "step": 27465
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937974006501941e-05,
+      "loss": 0.308,
+      "step": 27466
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937968310427022e-05,
+      "loss": 0.3504,
+      "step": 27467
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379626140989408e-05,
+      "loss": 0.277,
+      "step": 27468
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937956917517699e-05,
+      "loss": 0.3146,
+      "step": 27469
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379512206832987e-05,
+      "loss": 0.3067,
+      "step": 27470
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379455235957406e-05,
+      "loss": 0.3568,
+      "step": 27471
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379398262550267e-05,
+      "loss": 0.2761,
+      "step": 27472
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379341286611583e-05,
+      "loss": 0.2738,
+      "step": 27473
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379284308141373e-05,
+      "loss": 0.2669,
+      "step": 27474
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379227327139648e-05,
+      "loss": 0.3643,
+      "step": 27475
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379170343606427e-05,
+      "loss": 0.2687,
+      "step": 27476
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379113357541726e-05,
+      "loss": 0.3244,
+      "step": 27477
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9379056368945558e-05,
+      "loss": 0.3435,
+      "step": 27478
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378999377817934e-05,
+      "loss": 0.3485,
+      "step": 27479
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937894238415888e-05,
+      "loss": 0.3054,
+      "step": 27480
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378885387968402e-05,
+      "loss": 0.3176,
+      "step": 27481
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937882838924652e-05,
+      "loss": 0.3095,
+      "step": 27482
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937877138799325e-05,
+      "loss": 0.358,
+      "step": 27483
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378714384208605e-05,
+      "loss": 0.369,
+      "step": 27484
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.93786573778926e-05,
+      "loss": 0.3135,
+      "step": 27485
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937860036904525e-05,
+      "loss": 0.3189,
+      "step": 27486
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378543357666576e-05,
+      "loss": 0.2996,
+      "step": 27487
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378486343756588e-05,
+      "loss": 0.2955,
+      "step": 27488
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378429327315303e-05,
+      "loss": 0.2991,
+      "step": 27489
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378372308342735e-05,
+      "loss": 0.2887,
+      "step": 27490
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378315286838897e-05,
+      "loss": 0.3116,
+      "step": 27491
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378258262803812e-05,
+      "loss": 0.3022,
+      "step": 27492
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378201236237492e-05,
+      "loss": 0.2805,
+      "step": 27493
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378144207139953e-05,
+      "loss": 0.2618,
+      "step": 27494
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378087175511204e-05,
+      "loss": 0.3113,
+      "step": 27495
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9378030141351268e-05,
+      "loss": 0.3621,
+      "step": 27496
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937797310466016e-05,
+      "loss": 0.3109,
+      "step": 27497
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937791606543789e-05,
+      "loss": 0.286,
+      "step": 27498
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377859023684475e-05,
+      "loss": 0.3461,
+      "step": 27499
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377801979399937e-05,
+      "loss": 0.3084,
+      "step": 27500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377744932584284e-05,
+      "loss": 0.3395,
+      "step": 27501
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937768788323753e-05,
+      "loss": 0.3088,
+      "step": 27502
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.93776308313597e-05,
+      "loss": 0.3283,
+      "step": 27503
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.93775737769508e-05,
+      "loss": 0.2892,
+      "step": 27504
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377516720010852e-05,
+      "loss": 0.3555,
+      "step": 27505
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377459660539867e-05,
+      "loss": 0.2665,
+      "step": 27506
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377402598537866e-05,
+      "loss": 0.2935,
+      "step": 27507
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377345534004855e-05,
+      "loss": 0.3016,
+      "step": 27508
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377288466940857e-05,
+      "loss": 0.2883,
+      "step": 27509
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377231397345883e-05,
+      "loss": 0.4074,
+      "step": 27510
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937717432521995e-05,
+      "loss": 0.3524,
+      "step": 27511
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377117250563074e-05,
+      "loss": 0.3214,
+      "step": 27512
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937706017337527e-05,
+      "loss": 0.3228,
+      "step": 27513
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9377003093656554e-05,
+      "loss": 0.3018,
+      "step": 27514
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376946011406942e-05,
+      "loss": 0.2871,
+      "step": 27515
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937688892662645e-05,
+      "loss": 0.2901,
+      "step": 27516
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937683183931509e-05,
+      "loss": 0.3776,
+      "step": 27517
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376774749472878e-05,
+      "loss": 0.3486,
+      "step": 27518
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376717657099832e-05,
+      "loss": 0.317,
+      "step": 27519
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376660562195968e-05,
+      "loss": 0.3552,
+      "step": 27520
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376603464761298e-05,
+      "loss": 0.343,
+      "step": 27521
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376546364795836e-05,
+      "loss": 0.2553,
+      "step": 27522
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.93764892622996e-05,
+      "loss": 0.3842,
+      "step": 27523
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376432157272613e-05,
+      "loss": 0.3332,
+      "step": 27524
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376375049714876e-05,
+      "loss": 0.2737,
+      "step": 27525
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376317939626418e-05,
+      "loss": 0.2918,
+      "step": 27526
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376260827007242e-05,
+      "loss": 0.3054,
+      "step": 27527
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376203711857372e-05,
+      "loss": 0.3607,
+      "step": 27528
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937614659417682e-05,
+      "loss": 0.2639,
+      "step": 27529
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.93760894739656e-05,
+      "loss": 0.3043,
+      "step": 27530
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9376032351223736e-05,
+      "loss": 0.3559,
+      "step": 27531
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937597522595123e-05,
+      "loss": 0.3296,
+      "step": 27532
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375918098148108e-05,
+      "loss": 0.2827,
+      "step": 27533
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375860967814383e-05,
+      "loss": 0.3031,
+      "step": 27534
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375803834950066e-05,
+      "loss": 0.3228,
+      "step": 27535
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375746699555178e-05,
+      "loss": 0.3163,
+      "step": 27536
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937568956162973e-05,
+      "loss": 0.3521,
+      "step": 27537
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375632421173742e-05,
+      "loss": 0.3253,
+      "step": 27538
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375575278187226e-05,
+      "loss": 0.3352,
+      "step": 27539
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375518132670198e-05,
+      "loss": 0.3052,
+      "step": 27540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375460984622677e-05,
+      "loss": 0.308,
+      "step": 27541
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375403834044672e-05,
+      "loss": 0.3062,
+      "step": 27542
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.93753466809362e-05,
+      "loss": 0.2883,
+      "step": 27543
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937528952529728e-05,
+      "loss": 0.3468,
+      "step": 27544
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375232367127926e-05,
+      "loss": 0.293,
+      "step": 27545
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375175206428155e-05,
+      "loss": 0.2993,
+      "step": 27546
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.937511804319798e-05,
+      "loss": 0.3187,
+      "step": 27547
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375060877437413e-05,
+      "loss": 0.2959,
+      "step": 27548
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9375003709146475e-05,
+      "loss": 0.3055,
+      "step": 27549
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9374946538325182e-05,
+      "loss": 0.3306,
+      "step": 27550
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374889364973544e-05,
+      "loss": 0.3339,
+      "step": 27551
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374832189091578e-05,
+      "loss": 0.2895,
+      "step": 27552
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374775010679308e-05,
+      "loss": 0.3455,
+      "step": 27553
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374717829736737e-05,
+      "loss": 0.3507,
+      "step": 27554
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374660646263885e-05,
+      "loss": 0.3096,
+      "step": 27555
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937460346026077e-05,
+      "loss": 0.2859,
+      "step": 27556
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937454627172741e-05,
+      "loss": 0.294,
+      "step": 27557
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374489080663807e-05,
+      "loss": 0.2983,
+      "step": 27558
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374431887069993e-05,
+      "loss": 0.3177,
+      "step": 27559
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374374690945972e-05,
+      "loss": 0.3434,
+      "step": 27560
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374317492291766e-05,
+      "loss": 0.3428,
+      "step": 27561
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374260291107388e-05,
+      "loss": 0.3123,
+      "step": 27562
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937420308739285e-05,
+      "loss": 0.3532,
+      "step": 27563
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374145881148173e-05,
+      "loss": 0.3084,
+      "step": 27564
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374088672373373e-05,
+      "loss": 0.3116,
+      "step": 27565
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9374031461068456e-05,
+      "loss": 0.3137,
+      "step": 27566
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937397424723345e-05,
+      "loss": 0.3017,
+      "step": 27567
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373917030868365e-05,
+      "loss": 0.3058,
+      "step": 27568
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373859811973212e-05,
+      "loss": 0.2662,
+      "step": 27569
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937380259054801e-05,
+      "loss": 0.3144,
+      "step": 27570
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373745366592777e-05,
+      "loss": 0.3635,
+      "step": 27571
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373688140107528e-05,
+      "loss": 0.264,
+      "step": 27572
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373630911092273e-05,
+      "loss": 0.2872,
+      "step": 27573
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373573679547032e-05,
+      "loss": 0.3051,
+      "step": 27574
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937351644547182e-05,
+      "loss": 0.2872,
+      "step": 27575
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373459208866654e-05,
+      "loss": 0.314,
+      "step": 27576
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373401969731547e-05,
+      "loss": 0.2888,
+      "step": 27577
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373344728066516e-05,
+      "loss": 0.288,
+      "step": 27578
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373287483871573e-05,
+      "loss": 0.3096,
+      "step": 27579
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373230237146736e-05,
+      "loss": 0.3008,
+      "step": 27580
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373172987892022e-05,
+      "loss": 0.2926,
+      "step": 27581
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937311573610744e-05,
+      "loss": 0.2968,
+      "step": 27582
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373058481793016e-05,
+      "loss": 0.2806,
+      "step": 27583
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9373001224948757e-05,
+      "loss": 0.3002,
+      "step": 27584
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372943965574683e-05,
+      "loss": 0.3232,
+      "step": 27585
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937288670367081e-05,
+      "loss": 0.3166,
+      "step": 27586
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372829439237146e-05,
+      "loss": 0.3033,
+      "step": 27587
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372772172273713e-05,
+      "loss": 0.2986,
+      "step": 27588
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372714902780522e-05,
+      "loss": 0.2938,
+      "step": 27589
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372657630757596e-05,
+      "loss": 0.2599,
+      "step": 27590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372600356204945e-05,
+      "loss": 0.3022,
+      "step": 27591
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372543079122583e-05,
+      "loss": 0.3522,
+      "step": 27592
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937248579951053e-05,
+      "loss": 0.3264,
+      "step": 27593
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.93724285173688e-05,
+      "loss": 0.3845,
+      "step": 27594
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372371232697405e-05,
+      "loss": 0.3686,
+      "step": 27595
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372313945496363e-05,
+      "loss": 0.303,
+      "step": 27596
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372256655765695e-05,
+      "loss": 0.3462,
+      "step": 27597
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372199363505407e-05,
+      "loss": 0.3109,
+      "step": 27598
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937214206871552e-05,
+      "loss": 0.3095,
+      "step": 27599
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372084771396046e-05,
+      "loss": 0.3323,
+      "step": 27600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9372027471547e-05,
+      "loss": 0.3484,
+      "step": 27601
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937197016916841e-05,
+      "loss": 0.3359,
+      "step": 27602
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371912864260272e-05,
+      "loss": 0.323,
+      "step": 27603
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371855556822614e-05,
+      "loss": 0.3031,
+      "step": 27604
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371798246855448e-05,
+      "loss": 0.3183,
+      "step": 27605
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937174093435879e-05,
+      "loss": 0.3374,
+      "step": 27606
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371683619332656e-05,
+      "loss": 0.2933,
+      "step": 27607
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937162630177706e-05,
+      "loss": 0.2888,
+      "step": 27608
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371568981692018e-05,
+      "loss": 0.2783,
+      "step": 27609
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371511659077545e-05,
+      "loss": 0.3227,
+      "step": 27610
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937145433393366e-05,
+      "loss": 0.3127,
+      "step": 27611
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371397006260374e-05,
+      "loss": 0.3097,
+      "step": 27612
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371339676057706e-05,
+      "loss": 0.3427,
+      "step": 27613
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937128234332567e-05,
+      "loss": 0.3532,
+      "step": 27614
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371225008064276e-05,
+      "loss": 0.3473,
+      "step": 27615
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371167670273548e-05,
+      "loss": 0.3266,
+      "step": 27616
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9371110329953494e-05,
+      "loss": 0.2717,
+      "step": 27617
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937105298710414e-05,
+      "loss": 0.342,
+      "step": 27618
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937099564172549e-05,
+      "loss": 0.2497,
+      "step": 27619
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370938293817567e-05,
+      "loss": 0.2887,
+      "step": 27620
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370880943380387e-05,
+      "loss": 0.2572,
+      "step": 27621
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370823590413956e-05,
+      "loss": 0.2564,
+      "step": 27622
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370766234918297e-05,
+      "loss": 0.307,
+      "step": 27623
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937070887689343e-05,
+      "loss": 0.339,
+      "step": 27624
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937065151633936e-05,
+      "loss": 0.3179,
+      "step": 27625
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370594153256108e-05,
+      "loss": 0.3023,
+      "step": 27626
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937053678764369e-05,
+      "loss": 0.2986,
+      "step": 27627
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937047941950212e-05,
+      "loss": 0.312,
+      "step": 27628
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370422048831412e-05,
+      "loss": 0.2904,
+      "step": 27629
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370364675631587e-05,
+      "loss": 0.3017,
+      "step": 27630
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370307299902653e-05,
+      "loss": 0.3757,
+      "step": 27631
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937024992164463e-05,
+      "loss": 0.2855,
+      "step": 27632
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370192540857535e-05,
+      "loss": 0.3483,
+      "step": 27633
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370135157541378e-05,
+      "loss": 0.2844,
+      "step": 27634
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.937007777169618e-05,
+      "loss": 0.2389,
+      "step": 27635
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9370020383321956e-05,
+      "loss": 0.3585,
+      "step": 27636
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9369962992418718e-05,
+      "loss": 0.3459,
+      "step": 27637
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9369905598986477e-05,
+      "loss": 0.2924,
+      "step": 27638
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9369848203025263e-05,
+      "loss": 0.3053,
+      "step": 27639
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.936979080453508e-05,
+      "loss": 0.3471,
+      "step": 27640
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9369733403515948e-05,
+      "loss": 0.2722,
+      "step": 27641
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.936967599996788e-05,
+      "loss": 0.2694,
+      "step": 27642
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.936961859389089e-05,
+      "loss": 0.2907,
+      "step": 27643
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9369561185285e-05,
+      "loss": 0.296,
+      "step": 27644
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9369503774150224e-05,
+      "loss": 0.379,
+      "step": 27645
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.936944636048657e-05,
+      "loss": 0.267,
+      "step": 27646
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.936938894429406e-05,
+      "loss": 0.3198,
+      "step": 27647
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9369331525572708e-05,
+      "loss": 0.2775,
+      "step": 27648
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936927410432253e-05,
+      "loss": 0.3433,
+      "step": 27649
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936921668054354e-05,
+      "loss": 0.3208,
+      "step": 27650
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9369159254235755e-05,
+      "loss": 0.3096,
+      "step": 27651
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9369101825399194e-05,
+      "loss": 0.3058,
+      "step": 27652
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9369044394033863e-05,
+      "loss": 0.3248,
+      "step": 27653
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936898696013979e-05,
+      "loss": 0.3052,
+      "step": 27654
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9368929523716978e-05,
+      "loss": 0.2977,
+      "step": 27655
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936887208476545e-05,
+      "loss": 0.2951,
+      "step": 27656
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9368814643285218e-05,
+      "loss": 0.3299,
+      "step": 27657
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.93687571992763e-05,
+      "loss": 0.3224,
+      "step": 27658
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9368699752738712e-05,
+      "loss": 0.2941,
+      "step": 27659
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9368642303672466e-05,
+      "loss": 0.3152,
+      "step": 27660
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936858485207758e-05,
+      "loss": 0.3067,
+      "step": 27661
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936852739795407e-05,
+      "loss": 0.3201,
+      "step": 27662
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9368469941301952e-05,
+      "loss": 0.3013,
+      "step": 27663
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936841248212124e-05,
+      "loss": 0.3198,
+      "step": 27664
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9368355020411945e-05,
+      "loss": 0.3395,
+      "step": 27665
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9368297556174092e-05,
+      "loss": 0.3302,
+      "step": 27666
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936824008940769e-05,
+      "loss": 0.3603,
+      "step": 27667
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9368182620112756e-05,
+      "loss": 0.305,
+      "step": 27668
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9368125148289306e-05,
+      "loss": 0.3201,
+      "step": 27669
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9368067673937355e-05,
+      "loss": 0.3436,
+      "step": 27670
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936801019705692e-05,
+      "loss": 0.3063,
+      "step": 27671
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367952717648014e-05,
+      "loss": 0.3014,
+      "step": 27672
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367895235710656e-05,
+      "loss": 0.2892,
+      "step": 27673
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367837751244857e-05,
+      "loss": 0.322,
+      "step": 27674
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367780264250634e-05,
+      "loss": 0.3094,
+      "step": 27675
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367722774728004e-05,
+      "loss": 0.2918,
+      "step": 27676
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367665282676984e-05,
+      "loss": 0.3316,
+      "step": 27677
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367607788097587e-05,
+      "loss": 0.3288,
+      "step": 27678
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367550290989828e-05,
+      "loss": 0.3041,
+      "step": 27679
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367492791353723e-05,
+      "loss": 0.3179,
+      "step": 27680
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367435289189285e-05,
+      "loss": 0.3256,
+      "step": 27681
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367377784496536e-05,
+      "loss": 0.3005,
+      "step": 27682
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367320277275488e-05,
+      "loss": 0.3004,
+      "step": 27683
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367262767526155e-05,
+      "loss": 0.3641,
+      "step": 27684
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367205255248554e-05,
+      "loss": 0.3689,
+      "step": 27685
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367147740442702e-05,
+      "loss": 0.3111,
+      "step": 27686
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9367090223108613e-05,
+      "loss": 0.3377,
+      "step": 27687
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.93670327032463e-05,
+      "loss": 0.3581,
+      "step": 27688
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366975180855786e-05,
+      "loss": 0.2718,
+      "step": 27689
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936691765593708e-05,
+      "loss": 0.3682,
+      "step": 27690
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366860128490195e-05,
+      "loss": 0.3243,
+      "step": 27691
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366802598515152e-05,
+      "loss": 0.3286,
+      "step": 27692
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366745066011966e-05,
+      "loss": 0.3579,
+      "step": 27693
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366687530980655e-05,
+      "loss": 0.3905,
+      "step": 27694
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366629993421228e-05,
+      "loss": 0.3033,
+      "step": 27695
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366572453333706e-05,
+      "loss": 0.3453,
+      "step": 27696
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.93665149107181e-05,
+      "loss": 0.279,
+      "step": 27697
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366457365574427e-05,
+      "loss": 0.278,
+      "step": 27698
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366399817902707e-05,
+      "loss": 0.2568,
+      "step": 27699
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936634226770295e-05,
+      "loss": 0.3197,
+      "step": 27700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936628471497517e-05,
+      "loss": 0.3094,
+      "step": 27701
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936622715971939e-05,
+      "loss": 0.346,
+      "step": 27702
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366169601935622e-05,
+      "loss": 0.2557,
+      "step": 27703
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9366112041623877e-05,
+      "loss": 0.2963,
+      "step": 27704
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936605447878418e-05,
+      "loss": 0.2854,
+      "step": 27705
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365996913416535e-05,
+      "loss": 0.2641,
+      "step": 27706
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936593934552097e-05,
+      "loss": 0.3269,
+      "step": 27707
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936588177509749e-05,
+      "loss": 0.3294,
+      "step": 27708
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365824202146116e-05,
+      "loss": 0.3119,
+      "step": 27709
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936576662666686e-05,
+      "loss": 0.3785,
+      "step": 27710
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365709048659744e-05,
+      "loss": 0.3001,
+      "step": 27711
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936565146812478e-05,
+      "loss": 0.3019,
+      "step": 27712
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936559388506198e-05,
+      "loss": 0.3419,
+      "step": 27713
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365536299471363e-05,
+      "loss": 0.3268,
+      "step": 27714
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365478711352942e-05,
+      "loss": 0.3576,
+      "step": 27715
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365421120706735e-05,
+      "loss": 0.2855,
+      "step": 27716
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365363527532758e-05,
+      "loss": 0.3303,
+      "step": 27717
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365305931831028e-05,
+      "loss": 0.3457,
+      "step": 27718
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365248333601555e-05,
+      "loss": 0.3079,
+      "step": 27719
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936519073284436e-05,
+      "loss": 0.355,
+      "step": 27720
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365133129559455e-05,
+      "loss": 0.3367,
+      "step": 27721
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9365075523746855e-05,
+      "loss": 0.3475,
+      "step": 27722
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936501791540658e-05,
+      "loss": 0.2731,
+      "step": 27723
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936496030453864e-05,
+      "loss": 0.2974,
+      "step": 27724
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9364902691143056e-05,
+      "loss": 0.3453,
+      "step": 27725
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9364845075219844e-05,
+      "loss": 0.2891,
+      "step": 27726
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936478745676901e-05,
+      "loss": 0.3007,
+      "step": 27727
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936472983579058e-05,
+      "loss": 0.3166,
+      "step": 27728
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9364672212284564e-05,
+      "loss": 0.3221,
+      "step": 27729
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936461458625098e-05,
+      "loss": 0.2964,
+      "step": 27730
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9364556957689844e-05,
+      "loss": 0.3255,
+      "step": 27731
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936449932660117e-05,
+      "loss": 0.2883,
+      "step": 27732
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936444169298497e-05,
+      "loss": 0.2997,
+      "step": 27733
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9364384056841267e-05,
+      "loss": 0.317,
+      "step": 27734
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936432641817007e-05,
+      "loss": 0.3342,
+      "step": 27735
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9364268776971402e-05,
+      "loss": 0.2982,
+      "step": 27736
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9364211133245272e-05,
+      "loss": 0.2644,
+      "step": 27737
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9364153486991698e-05,
+      "loss": 0.3215,
+      "step": 27738
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9364095838210692e-05,
+      "loss": 0.3325,
+      "step": 27739
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9364038186902275e-05,
+      "loss": 0.2716,
+      "step": 27740
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9363980533066465e-05,
+      "loss": 0.3609,
+      "step": 27741
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9363922876703264e-05,
+      "loss": 0.3277,
+      "step": 27742
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9363865217812703e-05,
+      "loss": 0.3198,
+      "step": 27743
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936380755639479e-05,
+      "loss": 0.3255,
+      "step": 27744
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.936374989244954e-05,
+      "loss": 0.3486,
+      "step": 27745
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9363692225976973e-05,
+      "loss": 0.2871,
+      "step": 27746
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9363634556977094e-05,
+      "loss": 0.3415,
+      "step": 27747
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9363576885449937e-05,
+      "loss": 0.3141,
+      "step": 27748
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9363519211395497e-05,
+      "loss": 0.3458,
+      "step": 27749
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9363461534813806e-05,
+      "loss": 0.3175,
+      "step": 27750
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936340385570487e-05,
+      "loss": 0.3125,
+      "step": 27751
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936334617406871e-05,
+      "loss": 0.3329,
+      "step": 27752
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9363288489905336e-05,
+      "loss": 0.3197,
+      "step": 27753
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936323080321477e-05,
+      "loss": 0.3347,
+      "step": 27754
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9363173113997022e-05,
+      "loss": 0.2931,
+      "step": 27755
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936311542225211e-05,
+      "loss": 0.3277,
+      "step": 27756
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9363057727980047e-05,
+      "loss": 0.2963,
+      "step": 27757
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9363000031180856e-05,
+      "loss": 0.2883,
+      "step": 27758
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362942331854545e-05,
+      "loss": 0.3356,
+      "step": 27759
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362884630001134e-05,
+      "loss": 0.3164,
+      "step": 27760
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362826925620632e-05,
+      "loss": 0.3258,
+      "step": 27761
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362769218713064e-05,
+      "loss": 0.3367,
+      "step": 27762
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362711509278438e-05,
+      "loss": 0.3075,
+      "step": 27763
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362653797316774e-05,
+      "loss": 0.2923,
+      "step": 27764
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362596082828085e-05,
+      "loss": 0.3297,
+      "step": 27765
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362538365812388e-05,
+      "loss": 0.3402,
+      "step": 27766
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362480646269698e-05,
+      "loss": 0.3371,
+      "step": 27767
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936242292420003e-05,
+      "loss": 0.3029,
+      "step": 27768
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362365199603403e-05,
+      "loss": 0.3095,
+      "step": 27769
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362307472479828e-05,
+      "loss": 0.3036,
+      "step": 27770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936224974282932e-05,
+      "loss": 0.3016,
+      "step": 27771
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.93621920106519e-05,
+      "loss": 0.3812,
+      "step": 27772
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362134275947578e-05,
+      "loss": 0.2632,
+      "step": 27773
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362076538716373e-05,
+      "loss": 0.26,
+      "step": 27774
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9362018798958303e-05,
+      "loss": 0.3233,
+      "step": 27775
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936196105667338e-05,
+      "loss": 0.3058,
+      "step": 27776
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9361903311861617e-05,
+      "loss": 0.3013,
+      "step": 27777
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9361845564523036e-05,
+      "loss": 0.289,
+      "step": 27778
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9361787814657646e-05,
+      "loss": 0.3617,
+      "step": 27779
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9361730062265467e-05,
+      "loss": 0.3755,
+      "step": 27780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9361672307346513e-05,
+      "loss": 0.2885,
+      "step": 27781
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.93616145499008e-05,
+      "loss": 0.3381,
+      "step": 27782
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936155678992834e-05,
+      "loss": 0.2931,
+      "step": 27783
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9361499027429154e-05,
+      "loss": 0.3646,
+      "step": 27784
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936144126240326e-05,
+      "loss": 0.335,
+      "step": 27785
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9361383494850665e-05,
+      "loss": 0.3007,
+      "step": 27786
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936132572477139e-05,
+      "loss": 0.3094,
+      "step": 27787
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936126795216545e-05,
+      "loss": 0.3673,
+      "step": 27788
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936121017703286e-05,
+      "loss": 0.2807,
+      "step": 27789
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9361152399373635e-05,
+      "loss": 0.2534,
+      "step": 27790
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936109461918779e-05,
+      "loss": 0.2761,
+      "step": 27791
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9361036836475344e-05,
+      "loss": 0.289,
+      "step": 27792
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936097905123631e-05,
+      "loss": 0.3018,
+      "step": 27793
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360921263470703e-05,
+      "loss": 0.3309,
+      "step": 27794
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360863473178542e-05,
+      "loss": 0.3174,
+      "step": 27795
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360805680359838e-05,
+      "loss": 0.2677,
+      "step": 27796
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360747885014607e-05,
+      "loss": 0.2971,
+      "step": 27797
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936069008714287e-05,
+      "loss": 0.3625,
+      "step": 27798
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360632286744637e-05,
+      "loss": 0.3037,
+      "step": 27799
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360574483819925e-05,
+      "loss": 0.3081,
+      "step": 27800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360516678368754e-05,
+      "loss": 0.3199,
+      "step": 27801
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936045887039113e-05,
+      "loss": 0.3026,
+      "step": 27802
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936040105988708e-05,
+      "loss": 0.2665,
+      "step": 27803
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936034324685661e-05,
+      "loss": 0.3356,
+      "step": 27804
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936028543129974e-05,
+      "loss": 0.2879,
+      "step": 27805
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360227613216487e-05,
+      "loss": 0.33,
+      "step": 27806
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360169792606865e-05,
+      "loss": 0.2684,
+      "step": 27807
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360111969470887e-05,
+      "loss": 0.3016,
+      "step": 27808
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9360054143808575e-05,
+      "loss": 0.3285,
+      "step": 27809
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359996315619936e-05,
+      "loss": 0.2811,
+      "step": 27810
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935993848490499e-05,
+      "loss": 0.3353,
+      "step": 27811
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359880651663757e-05,
+      "loss": 0.2916,
+      "step": 27812
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359822815896247e-05,
+      "loss": 0.3315,
+      "step": 27813
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935976497760248e-05,
+      "loss": 0.2738,
+      "step": 27814
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935970713678246e-05,
+      "loss": 0.4135,
+      "step": 27815
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935964929343622e-05,
+      "loss": 0.3003,
+      "step": 27816
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359591447563765e-05,
+      "loss": 0.3056,
+      "step": 27817
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935953359916511e-05,
+      "loss": 0.3065,
+      "step": 27818
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359475748240274e-05,
+      "loss": 0.2622,
+      "step": 27819
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935941789478927e-05,
+      "loss": 0.2835,
+      "step": 27820
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359360038812117e-05,
+      "loss": 0.3104,
+      "step": 27821
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359302180308833e-05,
+      "loss": 0.3361,
+      "step": 27822
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359244319279423e-05,
+      "loss": 0.2609,
+      "step": 27823
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359186455723913e-05,
+      "loss": 0.3162,
+      "step": 27824
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359128589642314e-05,
+      "loss": 0.3346,
+      "step": 27825
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359070721034642e-05,
+      "loss": 0.3071,
+      "step": 27826
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9359012849900914e-05,
+      "loss": 0.3311,
+      "step": 27827
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358954976241145e-05,
+      "loss": 0.2867,
+      "step": 27828
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935889710005535e-05,
+      "loss": 0.313,
+      "step": 27829
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358839221343545e-05,
+      "loss": 0.2627,
+      "step": 27830
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358781340105744e-05,
+      "loss": 0.2801,
+      "step": 27831
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358723456341963e-05,
+      "loss": 0.3479,
+      "step": 27832
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935866557005222e-05,
+      "loss": 0.322,
+      "step": 27833
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935860768123653e-05,
+      "loss": 0.3829,
+      "step": 27834
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935854978989491e-05,
+      "loss": 0.363,
+      "step": 27835
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358491896027372e-05,
+      "loss": 0.3256,
+      "step": 27836
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358433999633933e-05,
+      "loss": 0.3077,
+      "step": 27837
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358376100714608e-05,
+      "loss": 0.2929,
+      "step": 27838
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358318199269417e-05,
+      "loss": 0.334,
+      "step": 27839
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935826029529837e-05,
+      "loss": 0.2891,
+      "step": 27840
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358202388801485e-05,
+      "loss": 0.3072,
+      "step": 27841
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358144479778778e-05,
+      "loss": 0.3124,
+      "step": 27842
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358086568230262e-05,
+      "loss": 0.3169,
+      "step": 27843
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9358028654155956e-05,
+      "loss": 0.3299,
+      "step": 27844
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9357970737555875e-05,
+      "loss": 0.2665,
+      "step": 27845
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9357912818430033e-05,
+      "loss": 0.2657,
+      "step": 27846
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9357854896778447e-05,
+      "loss": 0.3429,
+      "step": 27847
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935779697260113e-05,
+      "loss": 0.3075,
+      "step": 27848
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9357739045898103e-05,
+      "loss": 0.3012,
+      "step": 27849
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9357681116669377e-05,
+      "loss": 0.274,
+      "step": 27850
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935762318491497e-05,
+      "loss": 0.2709,
+      "step": 27851
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9357565250634898e-05,
+      "loss": 0.3197,
+      "step": 27852
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935750731382917e-05,
+      "loss": 0.2871,
+      "step": 27853
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9357449374497812e-05,
+      "loss": 0.3332,
+      "step": 27854
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9357391432640832e-05,
+      "loss": 0.3115,
+      "step": 27855
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9357333488258253e-05,
+      "loss": 0.2759,
+      "step": 27856
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9357275541350083e-05,
+      "loss": 0.3349,
+      "step": 27857
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935721759191634e-05,
+      "loss": 0.3256,
+      "step": 27858
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935715963995704e-05,
+      "loss": 0.3206,
+      "step": 27859
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9357101685472198e-05,
+      "loss": 0.3086,
+      "step": 27860
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9357043728461832e-05,
+      "loss": 0.3531,
+      "step": 27861
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9356985768925956e-05,
+      "loss": 0.3641,
+      "step": 27862
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9356927806864586e-05,
+      "loss": 0.3112,
+      "step": 27863
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9356869842277738e-05,
+      "loss": 0.3691,
+      "step": 27864
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9356811875165426e-05,
+      "loss": 0.2861,
+      "step": 27865
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9356753905527666e-05,
+      "loss": 0.2991,
+      "step": 27866
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9356695933364477e-05,
+      "loss": 0.3053,
+      "step": 27867
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935663795867587e-05,
+      "loss": 0.275,
+      "step": 27868
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9356579981461863e-05,
+      "loss": 0.3259,
+      "step": 27869
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9356522001722472e-05,
+      "loss": 0.3192,
+      "step": 27870
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935646401945771e-05,
+      "loss": 0.3466,
+      "step": 27871
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9356406034667598e-05,
+      "loss": 0.2957,
+      "step": 27872
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9356348047352145e-05,
+      "loss": 0.3561,
+      "step": 27873
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935629005751137e-05,
+      "loss": 0.345,
+      "step": 27874
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935623206514529e-05,
+      "loss": 0.3152,
+      "step": 27875
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935617407025392e-05,
+      "loss": 0.3105,
+      "step": 27876
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935611607283727e-05,
+      "loss": 0.314,
+      "step": 27877
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935605807289537e-05,
+      "loss": 0.2862,
+      "step": 27878
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935600007042822e-05,
+      "loss": 0.2734,
+      "step": 27879
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355942065435843e-05,
+      "loss": 0.2718,
+      "step": 27880
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935588405791825e-05,
+      "loss": 0.2572,
+      "step": 27881
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355826047875464e-05,
+      "loss": 0.3055,
+      "step": 27882
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.93557680353075e-05,
+      "loss": 0.2979,
+      "step": 27883
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355710020214366e-05,
+      "loss": 0.3086,
+      "step": 27884
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355652002596082e-05,
+      "loss": 0.3353,
+      "step": 27885
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355593982452664e-05,
+      "loss": 0.2946,
+      "step": 27886
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355535959784126e-05,
+      "loss": 0.3469,
+      "step": 27887
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355477934590484e-05,
+      "loss": 0.2911,
+      "step": 27888
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355419906871762e-05,
+      "loss": 0.2708,
+      "step": 27889
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355361876627964e-05,
+      "loss": 0.3341,
+      "step": 27890
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355303843859107e-05,
+      "loss": 0.3265,
+      "step": 27891
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355245808565214e-05,
+      "loss": 0.3143,
+      "step": 27892
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.93551877707463e-05,
+      "loss": 0.2935,
+      "step": 27893
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9355129730402368e-05,
+      "loss": 0.3031,
+      "step": 27894
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935507168753345e-05,
+      "loss": 0.2886,
+      "step": 27895
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935501364213955e-05,
+      "loss": 0.3254,
+      "step": 27896
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935495559422069e-05,
+      "loss": 0.4127,
+      "step": 27897
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9354897543776885e-05,
+      "loss": 0.3042,
+      "step": 27898
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9354839490808145e-05,
+      "loss": 0.3258,
+      "step": 27899
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9354781435314495e-05,
+      "loss": 0.3264,
+      "step": 27900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935472337729594e-05,
+      "loss": 0.2626,
+      "step": 27901
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935466531675251e-05,
+      "loss": 0.34,
+      "step": 27902
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9354607253684205e-05,
+      "loss": 0.3229,
+      "step": 27903
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9354549188091052e-05,
+      "loss": 0.2834,
+      "step": 27904
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935449111997306e-05,
+      "loss": 0.3206,
+      "step": 27905
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935443304933025e-05,
+      "loss": 0.2967,
+      "step": 27906
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935437497616263e-05,
+      "loss": 0.3474,
+      "step": 27907
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9354316900470222e-05,
+      "loss": 0.3374,
+      "step": 27908
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9354258822253043e-05,
+      "loss": 0.3043,
+      "step": 27909
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9354200741511106e-05,
+      "loss": 0.3022,
+      "step": 27910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935414265824442e-05,
+      "loss": 0.3381,
+      "step": 27911
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9354084572453012e-05,
+      "loss": 0.3132,
+      "step": 27912
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9354026484136893e-05,
+      "loss": 0.3244,
+      "step": 27913
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935396839329608e-05,
+      "loss": 0.3501,
+      "step": 27914
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9353910299930584e-05,
+      "loss": 0.3395,
+      "step": 27915
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9353852204040425e-05,
+      "loss": 0.3415,
+      "step": 27916
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9353794105625616e-05,
+      "loss": 0.35,
+      "step": 27917
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935373600468618e-05,
+      "loss": 0.3148,
+      "step": 27918
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9353677901222122e-05,
+      "loss": 0.3075,
+      "step": 27919
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935361979523346e-05,
+      "loss": 0.3143,
+      "step": 27920
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9353561686720216e-05,
+      "loss": 0.3342,
+      "step": 27921
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.93535035756824e-05,
+      "loss": 0.2716,
+      "step": 27922
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9353445462120034e-05,
+      "loss": 0.3458,
+      "step": 27923
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9353387346033126e-05,
+      "loss": 0.3571,
+      "step": 27924
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9353329227421698e-05,
+      "loss": 0.3461,
+      "step": 27925
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935327110628576e-05,
+      "loss": 0.3104,
+      "step": 27926
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935321298262533e-05,
+      "loss": 0.2943,
+      "step": 27927
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9353154856440424e-05,
+      "loss": 0.3185,
+      "step": 27928
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935309672773106e-05,
+      "loss": 0.3068,
+      "step": 27929
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935303859649725e-05,
+      "loss": 0.2996,
+      "step": 27930
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352980462739013e-05,
+      "loss": 0.2799,
+      "step": 27931
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352922326456356e-05,
+      "loss": 0.3058,
+      "step": 27932
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352864187649308e-05,
+      "loss": 0.3323,
+      "step": 27933
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935280604631788e-05,
+      "loss": 0.2778,
+      "step": 27934
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352747902462082e-05,
+      "loss": 0.3106,
+      "step": 27935
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352689756081936e-05,
+      "loss": 0.2708,
+      "step": 27936
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352631607177455e-05,
+      "loss": 0.3285,
+      "step": 27937
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352573455748653e-05,
+      "loss": 0.3446,
+      "step": 27938
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352515301795548e-05,
+      "loss": 0.2875,
+      "step": 27939
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352457145318155e-05,
+      "loss": 0.2861,
+      "step": 27940
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352398986316493e-05,
+      "loss": 0.2683,
+      "step": 27941
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.935234082479057e-05,
+      "loss": 0.2516,
+      "step": 27942
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352282660740412e-05,
+      "loss": 0.356,
+      "step": 27943
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352224494166024e-05,
+      "loss": 0.3204,
+      "step": 27944
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935216632506743e-05,
+      "loss": 0.3773,
+      "step": 27945
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9352108153444642e-05,
+      "loss": 0.3596,
+      "step": 27946
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9352049979297675e-05,
+      "loss": 0.2917,
+      "step": 27947
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935199180262655e-05,
+      "loss": 0.323,
+      "step": 27948
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351933623431276e-05,
+      "loss": 0.2988,
+      "step": 27949
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935187544171187e-05,
+      "loss": 0.3234,
+      "step": 27950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351817257468353e-05,
+      "loss": 0.2715,
+      "step": 27951
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351759070700733e-05,
+      "loss": 0.3027,
+      "step": 27952
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935170088140903e-05,
+      "loss": 0.3872,
+      "step": 27953
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935164268959326e-05,
+      "loss": 0.2875,
+      "step": 27954
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351584495253438e-05,
+      "loss": 0.3317,
+      "step": 27955
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935152629838958e-05,
+      "loss": 0.2951,
+      "step": 27956
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351468099001704e-05,
+      "loss": 0.286,
+      "step": 27957
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351409897089817e-05,
+      "loss": 0.2701,
+      "step": 27958
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351351692653947e-05,
+      "loss": 0.2954,
+      "step": 27959
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.93512934856941e-05,
+      "loss": 0.2866,
+      "step": 27960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351235276210293e-05,
+      "loss": 0.2966,
+      "step": 27961
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351177064202545e-05,
+      "loss": 0.3048,
+      "step": 27962
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351118849670873e-05,
+      "loss": 0.3268,
+      "step": 27963
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9351060632615287e-05,
+      "loss": 0.2941,
+      "step": 27964
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935100241303581e-05,
+      "loss": 0.3195,
+      "step": 27965
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350944190932453e-05,
+      "loss": 0.3011,
+      "step": 27966
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935088596630523e-05,
+      "loss": 0.3251,
+      "step": 27967
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350827739154162e-05,
+      "loss": 0.2743,
+      "step": 27968
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935076950947926e-05,
+      "loss": 0.3142,
+      "step": 27969
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350711277280543e-05,
+      "loss": 0.3009,
+      "step": 27970
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350653042558024e-05,
+      "loss": 0.3064,
+      "step": 27971
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935059480531172e-05,
+      "loss": 0.3406,
+      "step": 27972
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350536565541648e-05,
+      "loss": 0.3065,
+      "step": 27973
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935047832324782e-05,
+      "loss": 0.3199,
+      "step": 27974
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350420078430255e-05,
+      "loss": 0.3088,
+      "step": 27975
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350361831088972e-05,
+      "loss": 0.3288,
+      "step": 27976
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350303581223978e-05,
+      "loss": 0.3093,
+      "step": 27977
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350245328835293e-05,
+      "loss": 0.3172,
+      "step": 27978
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350187073922938e-05,
+      "loss": 0.3097,
+      "step": 27979
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935012881648692e-05,
+      "loss": 0.2797,
+      "step": 27980
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9350070556527258e-05,
+      "loss": 0.294,
+      "step": 27981
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.935001229404397e-05,
+      "loss": 0.2793,
+      "step": 27982
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934995402903707e-05,
+      "loss": 0.284,
+      "step": 27983
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349895761506572e-05,
+      "loss": 0.2705,
+      "step": 27984
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349837491452494e-05,
+      "loss": 0.3154,
+      "step": 27985
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934977921887485e-05,
+      "loss": 0.329,
+      "step": 27986
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934972094377366e-05,
+      "loss": 0.3079,
+      "step": 27987
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349662666148938e-05,
+      "loss": 0.2949,
+      "step": 27988
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349604386000693e-05,
+      "loss": 0.3257,
+      "step": 27989
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934954610332895e-05,
+      "loss": 0.3588,
+      "step": 27990
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349487818133717e-05,
+      "loss": 0.3063,
+      "step": 27991
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349429530415017e-05,
+      "loss": 0.3294,
+      "step": 27992
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934937124017286e-05,
+      "loss": 0.3246,
+      "step": 27993
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349312947407267e-05,
+      "loss": 0.2634,
+      "step": 27994
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934925465211825e-05,
+      "loss": 0.3181,
+      "step": 27995
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349196354305825e-05,
+      "loss": 0.3167,
+      "step": 27996
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349138053970007e-05,
+      "loss": 0.3018,
+      "step": 27997
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349079751110813e-05,
+      "loss": 0.3336,
+      "step": 27998
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9349021445728257e-05,
+      "loss": 0.3765,
+      "step": 27999
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934896313782236e-05,
+      "loss": 0.2989,
+      "step": 28000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348904827393134e-05,
+      "loss": 0.3368,
+      "step": 28001
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934884651444059e-05,
+      "loss": 0.3218,
+      "step": 28002
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348788198964753e-05,
+      "loss": 0.3478,
+      "step": 28003
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348729880965632e-05,
+      "loss": 0.3085,
+      "step": 28004
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934867156044325e-05,
+      "loss": 0.3583,
+      "step": 28005
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348613237397614e-05,
+      "loss": 0.3068,
+      "step": 28006
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348554911828742e-05,
+      "loss": 0.2755,
+      "step": 28007
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348496583736657e-05,
+      "loss": 0.2991,
+      "step": 28008
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348438253121363e-05,
+      "loss": 0.289,
+      "step": 28009
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348379919982886e-05,
+      "loss": 0.2589,
+      "step": 28010
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348321584321237e-05,
+      "loss": 0.2996,
+      "step": 28011
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934826324613643e-05,
+      "loss": 0.3022,
+      "step": 28012
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348204905428486e-05,
+      "loss": 0.2534,
+      "step": 28013
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348146562197415e-05,
+      "loss": 0.333,
+      "step": 28014
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348088216443236e-05,
+      "loss": 0.2613,
+      "step": 28015
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9348029868165966e-05,
+      "loss": 0.3149,
+      "step": 28016
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934797151736562e-05,
+      "loss": 0.318,
+      "step": 28017
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9347913164042208e-05,
+      "loss": 0.3316,
+      "step": 28018
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934785480819575e-05,
+      "loss": 0.3473,
+      "step": 28019
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934779644982627e-05,
+      "loss": 0.3374,
+      "step": 28020
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934773808893377e-05,
+      "loss": 0.3057,
+      "step": 28021
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9347679725518275e-05,
+      "loss": 0.3267,
+      "step": 28022
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9347621359579792e-05,
+      "loss": 0.3559,
+      "step": 28023
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9347562991118348e-05,
+      "loss": 0.3323,
+      "step": 28024
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934750462013395e-05,
+      "loss": 0.3237,
+      "step": 28025
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934744624662662e-05,
+      "loss": 0.3119,
+      "step": 28026
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934738787059637e-05,
+      "loss": 0.3064,
+      "step": 28027
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9347329492043214e-05,
+      "loss": 0.288,
+      "step": 28028
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934727111096717e-05,
+      "loss": 0.3226,
+      "step": 28029
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9347212727368255e-05,
+      "loss": 0.323,
+      "step": 28030
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9347154341246485e-05,
+      "loss": 0.3009,
+      "step": 28031
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934709595260187e-05,
+      "loss": 0.2969,
+      "step": 28032
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9347037561434435e-05,
+      "loss": 0.2979,
+      "step": 28033
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9346979167744188e-05,
+      "loss": 0.2681,
+      "step": 28034
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934692077153115e-05,
+      "loss": 0.3299,
+      "step": 28035
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9346862372795333e-05,
+      "loss": 0.3313,
+      "step": 28036
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9346803971536752e-05,
+      "loss": 0.2886,
+      "step": 28037
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.934674556775543e-05,
+      "loss": 0.3418,
+      "step": 28038
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9346687161451375e-05,
+      "loss": 0.3026,
+      "step": 28039
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9346628752624602e-05,
+      "loss": 0.3334,
+      "step": 28040
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9346570341275134e-05,
+      "loss": 0.3239,
+      "step": 28041
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9346511927402983e-05,
+      "loss": 0.2928,
+      "step": 28042
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9346453511008163e-05,
+      "loss": 0.317,
+      "step": 28043
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9346395092090694e-05,
+      "loss": 0.2909,
+      "step": 28044
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9346336670650587e-05,
+      "loss": 0.3018,
+      "step": 28045
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934627824668786e-05,
+      "loss": 0.3228,
+      "step": 28046
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934621982020253e-05,
+      "loss": 0.2859,
+      "step": 28047
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9346161391194612e-05,
+      "loss": 0.3086,
+      "step": 28048
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934610295966412e-05,
+      "loss": 0.2965,
+      "step": 28049
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934604452561107e-05,
+      "loss": 0.2954,
+      "step": 28050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934598608903548e-05,
+      "loss": 0.2953,
+      "step": 28051
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345927649937367e-05,
+      "loss": 0.2657,
+      "step": 28052
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345869208316744e-05,
+      "loss": 0.3171,
+      "step": 28053
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345810764173624e-05,
+      "loss": 0.24,
+      "step": 28054
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934575231750803e-05,
+      "loss": 0.3125,
+      "step": 28055
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934569386831997e-05,
+      "loss": 0.3184,
+      "step": 28056
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345635416609464e-05,
+      "loss": 0.3256,
+      "step": 28057
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345576962376528e-05,
+      "loss": 0.3287,
+      "step": 28058
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934551850562118e-05,
+      "loss": 0.3257,
+      "step": 28059
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934546004634343e-05,
+      "loss": 0.3584,
+      "step": 28060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345401584543297e-05,
+      "loss": 0.3068,
+      "step": 28061
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345343120220796e-05,
+      "loss": 0.2934,
+      "step": 28062
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345284653375946e-05,
+      "loss": 0.3543,
+      "step": 28063
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345226184008757e-05,
+      "loss": 0.3257,
+      "step": 28064
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345167712119246e-05,
+      "loss": 0.3036,
+      "step": 28065
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345109237707437e-05,
+      "loss": 0.3133,
+      "step": 28066
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9345050760773333e-05,
+      "loss": 0.3385,
+      "step": 28067
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934499228131696e-05,
+      "loss": 0.3918,
+      "step": 28068
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344933799338327e-05,
+      "loss": 0.3663,
+      "step": 28069
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344875314837454e-05,
+      "loss": 0.3117,
+      "step": 28070
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344816827814354e-05,
+      "loss": 0.3372,
+      "step": 28071
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344758338269048e-05,
+      "loss": 0.2836,
+      "step": 28072
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344699846201546e-05,
+      "loss": 0.3091,
+      "step": 28073
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344641351611867e-05,
+      "loss": 0.2791,
+      "step": 28074
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344582854500022e-05,
+      "loss": 0.2844,
+      "step": 28075
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344524354866032e-05,
+      "loss": 0.3146,
+      "step": 28076
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344465852709913e-05,
+      "loss": 0.3102,
+      "step": 28077
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934440734803168e-05,
+      "loss": 0.2742,
+      "step": 28078
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344348840831343e-05,
+      "loss": 0.2937,
+      "step": 28079
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344290331108926e-05,
+      "loss": 0.2896,
+      "step": 28080
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934423181886444e-05,
+      "loss": 0.3706,
+      "step": 28081
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344173304097903e-05,
+      "loss": 0.309,
+      "step": 28082
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934411478680933e-05,
+      "loss": 0.3037,
+      "step": 28083
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9344056266998734e-05,
+      "loss": 0.3408,
+      "step": 28084
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343997744666137e-05,
+      "loss": 0.2614,
+      "step": 28085
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934393921981155e-05,
+      "loss": 0.2858,
+      "step": 28086
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343880692434992e-05,
+      "loss": 0.2792,
+      "step": 28087
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343822162536474e-05,
+      "loss": 0.2793,
+      "step": 28088
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343763630116017e-05,
+      "loss": 0.3196,
+      "step": 28089
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343705095173634e-05,
+      "loss": 0.3077,
+      "step": 28090
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934364655770934e-05,
+      "loss": 0.3181,
+      "step": 28091
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343588017723154e-05,
+      "loss": 0.3581,
+      "step": 28092
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343529475215087e-05,
+      "loss": 0.3215,
+      "step": 28093
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934347093018516e-05,
+      "loss": 0.2992,
+      "step": 28094
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934341238263339e-05,
+      "loss": 0.2929,
+      "step": 28095
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934335383255978e-05,
+      "loss": 0.3814,
+      "step": 28096
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343295279964363e-05,
+      "loss": 0.2671,
+      "step": 28097
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343236724847147e-05,
+      "loss": 0.3017,
+      "step": 28098
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343178167208144e-05,
+      "loss": 0.3366,
+      "step": 28099
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343119607047375e-05,
+      "loss": 0.2961,
+      "step": 28100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9343061044364853e-05,
+      "loss": 0.3404,
+      "step": 28101
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.93430024791606e-05,
+      "loss": 0.3494,
+      "step": 28102
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934294391143462e-05,
+      "loss": 0.2772,
+      "step": 28103
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9342885341186938e-05,
+      "loss": 0.3246,
+      "step": 28104
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934282676841757e-05,
+      "loss": 0.3131,
+      "step": 28105
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9342768193126527e-05,
+      "loss": 0.2938,
+      "step": 28106
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934270961531383e-05,
+      "loss": 0.3246,
+      "step": 28107
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934265103497949e-05,
+      "loss": 0.2817,
+      "step": 28108
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9342592452123522e-05,
+      "loss": 0.296,
+      "step": 28109
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934253386674595e-05,
+      "loss": 0.347,
+      "step": 28110
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934247527884678e-05,
+      "loss": 0.3073,
+      "step": 28111
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9342416688426035e-05,
+      "loss": 0.3016,
+      "step": 28112
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9342358095483725e-05,
+      "loss": 0.3063,
+      "step": 28113
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9342299500019873e-05,
+      "loss": 0.3375,
+      "step": 28114
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934224090203449e-05,
+      "loss": 0.3408,
+      "step": 28115
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934218230152759e-05,
+      "loss": 0.3111,
+      "step": 28116
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9342123698499193e-05,
+      "loss": 0.3305,
+      "step": 28117
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9342065092949313e-05,
+      "loss": 0.3318,
+      "step": 28118
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9342006484877968e-05,
+      "loss": 0.3498,
+      "step": 28119
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934194787428517e-05,
+      "loss": 0.3369,
+      "step": 28120
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341889261170936e-05,
+      "loss": 0.3535,
+      "step": 28121
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341830645535282e-05,
+      "loss": 0.2787,
+      "step": 28122
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341772027378224e-05,
+      "loss": 0.3653,
+      "step": 28123
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934171340669978e-05,
+      "loss": 0.313,
+      "step": 28124
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341654783499963e-05,
+      "loss": 0.3223,
+      "step": 28125
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934159615777879e-05,
+      "loss": 0.3157,
+      "step": 28126
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341537529536273e-05,
+      "loss": 0.3588,
+      "step": 28127
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341478898772438e-05,
+      "loss": 0.2649,
+      "step": 28128
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934142026548729e-05,
+      "loss": 0.3138,
+      "step": 28129
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934136162968085e-05,
+      "loss": 0.312,
+      "step": 28130
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341302991353134e-05,
+      "loss": 0.3208,
+      "step": 28131
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341244350504152e-05,
+      "loss": 0.3189,
+      "step": 28132
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934118570713393e-05,
+      "loss": 0.4121,
+      "step": 28133
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341127061242475e-05,
+      "loss": 0.3211,
+      "step": 28134
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341068412829807e-05,
+      "loss": 0.2678,
+      "step": 28135
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9341009761895942e-05,
+      "loss": 0.287,
+      "step": 28136
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9340951108440894e-05,
+      "loss": 0.3033,
+      "step": 28137
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.934089245246468e-05,
+      "loss": 0.3128,
+      "step": 28138
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9340833793967312e-05,
+      "loss": 0.3265,
+      "step": 28139
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9340775132948812e-05,
+      "loss": 0.2676,
+      "step": 28140
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9340716469409196e-05,
+      "loss": 0.3238,
+      "step": 28141
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.934065780334847e-05,
+      "loss": 0.2765,
+      "step": 28142
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9340599134766665e-05,
+      "loss": 0.293,
+      "step": 28143
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9340540463663784e-05,
+      "loss": 0.3436,
+      "step": 28144
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9340481790039848e-05,
+      "loss": 0.2567,
+      "step": 28145
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.934042311389487e-05,
+      "loss": 0.3586,
+      "step": 28146
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9340364435228873e-05,
+      "loss": 0.3049,
+      "step": 28147
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9340305754041868e-05,
+      "loss": 0.3044,
+      "step": 28148
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9340247070333865e-05,
+      "loss": 0.2805,
+      "step": 28149
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.934018838410489e-05,
+      "loss": 0.3154,
+      "step": 28150
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.934012969535495e-05,
+      "loss": 0.3316,
+      "step": 28151
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9340071004084073e-05,
+      "loss": 0.3726,
+      "step": 28152
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9340012310292266e-05,
+      "loss": 0.3352,
+      "step": 28153
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933995361397954e-05,
+      "loss": 0.2832,
+      "step": 28154
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933989491514592e-05,
+      "loss": 0.3214,
+      "step": 28155
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933983621379142e-05,
+      "loss": 0.2979,
+      "step": 28156
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933977750991605e-05,
+      "loss": 0.3186,
+      "step": 28157
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9339718803519835e-05,
+      "loss": 0.3522,
+      "step": 28158
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9339660094602787e-05,
+      "loss": 0.3,
+      "step": 28159
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9339601383164918e-05,
+      "loss": 0.2884,
+      "step": 28160
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933954266920625e-05,
+      "loss": 0.3386,
+      "step": 28161
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9339483952726793e-05,
+      "loss": 0.3441,
+      "step": 28162
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9339425233726566e-05,
+      "loss": 0.3288,
+      "step": 28163
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933936651220558e-05,
+      "loss": 0.3014,
+      "step": 28164
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9339307788163865e-05,
+      "loss": 0.3357,
+      "step": 28165
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933924906160142e-05,
+      "loss": 0.3611,
+      "step": 28166
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9339190332518273e-05,
+      "loss": 0.3093,
+      "step": 28167
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933913160091443e-05,
+      "loss": 0.3603,
+      "step": 28168
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9339072866789915e-05,
+      "loss": 0.3367,
+      "step": 28169
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9339014130144736e-05,
+      "loss": 0.3389,
+      "step": 28170
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933895539097892e-05,
+      "loss": 0.2832,
+      "step": 28171
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933889664929247e-05,
+      "loss": 0.3058,
+      "step": 28172
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338837905085414e-05,
+      "loss": 0.3682,
+      "step": 28173
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338779158357757e-05,
+      "loss": 0.3089,
+      "step": 28174
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338720409109524e-05,
+      "loss": 0.3346,
+      "step": 28175
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338661657340723e-05,
+      "loss": 0.3197,
+      "step": 28176
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338602903051377e-05,
+      "loss": 0.291,
+      "step": 28177
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338544146241497e-05,
+      "loss": 0.3453,
+      "step": 28178
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338485386911098e-05,
+      "loss": 0.4194,
+      "step": 28179
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338426625060203e-05,
+      "loss": 0.3432,
+      "step": 28180
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933836786068882e-05,
+      "loss": 0.3351,
+      "step": 28181
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338309093796968e-05,
+      "loss": 0.335,
+      "step": 28182
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933825032438466e-05,
+      "loss": 0.2666,
+      "step": 28183
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338191552451923e-05,
+      "loss": 0.2565,
+      "step": 28184
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338132777998754e-05,
+      "loss": 0.3601,
+      "step": 28185
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338074001025187e-05,
+      "loss": 0.3279,
+      "step": 28186
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9338015221531226e-05,
+      "loss": 0.3375,
+      "step": 28187
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9337956439516895e-05,
+      "loss": 0.3056,
+      "step": 28188
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.93378976549822e-05,
+      "loss": 0.3104,
+      "step": 28189
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9337838867927167e-05,
+      "loss": 0.3214,
+      "step": 28190
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9337780078351806e-05,
+      "loss": 0.3353,
+      "step": 28191
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9337721286256137e-05,
+      "loss": 0.3139,
+      "step": 28192
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933766249164017e-05,
+      "loss": 0.3089,
+      "step": 28193
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9337603694503927e-05,
+      "loss": 0.3239,
+      "step": 28194
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933754489484742e-05,
+      "loss": 0.3545,
+      "step": 28195
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9337486092670667e-05,
+      "loss": 0.2857,
+      "step": 28196
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933742728797368e-05,
+      "loss": 0.3453,
+      "step": 28197
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933736848075648e-05,
+      "loss": 0.3271,
+      "step": 28198
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933730967101908e-05,
+      "loss": 0.3282,
+      "step": 28199
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.93372508587615e-05,
+      "loss": 0.2983,
+      "step": 28200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9337192043983744e-05,
+      "loss": 0.3009,
+      "step": 28201
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9337133226685843e-05,
+      "loss": 0.2737,
+      "step": 28202
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9337074406867804e-05,
+      "loss": 0.2806,
+      "step": 28203
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9337015584529647e-05,
+      "loss": 0.3386,
+      "step": 28204
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9336956759671383e-05,
+      "loss": 0.3524,
+      "step": 28205
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933689793229303e-05,
+      "loss": 0.2717,
+      "step": 28206
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9336839102394605e-05,
+      "loss": 0.2991,
+      "step": 28207
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933678026997613e-05,
+      "loss": 0.3479,
+      "step": 28208
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9336721435037607e-05,
+      "loss": 0.3446,
+      "step": 28209
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933666259757906e-05,
+      "loss": 0.3015,
+      "step": 28210
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9336603757600507e-05,
+      "loss": 0.3874,
+      "step": 28211
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933654491510196e-05,
+      "loss": 0.2865,
+      "step": 28212
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9336486070083437e-05,
+      "loss": 0.3294,
+      "step": 28213
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9336427222544948e-05,
+      "loss": 0.2973,
+      "step": 28214
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933636837248652e-05,
+      "loss": 0.2556,
+      "step": 28215
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9336309519908157e-05,
+      "loss": 0.3099,
+      "step": 28216
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9336250664809885e-05,
+      "loss": 0.3607,
+      "step": 28217
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933619180719171e-05,
+      "loss": 0.2731,
+      "step": 28218
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9336132947053657e-05,
+      "loss": 0.3295,
+      "step": 28219
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933607408439574e-05,
+      "loss": 0.2842,
+      "step": 28220
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9336015219217968e-05,
+      "loss": 0.273,
+      "step": 28221
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335956351520363e-05,
+      "loss": 0.2719,
+      "step": 28222
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335897481302942e-05,
+      "loss": 0.3322,
+      "step": 28223
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933583860856572e-05,
+      "loss": 0.2686,
+      "step": 28224
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933577973330871e-05,
+      "loss": 0.3062,
+      "step": 28225
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335720855531927e-05,
+      "loss": 0.3059,
+      "step": 28226
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933566197523539e-05,
+      "loss": 0.2559,
+      "step": 28227
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335603092419114e-05,
+      "loss": 0.3221,
+      "step": 28228
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335544207083118e-05,
+      "loss": 0.2915,
+      "step": 28229
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335485319227413e-05,
+      "loss": 0.3583,
+      "step": 28230
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335426428852018e-05,
+      "loss": 0.3535,
+      "step": 28231
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335367535956947e-05,
+      "loss": 0.3095,
+      "step": 28232
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335308640542215e-05,
+      "loss": 0.2924,
+      "step": 28233
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933524974260784e-05,
+      "loss": 0.3335,
+      "step": 28234
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335190842153838e-05,
+      "loss": 0.3252,
+      "step": 28235
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335131939180227e-05,
+      "loss": 0.2988,
+      "step": 28236
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.933507303368702e-05,
+      "loss": 0.3769,
+      "step": 28237
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9335014125674226e-05,
+      "loss": 0.3077,
+      "step": 28238
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9334955215141877e-05,
+      "loss": 0.3725,
+      "step": 28239
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9334896302089973e-05,
+      "loss": 0.2963,
+      "step": 28240
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9334837386518543e-05,
+      "loss": 0.3012,
+      "step": 28241
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9334778468427593e-05,
+      "loss": 0.3098,
+      "step": 28242
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933471954781714e-05,
+      "loss": 0.3584,
+      "step": 28243
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933466062468721e-05,
+      "loss": 0.3176,
+      "step": 28244
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9334601699037808e-05,
+      "loss": 0.2826,
+      "step": 28245
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933454277086895e-05,
+      "loss": 0.3176,
+      "step": 28246
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933448384018066e-05,
+      "loss": 0.2446,
+      "step": 28247
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9334424906972947e-05,
+      "loss": 0.3097,
+      "step": 28248
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933436597124583e-05,
+      "loss": 0.325,
+      "step": 28249
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933430703299932e-05,
+      "loss": 0.3247,
+      "step": 28250
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9334248092233445e-05,
+      "loss": 0.2935,
+      "step": 28251
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9334189148948205e-05,
+      "loss": 0.3301,
+      "step": 28252
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9334130203143628e-05,
+      "loss": 0.2898,
+      "step": 28253
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9334071254819724e-05,
+      "loss": 0.3377,
+      "step": 28254
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9334012303976514e-05,
+      "loss": 0.2842,
+      "step": 28255
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9333953350614004e-05,
+      "loss": 0.3322,
+      "step": 28256
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933389439473222e-05,
+      "loss": 0.3269,
+      "step": 28257
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9333835436331174e-05,
+      "loss": 0.3192,
+      "step": 28258
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933377647541088e-05,
+      "loss": 0.3407,
+      "step": 28259
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9333717511971363e-05,
+      "loss": 0.3037,
+      "step": 28260
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9333658546012626e-05,
+      "loss": 0.3211,
+      "step": 28261
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933359957753469e-05,
+      "loss": 0.3264,
+      "step": 28262
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9333540606537576e-05,
+      "loss": 0.3138,
+      "step": 28263
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9333481633021293e-05,
+      "loss": 0.327,
+      "step": 28264
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933342265698586e-05,
+      "loss": 0.3159,
+      "step": 28265
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9333363678431296e-05,
+      "loss": 0.3025,
+      "step": 28266
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933330469735761e-05,
+      "loss": 0.3034,
+      "step": 28267
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933324571376482e-05,
+      "loss": 0.3299,
+      "step": 28268
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9333186727652945e-05,
+      "loss": 0.2702,
+      "step": 28269
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9333127739022003e-05,
+      "loss": 0.3284,
+      "step": 28270
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9333068747872e-05,
+      "loss": 0.3129,
+      "step": 28271
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933300975420296e-05,
+      "loss": 0.324,
+      "step": 28272
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.93329507580149e-05,
+      "loss": 0.3059,
+      "step": 28273
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332891759307833e-05,
+      "loss": 0.3569,
+      "step": 28274
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332832758081772e-05,
+      "loss": 0.2787,
+      "step": 28275
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332773754336736e-05,
+      "loss": 0.3843,
+      "step": 28276
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933271474807274e-05,
+      "loss": 0.3558,
+      "step": 28277
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332655739289806e-05,
+      "loss": 0.3175,
+      "step": 28278
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933259672798794e-05,
+      "loss": 0.3092,
+      "step": 28279
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332537714167162e-05,
+      "loss": 0.292,
+      "step": 28280
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332478697827492e-05,
+      "loss": 0.3395,
+      "step": 28281
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332419678968936e-05,
+      "loss": 0.3853,
+      "step": 28282
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332360657591523e-05,
+      "loss": 0.3508,
+      "step": 28283
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933230163369526e-05,
+      "loss": 0.2981,
+      "step": 28284
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332242607280162e-05,
+      "loss": 0.2709,
+      "step": 28285
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332183578346254e-05,
+      "loss": 0.2955,
+      "step": 28286
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933212454689354e-05,
+      "loss": 0.3046,
+      "step": 28287
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9332065512922046e-05,
+      "loss": 0.3647,
+      "step": 28288
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933200647643178e-05,
+      "loss": 0.2942,
+      "step": 28289
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9331947437422768e-05,
+      "loss": 0.273,
+      "step": 28290
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9331888395895015e-05,
+      "loss": 0.2898,
+      "step": 28291
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9331829351848544e-05,
+      "loss": 0.3032,
+      "step": 28292
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9331770305283368e-05,
+      "loss": 0.2902,
+      "step": 28293
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9331711256199504e-05,
+      "loss": 0.3261,
+      "step": 28294
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9331652204596968e-05,
+      "loss": 0.3248,
+      "step": 28295
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9331593150475773e-05,
+      "loss": 0.3776,
+      "step": 28296
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933153409383594e-05,
+      "loss": 0.3206,
+      "step": 28297
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933147503467748e-05,
+      "loss": 0.3063,
+      "step": 28298
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933141597300041e-05,
+      "loss": 0.3045,
+      "step": 28299
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933135690880475e-05,
+      "loss": 0.3789,
+      "step": 28300
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9331297842090514e-05,
+      "loss": 0.3396,
+      "step": 28301
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9331238772857716e-05,
+      "loss": 0.3563,
+      "step": 28302
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933117970110637e-05,
+      "loss": 0.2931,
+      "step": 28303
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.93311206268365e-05,
+      "loss": 0.2868,
+      "step": 28304
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9331061550048114e-05,
+      "loss": 0.3017,
+      "step": 28305
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933100247074123e-05,
+      "loss": 0.3219,
+      "step": 28306
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330943388915865e-05,
+      "loss": 0.2951,
+      "step": 28307
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933088430457204e-05,
+      "loss": 0.2851,
+      "step": 28308
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933082521770976e-05,
+      "loss": 0.3652,
+      "step": 28309
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330766128329047e-05,
+      "loss": 0.3552,
+      "step": 28310
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330707036429918e-05,
+      "loss": 0.2747,
+      "step": 28311
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933064794201239e-05,
+      "loss": 0.3358,
+      "step": 28312
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330588845076472e-05,
+      "loss": 0.2946,
+      "step": 28313
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330529745622185e-05,
+      "loss": 0.3156,
+      "step": 28314
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330470643649547e-05,
+      "loss": 0.29,
+      "step": 28315
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330411539158572e-05,
+      "loss": 0.2998,
+      "step": 28316
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330352432149273e-05,
+      "loss": 0.2938,
+      "step": 28317
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933029332262167e-05,
+      "loss": 0.2805,
+      "step": 28318
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330234210575775e-05,
+      "loss": 0.2987,
+      "step": 28319
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330175096011606e-05,
+      "loss": 0.3556,
+      "step": 28320
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.933011597892918e-05,
+      "loss": 0.2886,
+      "step": 28321
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9330056859328514e-05,
+      "loss": 0.3972,
+      "step": 28322
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.932999773720962e-05,
+      "loss": 0.3238,
+      "step": 28323
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329938612572516e-05,
+      "loss": 0.2959,
+      "step": 28324
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.932987948541722e-05,
+      "loss": 0.2825,
+      "step": 28325
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329820355743745e-05,
+      "loss": 0.3562,
+      "step": 28326
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329761223552106e-05,
+      "loss": 0.3339,
+      "step": 28327
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329702088842322e-05,
+      "loss": 0.3094,
+      "step": 28328
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329642951614408e-05,
+      "loss": 0.2723,
+      "step": 28329
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.932958381186838e-05,
+      "loss": 0.3008,
+      "step": 28330
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329524669604255e-05,
+      "loss": 0.2942,
+      "step": 28331
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329465524822044e-05,
+      "loss": 0.2734,
+      "step": 28332
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.932940637752177e-05,
+      "loss": 0.3369,
+      "step": 28333
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329347227703446e-05,
+      "loss": 0.2868,
+      "step": 28334
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329288075367083e-05,
+      "loss": 0.3178,
+      "step": 28335
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329228920512704e-05,
+      "loss": 0.3482,
+      "step": 28336
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329169763140324e-05,
+      "loss": 0.2826,
+      "step": 28337
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9329110603249955e-05,
+      "loss": 0.377,
+      "step": 28338
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9329051440841618e-05,
+      "loss": 0.3135,
+      "step": 28339
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9328992275915327e-05,
+      "loss": 0.2778,
+      "step": 28340
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9328933108471094e-05,
+      "loss": 0.3434,
+      "step": 28341
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9328873938508938e-05,
+      "loss": 0.2647,
+      "step": 28342
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932881476602888e-05,
+      "loss": 0.2595,
+      "step": 28343
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9328755591030928e-05,
+      "loss": 0.2582,
+      "step": 28344
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.93286964135151e-05,
+      "loss": 0.3238,
+      "step": 28345
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9328637233481418e-05,
+      "loss": 0.3302,
+      "step": 28346
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932857805092989e-05,
+      "loss": 0.2901,
+      "step": 28347
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9328518865860534e-05,
+      "loss": 0.2946,
+      "step": 28348
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932845967827337e-05,
+      "loss": 0.3484,
+      "step": 28349
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9328400488168407e-05,
+      "loss": 0.3509,
+      "step": 28350
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932834129554567e-05,
+      "loss": 0.3501,
+      "step": 28351
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9328282100405168e-05,
+      "loss": 0.2246,
+      "step": 28352
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9328222902746918e-05,
+      "loss": 0.2987,
+      "step": 28353
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932816370257094e-05,
+      "loss": 0.3233,
+      "step": 28354
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9328104499877246e-05,
+      "loss": 0.3242,
+      "step": 28355
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932804529466585e-05,
+      "loss": 0.3061,
+      "step": 28356
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9327986086936772e-05,
+      "loss": 0.3274,
+      "step": 28357
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932792687669003e-05,
+      "loss": 0.3309,
+      "step": 28358
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9327867663925635e-05,
+      "loss": 0.3176,
+      "step": 28359
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.93278084486436e-05,
+      "loss": 0.3377,
+      "step": 28360
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9327749230843955e-05,
+      "loss": 0.3223,
+      "step": 28361
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.93276900105267e-05,
+      "loss": 0.2953,
+      "step": 28362
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932763078769186e-05,
+      "loss": 0.3519,
+      "step": 28363
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932757156233945e-05,
+      "loss": 0.3012,
+      "step": 28364
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9327512334469486e-05,
+      "loss": 0.3328,
+      "step": 28365
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932745310408198e-05,
+      "loss": 0.3792,
+      "step": 28366
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9327393871176952e-05,
+      "loss": 0.3034,
+      "step": 28367
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9327334635754416e-05,
+      "loss": 0.2785,
+      "step": 28368
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932727539781439e-05,
+      "loss": 0.3258,
+      "step": 28369
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9327216157356886e-05,
+      "loss": 0.3197,
+      "step": 28370
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9327156914381926e-05,
+      "loss": 0.3202,
+      "step": 28371
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932709766888952e-05,
+      "loss": 0.3046,
+      "step": 28372
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9327038420879688e-05,
+      "loss": 0.2698,
+      "step": 28373
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9326979170352446e-05,
+      "loss": 0.3847,
+      "step": 28374
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932691991730781e-05,
+      "loss": 0.3182,
+      "step": 28375
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932686066174579e-05,
+      "loss": 0.3362,
+      "step": 28376
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932680140366641e-05,
+      "loss": 0.3132,
+      "step": 28377
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9326742143069683e-05,
+      "loss": 0.2658,
+      "step": 28378
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9326682879955624e-05,
+      "loss": 0.2856,
+      "step": 28379
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9326623614324247e-05,
+      "loss": 0.3408,
+      "step": 28380
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9326564346175573e-05,
+      "loss": 0.3271,
+      "step": 28381
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9326505075509615e-05,
+      "loss": 0.307,
+      "step": 28382
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932644580232639e-05,
+      "loss": 0.3213,
+      "step": 28383
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9326386526625916e-05,
+      "loss": 0.3061,
+      "step": 28384
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9326327248408203e-05,
+      "loss": 0.2871,
+      "step": 28385
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9326267967673273e-05,
+      "loss": 0.2902,
+      "step": 28386
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932620868442114e-05,
+      "loss": 0.2519,
+      "step": 28387
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932614939865182e-05,
+      "loss": 0.283,
+      "step": 28388
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9326090110365324e-05,
+      "loss": 0.3215,
+      "step": 28389
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932603081956168e-05,
+      "loss": 0.4151,
+      "step": 28390
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932597152624089e-05,
+      "loss": 0.3005,
+      "step": 28391
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932591223040298e-05,
+      "loss": 0.3364,
+      "step": 28392
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9325852932047963e-05,
+      "loss": 0.2789,
+      "step": 28393
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932579363117585e-05,
+      "loss": 0.2823,
+      "step": 28394
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932573432778667e-05,
+      "loss": 0.3095,
+      "step": 28395
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9325675021880425e-05,
+      "loss": 0.2934,
+      "step": 28396
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9325615713457137e-05,
+      "loss": 0.31,
+      "step": 28397
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9325556402516822e-05,
+      "loss": 0.2848,
+      "step": 28398
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.93254970890595e-05,
+      "loss": 0.375,
+      "step": 28399
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9325437773085176e-05,
+      "loss": 0.2871,
+      "step": 28400
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9325378454593878e-05,
+      "loss": 0.2653,
+      "step": 28401
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9325319133585615e-05,
+      "loss": 0.329,
+      "step": 28402
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9325259810060404e-05,
+      "loss": 0.2749,
+      "step": 28403
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932520048401826e-05,
+      "loss": 0.3813,
+      "step": 28404
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9325141155459204e-05,
+      "loss": 0.3374,
+      "step": 28405
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932508182438325e-05,
+      "loss": 0.3553,
+      "step": 28406
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932502249079041e-05,
+      "loss": 0.3445,
+      "step": 28407
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324963154680706e-05,
+      "loss": 0.3371,
+      "step": 28408
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324903816054147e-05,
+      "loss": 0.3295,
+      "step": 28409
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324844474910752e-05,
+      "loss": 0.3132,
+      "step": 28410
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932478513125054e-05,
+      "loss": 0.367,
+      "step": 28411
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324725785073527e-05,
+      "loss": 0.3307,
+      "step": 28412
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324666436379724e-05,
+      "loss": 0.3285,
+      "step": 28413
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932460708516915e-05,
+      "loss": 0.3381,
+      "step": 28414
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324547731441823e-05,
+      "loss": 0.2693,
+      "step": 28415
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324488375197755e-05,
+      "loss": 0.4012,
+      "step": 28416
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324429016436966e-05,
+      "loss": 0.3155,
+      "step": 28417
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932436965515947e-05,
+      "loss": 0.3139,
+      "step": 28418
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932431029136528e-05,
+      "loss": 0.3793,
+      "step": 28419
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324250925054416e-05,
+      "loss": 0.2858,
+      "step": 28420
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324191556226895e-05,
+      "loss": 0.2704,
+      "step": 28421
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324132184882733e-05,
+      "loss": 0.286,
+      "step": 28422
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932407281102194e-05,
+      "loss": 0.2983,
+      "step": 28423
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9324013434644537e-05,
+      "loss": 0.3235,
+      "step": 28424
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932395405575054e-05,
+      "loss": 0.351,
+      "step": 28425
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932389467433996e-05,
+      "loss": 0.3006,
+      "step": 28426
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9323835290412824e-05,
+      "loss": 0.3066,
+      "step": 28427
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932377590396914e-05,
+      "loss": 0.3088,
+      "step": 28428
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9323716515008924e-05,
+      "loss": 0.3085,
+      "step": 28429
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932365712353219e-05,
+      "loss": 0.3035,
+      "step": 28430
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932359772953896e-05,
+      "loss": 0.3273,
+      "step": 28431
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932353833302925e-05,
+      "loss": 0.3318,
+      "step": 28432
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9323478934003074e-05,
+      "loss": 0.3673,
+      "step": 28433
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9323419532460443e-05,
+      "loss": 0.3483,
+      "step": 28434
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9323360128401378e-05,
+      "loss": 0.3426,
+      "step": 28435
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9323300721825897e-05,
+      "loss": 0.2893,
+      "step": 28436
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932324131273401e-05,
+      "loss": 0.3063,
+      "step": 28437
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932318190112574e-05,
+      "loss": 0.4013,
+      "step": 28438
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.93231224870011e-05,
+      "loss": 0.3738,
+      "step": 28439
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9323063070360103e-05,
+      "loss": 0.2939,
+      "step": 28440
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932300365120277e-05,
+      "loss": 0.3521,
+      "step": 28441
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932294422952911e-05,
+      "loss": 0.3189,
+      "step": 28442
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322884805339147e-05,
+      "loss": 0.3112,
+      "step": 28443
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322825378632898e-05,
+      "loss": 0.3215,
+      "step": 28444
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932276594941037e-05,
+      "loss": 0.3409,
+      "step": 28445
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322706517671586e-05,
+      "loss": 0.2968,
+      "step": 28446
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322647083416556e-05,
+      "loss": 0.3204,
+      "step": 28447
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322587646645306e-05,
+      "loss": 0.3285,
+      "step": 28448
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932252820735784e-05,
+      "loss": 0.3124,
+      "step": 28449
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322468765554184e-05,
+      "loss": 0.3299,
+      "step": 28450
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322409321234346e-05,
+      "loss": 0.3448,
+      "step": 28451
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322349874398353e-05,
+      "loss": 0.2821,
+      "step": 28452
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322290425046207e-05,
+      "loss": 0.3536,
+      "step": 28453
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322230973177937e-05,
+      "loss": 0.3122,
+      "step": 28454
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322171518793552e-05,
+      "loss": 0.3226,
+      "step": 28455
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932211206189307e-05,
+      "loss": 0.3297,
+      "step": 28456
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9322052602476498e-05,
+      "loss": 0.2677,
+      "step": 28457
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321993140543867e-05,
+      "loss": 0.3035,
+      "step": 28458
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932193367609519e-05,
+      "loss": 0.3065,
+      "step": 28459
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321874209130474e-05,
+      "loss": 0.294,
+      "step": 28460
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321814739649746e-05,
+      "loss": 0.2872,
+      "step": 28461
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932175526765301e-05,
+      "loss": 0.322,
+      "step": 28462
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321695793140294e-05,
+      "loss": 0.3283,
+      "step": 28463
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321636316111605e-05,
+      "loss": 0.2997,
+      "step": 28464
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321576836566963e-05,
+      "loss": 0.3578,
+      "step": 28465
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321517354506386e-05,
+      "loss": 0.3108,
+      "step": 28466
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321457869929888e-05,
+      "loss": 0.3095,
+      "step": 28467
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321398382837485e-05,
+      "loss": 0.2873,
+      "step": 28468
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932133889322919e-05,
+      "loss": 0.2868,
+      "step": 28469
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321279401105025e-05,
+      "loss": 0.2868,
+      "step": 28470
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321219906465003e-05,
+      "loss": 0.356,
+      "step": 28471
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932116040930914e-05,
+      "loss": 0.2917,
+      "step": 28472
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9321100909637448e-05,
+      "loss": 0.3318,
+      "step": 28473
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932104140744995e-05,
+      "loss": 0.323,
+      "step": 28474
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320981902746663e-05,
+      "loss": 0.3543,
+      "step": 28475
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320922395527598e-05,
+      "loss": 0.3353,
+      "step": 28476
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932086288579277e-05,
+      "loss": 0.3418,
+      "step": 28477
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.93208033735422e-05,
+      "loss": 0.3333,
+      "step": 28478
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.93207438587759e-05,
+      "loss": 0.3498,
+      "step": 28479
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320684341493888e-05,
+      "loss": 0.3,
+      "step": 28480
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320624821696178e-05,
+      "loss": 0.312,
+      "step": 28481
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.932056529938279e-05,
+      "loss": 0.3395,
+      "step": 28482
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320505774553737e-05,
+      "loss": 0.3043,
+      "step": 28483
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320446247209037e-05,
+      "loss": 0.3403,
+      "step": 28484
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320386717348704e-05,
+      "loss": 0.336,
+      "step": 28485
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320327184972757e-05,
+      "loss": 0.2988,
+      "step": 28486
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320267650081207e-05,
+      "loss": 0.2645,
+      "step": 28487
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320208112674075e-05,
+      "loss": 0.307,
+      "step": 28488
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320148572751373e-05,
+      "loss": 0.2773,
+      "step": 28489
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320089030313122e-05,
+      "loss": 0.3453,
+      "step": 28490
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9320029485359336e-05,
+      "loss": 0.3418,
+      "step": 28491
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319969937890028e-05,
+      "loss": 0.3048,
+      "step": 28492
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319910387905218e-05,
+      "loss": 0.2955,
+      "step": 28493
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.931985083540492e-05,
+      "loss": 0.3673,
+      "step": 28494
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319791280389152e-05,
+      "loss": 0.3339,
+      "step": 28495
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319731722857926e-05,
+      "loss": 0.2913,
+      "step": 28496
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319672162811263e-05,
+      "loss": 0.3062,
+      "step": 28497
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319612600249176e-05,
+      "loss": 0.3463,
+      "step": 28498
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319553035171683e-05,
+      "loss": 0.3097,
+      "step": 28499
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319493467578797e-05,
+      "loss": 0.3114,
+      "step": 28500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319433897470537e-05,
+      "loss": 0.3264,
+      "step": 28501
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319374324846922e-05,
+      "loss": 0.3491,
+      "step": 28502
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319314749707962e-05,
+      "loss": 0.2998,
+      "step": 28503
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319255172053673e-05,
+      "loss": 0.3189,
+      "step": 28504
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319195591884075e-05,
+      "loss": 0.2855,
+      "step": 28505
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319136009199182e-05,
+      "loss": 0.2806,
+      "step": 28506
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9319076423999012e-05,
+      "loss": 0.3176,
+      "step": 28507
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.931901683628358e-05,
+      "loss": 0.2991,
+      "step": 28508
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.93189572460529e-05,
+      "loss": 0.3373,
+      "step": 28509
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318897653306992e-05,
+      "loss": 0.3269,
+      "step": 28510
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318838058045866e-05,
+      "loss": 0.3962,
+      "step": 28511
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318778460269543e-05,
+      "loss": 0.3163,
+      "step": 28512
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318718859978045e-05,
+      "loss": 0.2809,
+      "step": 28513
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318659257171376e-05,
+      "loss": 0.2995,
+      "step": 28514
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318599651849555e-05,
+      "loss": 0.355,
+      "step": 28515
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318540044012605e-05,
+      "loss": 0.3079,
+      "step": 28516
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318480433660533e-05,
+      "loss": 0.3244,
+      "step": 28517
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318420820793363e-05,
+      "loss": 0.311,
+      "step": 28518
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.931836120541111e-05,
+      "loss": 0.3088,
+      "step": 28519
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.931830158751378e-05,
+      "loss": 0.2699,
+      "step": 28520
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318241967101402e-05,
+      "loss": 0.3402,
+      "step": 28521
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318182344173984e-05,
+      "loss": 0.3321,
+      "step": 28522
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318122718731546e-05,
+      "loss": 0.3359,
+      "step": 28523
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318063090774105e-05,
+      "loss": 0.2972,
+      "step": 28524
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9318003460301677e-05,
+      "loss": 0.2827,
+      "step": 28525
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9317943827314272e-05,
+      "loss": 0.3133,
+      "step": 28526
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.931788419181191e-05,
+      "loss": 0.3033,
+      "step": 28527
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9317824553794612e-05,
+      "loss": 0.3028,
+      "step": 28528
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9317764913262387e-05,
+      "loss": 0.3332,
+      "step": 28529
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9317705270215254e-05,
+      "loss": 0.3752,
+      "step": 28530
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9317645624653228e-05,
+      "loss": 0.3028,
+      "step": 28531
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9317585976576328e-05,
+      "loss": 0.3074,
+      "step": 28532
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9317526325984564e-05,
+      "loss": 0.2942,
+      "step": 28533
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.931746667287796e-05,
+      "loss": 0.3184,
+      "step": 28534
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9317407017256526e-05,
+      "loss": 0.364,
+      "step": 28535
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.931734735912028e-05,
+      "loss": 0.3202,
+      "step": 28536
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931728769846924e-05,
+      "loss": 0.3367,
+      "step": 28537
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931722803530342e-05,
+      "loss": 0.2921,
+      "step": 28538
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9317168369622838e-05,
+      "loss": 0.2777,
+      "step": 28539
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9317108701427506e-05,
+      "loss": 0.3113,
+      "step": 28540
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9317049030717445e-05,
+      "loss": 0.3088,
+      "step": 28541
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9316989357492666e-05,
+      "loss": 0.3532,
+      "step": 28542
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931692968175319e-05,
+      "loss": 0.2641,
+      "step": 28543
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9316870003499035e-05,
+      "loss": 0.3022,
+      "step": 28544
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9316810322730207e-05,
+      "loss": 0.2703,
+      "step": 28545
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931675063944673e-05,
+      "loss": 0.3294,
+      "step": 28546
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931669095364862e-05,
+      "loss": 0.3041,
+      "step": 28547
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931663126533589e-05,
+      "loss": 0.3518,
+      "step": 28548
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931657157450856e-05,
+      "loss": 0.3065,
+      "step": 28549
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931651188116664e-05,
+      "loss": 0.3152,
+      "step": 28550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9316452185310154e-05,
+      "loss": 0.3337,
+      "step": 28551
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931639248693911e-05,
+      "loss": 0.2819,
+      "step": 28552
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9316332786053532e-05,
+      "loss": 0.2985,
+      "step": 28553
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931627308265343e-05,
+      "loss": 0.3096,
+      "step": 28554
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931621337673882e-05,
+      "loss": 0.3227,
+      "step": 28555
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931615366830973e-05,
+      "loss": 0.3571,
+      "step": 28556
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9316093957366155e-05,
+      "loss": 0.3408,
+      "step": 28557
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9316034243908127e-05,
+      "loss": 0.349,
+      "step": 28558
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315974527935658e-05,
+      "loss": 0.3129,
+      "step": 28559
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315914809448765e-05,
+      "loss": 0.4171,
+      "step": 28560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931585508844746e-05,
+      "loss": 0.3186,
+      "step": 28561
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315795364931767e-05,
+      "loss": 0.2833,
+      "step": 28562
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315735638901697e-05,
+      "loss": 0.2887,
+      "step": 28563
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931567591035726e-05,
+      "loss": 0.282,
+      "step": 28564
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315616179298486e-05,
+      "loss": 0.3068,
+      "step": 28565
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931555644572538e-05,
+      "loss": 0.3303,
+      "step": 28566
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315496709637963e-05,
+      "loss": 0.2973,
+      "step": 28567
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315436971036246e-05,
+      "loss": 0.3187,
+      "step": 28568
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315377229920255e-05,
+      "loss": 0.3663,
+      "step": 28569
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315317486289995e-05,
+      "loss": 0.3004,
+      "step": 28570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931525774014549e-05,
+      "loss": 0.2836,
+      "step": 28571
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315197991486752e-05,
+      "loss": 0.347,
+      "step": 28572
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.93151382403138e-05,
+      "loss": 0.2906,
+      "step": 28573
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931507848662665e-05,
+      "loss": 0.3621,
+      "step": 28574
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9315018730425313e-05,
+      "loss": 0.3271,
+      "step": 28575
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9314958971709813e-05,
+      "loss": 0.2591,
+      "step": 28576
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9314899210480157e-05,
+      "loss": 0.3416,
+      "step": 28577
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9314839446736368e-05,
+      "loss": 0.3241,
+      "step": 28578
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9314779680478463e-05,
+      "loss": 0.3956,
+      "step": 28579
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9314719911706452e-05,
+      "loss": 0.2796,
+      "step": 28580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9314660140420357e-05,
+      "loss": 0.3784,
+      "step": 28581
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931460036662019e-05,
+      "loss": 0.347,
+      "step": 28582
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931454059030597e-05,
+      "loss": 0.3082,
+      "step": 28583
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931448081147771e-05,
+      "loss": 0.3325,
+      "step": 28584
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931442103013543e-05,
+      "loss": 0.3016,
+      "step": 28585
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9314361246279145e-05,
+      "loss": 0.3026,
+      "step": 28586
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931430145990887e-05,
+      "loss": 0.329,
+      "step": 28587
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931424167102462e-05,
+      "loss": 0.3714,
+      "step": 28588
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9314181879626416e-05,
+      "loss": 0.2813,
+      "step": 28589
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9314122085714265e-05,
+      "loss": 0.2861,
+      "step": 28590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931406228928819e-05,
+      "loss": 0.2931,
+      "step": 28591
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931400249034821e-05,
+      "loss": 0.3219,
+      "step": 28592
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313942688894335e-05,
+      "loss": 0.3382,
+      "step": 28593
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313882884926585e-05,
+      "loss": 0.3249,
+      "step": 28594
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313823078444972e-05,
+      "loss": 0.3185,
+      "step": 28595
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313763269449515e-05,
+      "loss": 0.3024,
+      "step": 28596
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931370345794023e-05,
+      "loss": 0.3067,
+      "step": 28597
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313643643917135e-05,
+      "loss": 0.2757,
+      "step": 28598
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313583827380242e-05,
+      "loss": 0.3397,
+      "step": 28599
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931352400832957e-05,
+      "loss": 0.3282,
+      "step": 28600
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313464186765133e-05,
+      "loss": 0.3429,
+      "step": 28601
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313404362686953e-05,
+      "loss": 0.3237,
+      "step": 28602
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313344536095037e-05,
+      "loss": 0.408,
+      "step": 28603
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313284706989405e-05,
+      "loss": 0.2623,
+      "step": 28604
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313224875370074e-05,
+      "loss": 0.282,
+      "step": 28605
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931316504123706e-05,
+      "loss": 0.2925,
+      "step": 28606
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313105204590387e-05,
+      "loss": 0.3004,
+      "step": 28607
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9313045365430054e-05,
+      "loss": 0.314,
+      "step": 28608
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931298552375609e-05,
+      "loss": 0.2897,
+      "step": 28609
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9312925679568507e-05,
+      "loss": 0.2845,
+      "step": 28610
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9312865832867324e-05,
+      "loss": 0.3234,
+      "step": 28611
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9312805983652554e-05,
+      "loss": 0.2939,
+      "step": 28612
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931274613192421e-05,
+      "loss": 0.3273,
+      "step": 28613
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931268627768232e-05,
+      "loss": 0.2516,
+      "step": 28614
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9312626420926887e-05,
+      "loss": 0.285,
+      "step": 28615
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9312566561657934e-05,
+      "loss": 0.3429,
+      "step": 28616
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9312506699875475e-05,
+      "loss": 0.2908,
+      "step": 28617
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9312446835579524e-05,
+      "loss": 0.2591,
+      "step": 28618
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9312386968770108e-05,
+      "loss": 0.271,
+      "step": 28619
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931232709944723e-05,
+      "loss": 0.3326,
+      "step": 28620
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9312267227610914e-05,
+      "loss": 0.3127,
+      "step": 28621
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931220735326117e-05,
+      "loss": 0.3086,
+      "step": 28622
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931214747639802e-05,
+      "loss": 0.317,
+      "step": 28623
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9312087597021475e-05,
+      "loss": 0.3168,
+      "step": 28624
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931202771513156e-05,
+      "loss": 0.3055,
+      "step": 28625
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931196783072828e-05,
+      "loss": 0.321,
+      "step": 28626
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931190794381166e-05,
+      "loss": 0.3633,
+      "step": 28627
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931184805438171e-05,
+      "loss": 0.3437,
+      "step": 28628
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9311788162438447e-05,
+      "loss": 0.2983,
+      "step": 28629
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9311728267981892e-05,
+      "loss": 0.3215,
+      "step": 28630
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9311668371012055e-05,
+      "loss": 0.2923,
+      "step": 28631
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.931160847152896e-05,
+      "loss": 0.3312,
+      "step": 28632
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9311548569532612e-05,
+      "loss": 0.2547,
+      "step": 28633
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9311488665023036e-05,
+      "loss": 0.2984,
+      "step": 28634
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931142875800025e-05,
+      "loss": 0.3032,
+      "step": 28635
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931136884846426e-05,
+      "loss": 0.3389,
+      "step": 28636
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931130893641509e-05,
+      "loss": 0.3373,
+      "step": 28637
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9311249021852755e-05,
+      "loss": 0.3157,
+      "step": 28638
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9311189104777273e-05,
+      "loss": 0.3233,
+      "step": 28639
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9311129185188655e-05,
+      "loss": 0.3181,
+      "step": 28640
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931106926308692e-05,
+      "loss": 0.4227,
+      "step": 28641
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931100933847208e-05,
+      "loss": 0.2722,
+      "step": 28642
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931094941134416e-05,
+      "loss": 0.3464,
+      "step": 28643
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931088948170317e-05,
+      "loss": 0.3339,
+      "step": 28644
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9310829549549126e-05,
+      "loss": 0.3708,
+      "step": 28645
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9310769614882047e-05,
+      "loss": 0.299,
+      "step": 28646
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9310709677701946e-05,
+      "loss": 0.3289,
+      "step": 28647
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931064973800884e-05,
+      "loss": 0.3416,
+      "step": 28648
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931058979580275e-05,
+      "loss": 0.2931,
+      "step": 28649
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9310529851083686e-05,
+      "loss": 0.2827,
+      "step": 28650
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9310469903851666e-05,
+      "loss": 0.3099,
+      "step": 28651
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931040995410671e-05,
+      "loss": 0.2976,
+      "step": 28652
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931035000184883e-05,
+      "loss": 0.3018,
+      "step": 28653
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931029004707804e-05,
+      "loss": 0.3297,
+      "step": 28654
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931023008979436e-05,
+      "loss": 0.3099,
+      "step": 28655
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9310170129997807e-05,
+      "loss": 0.2927,
+      "step": 28656
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9310110167688394e-05,
+      "loss": 0.3256,
+      "step": 28657
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.931005020286614e-05,
+      "loss": 0.3832,
+      "step": 28658
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930999023553106e-05,
+      "loss": 0.2948,
+      "step": 28659
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930993026568317e-05,
+      "loss": 0.3124,
+      "step": 28660
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9309870293322482e-05,
+      "loss": 0.2823,
+      "step": 28661
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9309810318449023e-05,
+      "loss": 0.3108,
+      "step": 28662
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.93097503410628e-05,
+      "loss": 0.3388,
+      "step": 28663
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9309690361163832e-05,
+      "loss": 0.3036,
+      "step": 28664
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9309630378752135e-05,
+      "loss": 0.3196,
+      "step": 28665
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9309570393827725e-05,
+      "loss": 0.295,
+      "step": 28666
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930951040639062e-05,
+      "loss": 0.3724,
+      "step": 28667
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930945041644083e-05,
+      "loss": 0.2891,
+      "step": 28668
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930939042397838e-05,
+      "loss": 0.2975,
+      "step": 28669
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9309330429003282e-05,
+      "loss": 0.2949,
+      "step": 28670
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930927043151555e-05,
+      "loss": 0.3198,
+      "step": 28671
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9309210431515204e-05,
+      "loss": 0.3069,
+      "step": 28672
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930915042900226e-05,
+      "loss": 0.3382,
+      "step": 28673
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9309090423976728e-05,
+      "loss": 0.2894,
+      "step": 28674
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9309030416438632e-05,
+      "loss": 0.2577,
+      "step": 28675
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308970406387986e-05,
+      "loss": 0.2715,
+      "step": 28676
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308910393824806e-05,
+      "loss": 0.2834,
+      "step": 28677
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308850378749106e-05,
+      "loss": 0.311,
+      "step": 28678
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308790361160903e-05,
+      "loss": 0.3012,
+      "step": 28679
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308730341060214e-05,
+      "loss": 0.3067,
+      "step": 28680
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308670318447056e-05,
+      "loss": 0.3119,
+      "step": 28681
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308610293321446e-05,
+      "loss": 0.2716,
+      "step": 28682
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308550265683393e-05,
+      "loss": 0.2887,
+      "step": 28683
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308490235532926e-05,
+      "loss": 0.3325,
+      "step": 28684
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930843020287005e-05,
+      "loss": 0.3476,
+      "step": 28685
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308370167694784e-05,
+      "loss": 0.2984,
+      "step": 28686
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308310130007147e-05,
+      "loss": 0.3157,
+      "step": 28687
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308250089807155e-05,
+      "loss": 0.2893,
+      "step": 28688
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930819004709482e-05,
+      "loss": 0.3514,
+      "step": 28689
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308130001870165e-05,
+      "loss": 0.3391,
+      "step": 28690
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.93080699541332e-05,
+      "loss": 0.3449,
+      "step": 28691
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9308009903883942e-05,
+      "loss": 0.3265,
+      "step": 28692
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307949851122408e-05,
+      "loss": 0.3378,
+      "step": 28693
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930788979584862e-05,
+      "loss": 0.3166,
+      "step": 28694
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307829738062583e-05,
+      "loss": 0.2818,
+      "step": 28695
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307769677764323e-05,
+      "loss": 0.2951,
+      "step": 28696
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930770961495385e-05,
+      "loss": 0.2709,
+      "step": 28697
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307649549631184e-05,
+      "loss": 0.3802,
+      "step": 28698
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930758948179634e-05,
+      "loss": 0.3291,
+      "step": 28699
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307529411449334e-05,
+      "loss": 0.2823,
+      "step": 28700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307469338590182e-05,
+      "loss": 0.2918,
+      "step": 28701
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307409263218902e-05,
+      "loss": 0.3198,
+      "step": 28702
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930734918533551e-05,
+      "loss": 0.2877,
+      "step": 28703
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307289104940017e-05,
+      "loss": 0.3421,
+      "step": 28704
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307229022032445e-05,
+      "loss": 0.2988,
+      "step": 28705
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307168936612807e-05,
+      "loss": 0.3467,
+      "step": 28706
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307108848681123e-05,
+      "loss": 0.3046,
+      "step": 28707
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9307048758237404e-05,
+      "loss": 0.3156,
+      "step": 28708
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306988665281673e-05,
+      "loss": 0.3566,
+      "step": 28709
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306928569813937e-05,
+      "loss": 0.3135,
+      "step": 28710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930686847183422e-05,
+      "loss": 0.3526,
+      "step": 28711
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306808371342535e-05,
+      "loss": 0.3066,
+      "step": 28712
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306748268338903e-05,
+      "loss": 0.3234,
+      "step": 28713
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930668816282333e-05,
+      "loss": 0.2978,
+      "step": 28714
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306628054795844e-05,
+      "loss": 0.3574,
+      "step": 28715
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930656794425645e-05,
+      "loss": 0.2542,
+      "step": 28716
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306507831205176e-05,
+      "loss": 0.3254,
+      "step": 28717
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306447715642027e-05,
+      "loss": 0.3569,
+      "step": 28718
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306387597567026e-05,
+      "loss": 0.2707,
+      "step": 28719
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306327476980186e-05,
+      "loss": 0.312,
+      "step": 28720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306267353881527e-05,
+      "loss": 0.2702,
+      "step": 28721
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930620722827106e-05,
+      "loss": 0.2997,
+      "step": 28722
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930614710014881e-05,
+      "loss": 0.3212,
+      "step": 28723
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306086969514782e-05,
+      "loss": 0.2903,
+      "step": 28724
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9306026836369e-05,
+      "loss": 0.3156,
+      "step": 28725
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9305966700711477e-05,
+      "loss": 0.3681,
+      "step": 28726
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930590656254223e-05,
+      "loss": 0.3009,
+      "step": 28727
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9305846421861276e-05,
+      "loss": 0.2502,
+      "step": 28728
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930578627866863e-05,
+      "loss": 0.345,
+      "step": 28729
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930572613296431e-05,
+      "loss": 0.3008,
+      "step": 28730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.930566598474833e-05,
+      "loss": 0.3255,
+      "step": 28731
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9305605834020706e-05,
+      "loss": 0.2814,
+      "step": 28732
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9305545680781457e-05,
+      "loss": 0.3064,
+      "step": 28733
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.93054855250306e-05,
+      "loss": 0.4274,
+      "step": 28734
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930542536676814e-05,
+      "loss": 0.3482,
+      "step": 28735
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9305365205994114e-05,
+      "loss": 0.3072,
+      "step": 28736
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930530504270852e-05,
+      "loss": 0.3267,
+      "step": 28737
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930524487691138e-05,
+      "loss": 0.3119,
+      "step": 28738
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9305184708602714e-05,
+      "loss": 0.2954,
+      "step": 28739
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930512453778253e-05,
+      "loss": 0.3103,
+      "step": 28740
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9305064364450858e-05,
+      "loss": 0.3004,
+      "step": 28741
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9305004188607698e-05,
+      "loss": 0.2919,
+      "step": 28742
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930494401025308e-05,
+      "loss": 0.3246,
+      "step": 28743
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930488382938701e-05,
+      "loss": 0.2872,
+      "step": 28744
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9304823646009506e-05,
+      "loss": 0.314,
+      "step": 28745
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930476346012059e-05,
+      "loss": 0.2706,
+      "step": 28746
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9304703271720275e-05,
+      "loss": 0.3194,
+      "step": 28747
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9304643080808575e-05,
+      "loss": 0.31,
+      "step": 28748
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930458288738551e-05,
+      "loss": 0.3127,
+      "step": 28749
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.93045226914511e-05,
+      "loss": 0.2936,
+      "step": 28750
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9304462493005346e-05,
+      "loss": 0.3003,
+      "step": 28751
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930440229204828e-05,
+      "loss": 0.313,
+      "step": 28752
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930434208857991e-05,
+      "loss": 0.2899,
+      "step": 28753
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9304281882600256e-05,
+      "loss": 0.3533,
+      "step": 28754
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9304221674109333e-05,
+      "loss": 0.2859,
+      "step": 28755
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9304161463107153e-05,
+      "loss": 0.3791,
+      "step": 28756
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930410124959374e-05,
+      "loss": 0.3168,
+      "step": 28757
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9304041033569107e-05,
+      "loss": 0.2929,
+      "step": 28758
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930398081503327e-05,
+      "loss": 0.3205,
+      "step": 28759
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303920593986243e-05,
+      "loss": 0.3986,
+      "step": 28760
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303860370428044e-05,
+      "loss": 0.301,
+      "step": 28761
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303800144358693e-05,
+      "loss": 0.3208,
+      "step": 28762
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303739915778202e-05,
+      "loss": 0.2971,
+      "step": 28763
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303679684686586e-05,
+      "loss": 0.3192,
+      "step": 28764
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303619451083864e-05,
+      "loss": 0.3503,
+      "step": 28765
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930355921497005e-05,
+      "loss": 0.2667,
+      "step": 28766
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303498976345168e-05,
+      "loss": 0.3429,
+      "step": 28767
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303438735209222e-05,
+      "loss": 0.3195,
+      "step": 28768
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303378491562234e-05,
+      "loss": 0.3688,
+      "step": 28769
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303318245404226e-05,
+      "loss": 0.3046,
+      "step": 28770
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303257996735207e-05,
+      "loss": 0.3308,
+      "step": 28771
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303197745555192e-05,
+      "loss": 0.3156,
+      "step": 28772
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303137491864202e-05,
+      "loss": 0.2656,
+      "step": 28773
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303077235662254e-05,
+      "loss": 0.2588,
+      "step": 28774
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9303016976949358e-05,
+      "loss": 0.3108,
+      "step": 28775
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9302956715725538e-05,
+      "loss": 0.2704,
+      "step": 28776
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9302896451990804e-05,
+      "loss": 0.2846,
+      "step": 28777
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9302836185745177e-05,
+      "loss": 0.2949,
+      "step": 28778
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930277591698867e-05,
+      "loss": 0.3426,
+      "step": 28779
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9302715645721298e-05,
+      "loss": 0.2996,
+      "step": 28780
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930265537194308e-05,
+      "loss": 0.3719,
+      "step": 28781
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930259509565404e-05,
+      "loss": 0.3616,
+      "step": 28782
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9302534816854176e-05,
+      "loss": 0.2977,
+      "step": 28783
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930247453554352e-05,
+      "loss": 0.2994,
+      "step": 28784
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930241425172208e-05,
+      "loss": 0.3104,
+      "step": 28785
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9302353965389872e-05,
+      "loss": 0.2932,
+      "step": 28786
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930229367654692e-05,
+      "loss": 0.3055,
+      "step": 28787
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9302233385193237e-05,
+      "loss": 0.2972,
+      "step": 28788
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9302173091328833e-05,
+      "loss": 0.2932,
+      "step": 28789
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9302112794953732e-05,
+      "loss": 0.3312,
+      "step": 28790
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9302052496067944e-05,
+      "loss": 0.3101,
+      "step": 28791
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930199219467149e-05,
+      "loss": 0.3677,
+      "step": 28792
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9301931890764388e-05,
+      "loss": 0.2975,
+      "step": 28793
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9301871584346647e-05,
+      "loss": 0.3675,
+      "step": 28794
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930181127541829e-05,
+      "loss": 0.3136,
+      "step": 28795
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930175096397933e-05,
+      "loss": 0.3554,
+      "step": 28796
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930169065002978e-05,
+      "loss": 0.3328,
+      "step": 28797
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9301630333569666e-05,
+      "loss": 0.2693,
+      "step": 28798
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9301570014598995e-05,
+      "loss": 0.3003,
+      "step": 28799
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9301509693117787e-05,
+      "loss": 0.2894,
+      "step": 28800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930144936912606e-05,
+      "loss": 0.2825,
+      "step": 28801
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9301389042623827e-05,
+      "loss": 0.3446,
+      "step": 28802
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9301328713611107e-05,
+      "loss": 0.2894,
+      "step": 28803
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9301268382087913e-05,
+      "loss": 0.2734,
+      "step": 28804
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9301208048054266e-05,
+      "loss": 0.3253,
+      "step": 28805
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930114771151018e-05,
+      "loss": 0.3735,
+      "step": 28806
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9301087372455666e-05,
+      "loss": 0.331,
+      "step": 28807
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930102703089075e-05,
+      "loss": 0.3402,
+      "step": 28808
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930096668681544e-05,
+      "loss": 0.3049,
+      "step": 28809
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9300906340229753e-05,
+      "loss": 0.3499,
+      "step": 28810
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9300845991133712e-05,
+      "loss": 0.2981,
+      "step": 28811
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930078563952733e-05,
+      "loss": 0.3159,
+      "step": 28812
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930072528541062e-05,
+      "loss": 0.3185,
+      "step": 28813
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.93006649287836e-05,
+      "loss": 0.3262,
+      "step": 28814
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930060456964629e-05,
+      "loss": 0.3231,
+      "step": 28815
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.93005442079987e-05,
+      "loss": 0.3577,
+      "step": 28816
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9300483843840852e-05,
+      "loss": 0.3817,
+      "step": 28817
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9300423477172764e-05,
+      "loss": 0.3398,
+      "step": 28818
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930036310799444e-05,
+      "loss": 0.3146,
+      "step": 28819
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9300302736305908e-05,
+      "loss": 0.2808,
+      "step": 28820
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930024236210718e-05,
+      "loss": 0.3243,
+      "step": 28821
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930018198539828e-05,
+      "loss": 0.3789,
+      "step": 28822
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930012160617921e-05,
+      "loss": 0.2697,
+      "step": 28823
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9300061224449994e-05,
+      "loss": 0.3273,
+      "step": 28824
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.930000084021065e-05,
+      "loss": 0.2895,
+      "step": 28825
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9299940453461193e-05,
+      "loss": 0.3535,
+      "step": 28826
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9299880064201635e-05,
+      "loss": 0.2968,
+      "step": 28827
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9299819672432e-05,
+      "loss": 0.3349,
+      "step": 28828
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.92997592781523e-05,
+      "loss": 0.3013,
+      "step": 28829
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9299698881362548e-05,
+      "loss": 0.2502,
+      "step": 28830
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9299638482062762e-05,
+      "loss": 0.3575,
+      "step": 28831
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9299578080252964e-05,
+      "loss": 0.2941,
+      "step": 28832
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929951767593317e-05,
+      "loss": 0.3161,
+      "step": 28833
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9299457269103387e-05,
+      "loss": 0.3816,
+      "step": 28834
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9299396859763638e-05,
+      "loss": 0.3063,
+      "step": 28835
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929933644791394e-05,
+      "loss": 0.2934,
+      "step": 28836
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929927603355431e-05,
+      "loss": 0.3367,
+      "step": 28837
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9299215616684757e-05,
+      "loss": 0.284,
+      "step": 28838
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9299155197305304e-05,
+      "loss": 0.2677,
+      "step": 28839
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9299094775415966e-05,
+      "loss": 0.3117,
+      "step": 28840
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9299034351016758e-05,
+      "loss": 0.289,
+      "step": 28841
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.92989739241077e-05,
+      "loss": 0.3411,
+      "step": 28842
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.92989134946888e-05,
+      "loss": 0.3017,
+      "step": 28843
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298853062760084e-05,
+      "loss": 0.2601,
+      "step": 28844
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298792628321563e-05,
+      "loss": 0.2859,
+      "step": 28845
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298732191373254e-05,
+      "loss": 0.2781,
+      "step": 28846
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298671751915175e-05,
+      "loss": 0.3118,
+      "step": 28847
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298611309947338e-05,
+      "loss": 0.2953,
+      "step": 28848
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298550865469766e-05,
+      "loss": 0.2725,
+      "step": 28849
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929849041848247e-05,
+      "loss": 0.3597,
+      "step": 28850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929842996898547e-05,
+      "loss": 0.2755,
+      "step": 28851
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298369516978777e-05,
+      "loss": 0.3213,
+      "step": 28852
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929830906246241e-05,
+      "loss": 0.2814,
+      "step": 28853
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298248605436388e-05,
+      "loss": 0.2877,
+      "step": 28854
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298188145900725e-05,
+      "loss": 0.3146,
+      "step": 28855
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929812768385544e-05,
+      "loss": 0.2551,
+      "step": 28856
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298067219300544e-05,
+      "loss": 0.3107,
+      "step": 28857
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9298006752236056e-05,
+      "loss": 0.3135,
+      "step": 28858
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297946282661996e-05,
+      "loss": 0.3008,
+      "step": 28859
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297885810578373e-05,
+      "loss": 0.3933,
+      "step": 28860
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929782533598521e-05,
+      "loss": 0.3365,
+      "step": 28861
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.2638,
+      "step": 28862
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297704379270316e-05,
+      "loss": 0.2903,
+      "step": 28863
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297643897148618e-05,
+      "loss": 0.2788,
+      "step": 28864
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297583412517446e-05,
+      "loss": 0.3158,
+      "step": 28865
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929752292537681e-05,
+      "loss": 0.3091,
+      "step": 28866
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297462435726734e-05,
+      "loss": 0.2949,
+      "step": 28867
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297401943567228e-05,
+      "loss": 0.3023,
+      "step": 28868
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297341448898305e-05,
+      "loss": 0.2741,
+      "step": 28869
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929728095171999e-05,
+      "loss": 0.3432,
+      "step": 28870
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297220452032295e-05,
+      "loss": 0.2725,
+      "step": 28871
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297159949835235e-05,
+      "loss": 0.3087,
+      "step": 28872
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929709944512883e-05,
+      "loss": 0.2749,
+      "step": 28873
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297038937913097e-05,
+      "loss": 0.3366,
+      "step": 28874
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929697842818804e-05,
+      "loss": 0.3292,
+      "step": 28875
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296917915953695e-05,
+      "loss": 0.2804,
+      "step": 28876
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296857401210065e-05,
+      "loss": 0.319,
+      "step": 28877
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929679688395717e-05,
+      "loss": 0.2801,
+      "step": 28878
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296736364195026e-05,
+      "loss": 0.3116,
+      "step": 28879
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929667584192365e-05,
+      "loss": 0.3051,
+      "step": 28880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296615317143056e-05,
+      "loss": 0.2983,
+      "step": 28881
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296554789853265e-05,
+      "loss": 0.4146,
+      "step": 28882
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296494260054288e-05,
+      "loss": 0.3488,
+      "step": 28883
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296433727746145e-05,
+      "loss": 0.3929,
+      "step": 28884
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296373192928847e-05,
+      "loss": 0.3273,
+      "step": 28885
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929631265560242e-05,
+      "loss": 0.3466,
+      "step": 28886
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296252115766872e-05,
+      "loss": 0.2916,
+      "step": 28887
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296191573422222e-05,
+      "loss": 0.3049,
+      "step": 28888
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296131028568488e-05,
+      "loss": 0.3033,
+      "step": 28889
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296070481205683e-05,
+      "loss": 0.2905,
+      "step": 28890
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296009931333824e-05,
+      "loss": 0.3164,
+      "step": 28891
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9295949378952932e-05,
+      "loss": 0.3355,
+      "step": 28892
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929588882406302e-05,
+      "loss": 0.2928,
+      "step": 28893
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.92958282666641e-05,
+      "loss": 0.3028,
+      "step": 28894
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9295767706756196e-05,
+      "loss": 0.3712,
+      "step": 28895
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9295707144339318e-05,
+      "loss": 0.3024,
+      "step": 28896
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929564657941349e-05,
+      "loss": 0.296,
+      "step": 28897
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9295586011978717e-05,
+      "loss": 0.3205,
+      "step": 28898
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9295525442035026e-05,
+      "loss": 0.2617,
+      "step": 28899
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929546486958243e-05,
+      "loss": 0.2961,
+      "step": 28900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9295404294620942e-05,
+      "loss": 0.3128,
+      "step": 28901
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9295343717150584e-05,
+      "loss": 0.306,
+      "step": 28902
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9295283137171366e-05,
+      "loss": 0.3509,
+      "step": 28903
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9295222554683308e-05,
+      "loss": 0.2612,
+      "step": 28904
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929516196968643e-05,
+      "loss": 0.2847,
+      "step": 28905
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9295101382180738e-05,
+      "loss": 0.367,
+      "step": 28906
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929504079216626e-05,
+      "loss": 0.3143,
+      "step": 28907
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294980199643002e-05,
+      "loss": 0.3309,
+      "step": 28908
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929491960461099e-05,
+      "loss": 0.2969,
+      "step": 28909
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294859007070233e-05,
+      "loss": 0.3155,
+      "step": 28910
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294798407020752e-05,
+      "loss": 0.3117,
+      "step": 28911
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929473780446256e-05,
+      "loss": 0.283,
+      "step": 28912
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294677199395676e-05,
+      "loss": 0.3042,
+      "step": 28913
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294616591820118e-05,
+      "loss": 0.3379,
+      "step": 28914
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294555981735896e-05,
+      "loss": 0.3327,
+      "step": 28915
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294495369143026e-05,
+      "loss": 0.2783,
+      "step": 28916
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294434754041533e-05,
+      "loss": 0.2788,
+      "step": 28917
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929437413643143e-05,
+      "loss": 0.3905,
+      "step": 28918
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929431351631273e-05,
+      "loss": 0.2988,
+      "step": 28919
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929425289368545e-05,
+      "loss": 0.3381,
+      "step": 28920
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929419226854961e-05,
+      "loss": 0.3354,
+      "step": 28921
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929413164090522e-05,
+      "loss": 0.3459,
+      "step": 28922
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294071010752303e-05,
+      "loss": 0.3434,
+      "step": 28923
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294010378090875e-05,
+      "loss": 0.2821,
+      "step": 28924
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929394974292095e-05,
+      "loss": 0.3814,
+      "step": 28925
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929388910524254e-05,
+      "loss": 0.3269,
+      "step": 28926
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9293828465055672e-05,
+      "loss": 0.3065,
+      "step": 28927
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9293767822360353e-05,
+      "loss": 0.3308,
+      "step": 28928
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9293707177156603e-05,
+      "loss": 0.2751,
+      "step": 28929
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9293646529444436e-05,
+      "loss": 0.2501,
+      "step": 28930
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9293585879223872e-05,
+      "loss": 0.2995,
+      "step": 28931
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9293525226494925e-05,
+      "loss": 0.2938,
+      "step": 28932
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9293464571257613e-05,
+      "loss": 0.3299,
+      "step": 28933
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929340391351195e-05,
+      "loss": 0.3309,
+      "step": 28934
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9293343253257957e-05,
+      "loss": 0.3285,
+      "step": 28935
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9293282590495644e-05,
+      "loss": 0.3044,
+      "step": 28936
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9293221925225033e-05,
+      "loss": 0.3672,
+      "step": 28937
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9293161257446137e-05,
+      "loss": 0.3888,
+      "step": 28938
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9293100587158973e-05,
+      "loss": 0.3166,
+      "step": 28939
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929303991436356e-05,
+      "loss": 0.2681,
+      "step": 28940
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929297923905991e-05,
+      "loss": 0.299,
+      "step": 28941
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929291856124804e-05,
+      "loss": 0.3578,
+      "step": 28942
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292857880927972e-05,
+      "loss": 0.2779,
+      "step": 28943
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292797198099716e-05,
+      "loss": 0.3376,
+      "step": 28944
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292736512763288e-05,
+      "loss": 0.3758,
+      "step": 28945
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929267582491871e-05,
+      "loss": 0.2701,
+      "step": 28946
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292615134565997e-05,
+      "loss": 0.3042,
+      "step": 28947
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292554441705157e-05,
+      "loss": 0.3117,
+      "step": 28948
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929249374633622e-05,
+      "loss": 0.2945,
+      "step": 28949
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929243304845919e-05,
+      "loss": 0.3185,
+      "step": 28950
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292372348074094e-05,
+      "loss": 0.2862,
+      "step": 28951
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292311645180942e-05,
+      "loss": 0.327,
+      "step": 28952
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929225093977975e-05,
+      "loss": 0.319,
+      "step": 28953
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292190231870537e-05,
+      "loss": 0.2505,
+      "step": 28954
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292129521453317e-05,
+      "loss": 0.2695,
+      "step": 28955
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292068808528107e-05,
+      "loss": 0.3115,
+      "step": 28956
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9292008093094928e-05,
+      "loss": 0.3344,
+      "step": 28957
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929194737515379e-05,
+      "loss": 0.3137,
+      "step": 28958
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9291886654704714e-05,
+      "loss": 0.3282,
+      "step": 28959
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9291825931747713e-05,
+      "loss": 0.2915,
+      "step": 28960
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9291765206282804e-05,
+      "loss": 0.2972,
+      "step": 28961
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9291704478310006e-05,
+      "loss": 0.3014,
+      "step": 28962
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929164374782933e-05,
+      "loss": 0.3856,
+      "step": 28963
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.92915830148408e-05,
+      "loss": 0.29,
+      "step": 28964
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929152227934443e-05,
+      "loss": 0.3425,
+      "step": 28965
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9291461541340227e-05,
+      "loss": 0.355,
+      "step": 28966
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929140080082822e-05,
+      "loss": 0.2992,
+      "step": 28967
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9291340057808422e-05,
+      "loss": 0.3482,
+      "step": 28968
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9291279312280843e-05,
+      "loss": 0.3128,
+      "step": 28969
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9291218564245507e-05,
+      "loss": 0.3468,
+      "step": 28970
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929115781370243e-05,
+      "loss": 0.3324,
+      "step": 28971
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9291097060651626e-05,
+      "loss": 0.3608,
+      "step": 28972
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9291036305093106e-05,
+      "loss": 0.3416,
+      "step": 28973
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290975547026895e-05,
+      "loss": 0.3553,
+      "step": 28974
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290914786453005e-05,
+      "loss": 0.2606,
+      "step": 28975
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290854023371458e-05,
+      "loss": 0.3144,
+      "step": 28976
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290793257782262e-05,
+      "loss": 0.3028,
+      "step": 28977
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929073248968544e-05,
+      "loss": 0.333,
+      "step": 28978
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290671719081003e-05,
+      "loss": 0.3373,
+      "step": 28979
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290610945968972e-05,
+      "loss": 0.3163,
+      "step": 28980
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290550170349362e-05,
+      "loss": 0.3189,
+      "step": 28981
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929048939222219e-05,
+      "loss": 0.2993,
+      "step": 28982
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.929042861158747e-05,
+      "loss": 0.3653,
+      "step": 28983
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290367828445216e-05,
+      "loss": 0.327,
+      "step": 28984
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290307042795455e-05,
+      "loss": 0.3617,
+      "step": 28985
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290246254638195e-05,
+      "loss": 0.3097,
+      "step": 28986
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290185463973453e-05,
+      "loss": 0.3218,
+      "step": 28987
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290124670801247e-05,
+      "loss": 0.2986,
+      "step": 28988
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290063875121592e-05,
+      "loss": 0.296,
+      "step": 28989
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9290003076934503e-05,
+      "loss": 0.2905,
+      "step": 28990
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289942276240004e-05,
+      "loss": 0.2808,
+      "step": 28991
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289881473038104e-05,
+      "loss": 0.291,
+      "step": 28992
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.928982066732882e-05,
+      "loss": 0.3333,
+      "step": 28993
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289759859112173e-05,
+      "loss": 0.3193,
+      "step": 28994
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289699048388173e-05,
+      "loss": 0.3602,
+      "step": 28995
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.928963823515684e-05,
+      "loss": 0.3291,
+      "step": 28996
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289577419418192e-05,
+      "loss": 0.2931,
+      "step": 28997
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289516601172245e-05,
+      "loss": 0.2717,
+      "step": 28998
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289455780419013e-05,
+      "loss": 0.3555,
+      "step": 28999
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289394957158515e-05,
+      "loss": 0.342,
+      "step": 29000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289334131390763e-05,
+      "loss": 0.3235,
+      "step": 29001
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289273303115776e-05,
+      "loss": 0.3147,
+      "step": 29002
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289212472333572e-05,
+      "loss": 0.2668,
+      "step": 29003
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289151639044164e-05,
+      "loss": 0.2835,
+      "step": 29004
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289090803247572e-05,
+      "loss": 0.3028,
+      "step": 29005
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9289029964943813e-05,
+      "loss": 0.2677,
+      "step": 29006
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.92889691241329e-05,
+      "loss": 0.2522,
+      "step": 29007
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288908280814847e-05,
+      "loss": 0.3573,
+      "step": 29008
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.928884743498968e-05,
+      "loss": 0.3676,
+      "step": 29009
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.928878658665741e-05,
+      "loss": 0.3148,
+      "step": 29010
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.928872573581805e-05,
+      "loss": 0.3468,
+      "step": 29011
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.928866488247162e-05,
+      "loss": 0.3636,
+      "step": 29012
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288604026618137e-05,
+      "loss": 0.3834,
+      "step": 29013
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288543168257614e-05,
+      "loss": 0.3373,
+      "step": 29014
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.928848230739007e-05,
+      "loss": 0.3192,
+      "step": 29015
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288421444015522e-05,
+      "loss": 0.3995,
+      "step": 29016
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288360578133987e-05,
+      "loss": 0.3007,
+      "step": 29017
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288299709745478e-05,
+      "loss": 0.3241,
+      "step": 29018
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288238838850016e-05,
+      "loss": 0.2954,
+      "step": 29019
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288177965447614e-05,
+      "loss": 0.2684,
+      "step": 29020
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288117089538287e-05,
+      "loss": 0.3315,
+      "step": 29021
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9288056211122056e-05,
+      "loss": 0.3041,
+      "step": 29022
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9287995330198934e-05,
+      "loss": 0.3309,
+      "step": 29023
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9287934446768943e-05,
+      "loss": 0.3492,
+      "step": 29024
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.928787356083209e-05,
+      "loss": 0.2859,
+      "step": 29025
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.92878126723884e-05,
+      "loss": 0.3142,
+      "step": 29026
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9287751781437884e-05,
+      "loss": 0.3101,
+      "step": 29027
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9287690887980557e-05,
+      "loss": 0.2919,
+      "step": 29028
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928762999201644e-05,
+      "loss": 0.3541,
+      "step": 29029
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9287569093545553e-05,
+      "loss": 0.2804,
+      "step": 29030
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9287508192567906e-05,
+      "loss": 0.2853,
+      "step": 29031
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9287447289083515e-05,
+      "loss": 0.2854,
+      "step": 29032
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9287386383092398e-05,
+      "loss": 0.2821,
+      "step": 29033
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9287325474594575e-05,
+      "loss": 0.2644,
+      "step": 29034
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928726456359006e-05,
+      "loss": 0.3041,
+      "step": 29035
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9287203650078866e-05,
+      "loss": 0.3039,
+      "step": 29036
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9287142734061014e-05,
+      "loss": 0.3812,
+      "step": 29037
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9287081815536517e-05,
+      "loss": 0.3392,
+      "step": 29038
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9287020894505395e-05,
+      "loss": 0.3417,
+      "step": 29039
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286959970967662e-05,
+      "loss": 0.3508,
+      "step": 29040
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286899044923335e-05,
+      "loss": 0.2967,
+      "step": 29041
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928683811637243e-05,
+      "loss": 0.2712,
+      "step": 29042
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286777185314966e-05,
+      "loss": 0.3848,
+      "step": 29043
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286716251750955e-05,
+      "loss": 0.313,
+      "step": 29044
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286655315680417e-05,
+      "loss": 0.3113,
+      "step": 29045
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286594377103367e-05,
+      "loss": 0.3426,
+      "step": 29046
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286533436019824e-05,
+      "loss": 0.3343,
+      "step": 29047
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.92864724924298e-05,
+      "loss": 0.305,
+      "step": 29048
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286411546333316e-05,
+      "loss": 0.263,
+      "step": 29049
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928635059773038e-05,
+      "loss": 0.281,
+      "step": 29050
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928628964662102e-05,
+      "loss": 0.3087,
+      "step": 29051
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286228693005246e-05,
+      "loss": 0.3408,
+      "step": 29052
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286167736883073e-05,
+      "loss": 0.3609,
+      "step": 29053
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9286106778254524e-05,
+      "loss": 0.3094,
+      "step": 29054
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928604581711961e-05,
+      "loss": 0.349,
+      "step": 29055
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285984853478347e-05,
+      "loss": 0.276,
+      "step": 29056
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285923887330754e-05,
+      "loss": 0.3162,
+      "step": 29057
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285862918676848e-05,
+      "loss": 0.3238,
+      "step": 29058
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285801947516642e-05,
+      "loss": 0.3631,
+      "step": 29059
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285740973850154e-05,
+      "loss": 0.3464,
+      "step": 29060
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285679997677408e-05,
+      "loss": 0.3552,
+      "step": 29061
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285619018998406e-05,
+      "loss": 0.2795,
+      "step": 29062
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285558037813176e-05,
+      "loss": 0.2859,
+      "step": 29063
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285497054121728e-05,
+      "loss": 0.2661,
+      "step": 29064
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285436067924082e-05,
+      "loss": 0.3542,
+      "step": 29065
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285375079220256e-05,
+      "loss": 0.3355,
+      "step": 29066
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285314088010262e-05,
+      "loss": 0.268,
+      "step": 29067
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285253094294118e-05,
+      "loss": 0.3347,
+      "step": 29068
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285192098071837e-05,
+      "loss": 0.3198,
+      "step": 29069
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9285131099343444e-05,
+      "loss": 0.2785,
+      "step": 29070
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928507009810895e-05,
+      "loss": 0.3103,
+      "step": 29071
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928500909436837e-05,
+      "loss": 0.2546,
+      "step": 29072
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9284948088121727e-05,
+      "loss": 0.3802,
+      "step": 29073
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9284887079369027e-05,
+      "loss": 0.3298,
+      "step": 29074
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.92848260681103e-05,
+      "loss": 0.3168,
+      "step": 29075
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928476505434555e-05,
+      "loss": 0.2916,
+      "step": 29076
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.92847040380748e-05,
+      "loss": 0.3076,
+      "step": 29077
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9284643019298063e-05,
+      "loss": 0.3011,
+      "step": 29078
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928458199801536e-05,
+      "loss": 0.3578,
+      "step": 29079
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9284520974226704e-05,
+      "loss": 0.3402,
+      "step": 29080
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928445994793211e-05,
+      "loss": 0.3221,
+      "step": 29081
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9284398919131598e-05,
+      "loss": 0.2816,
+      "step": 29082
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9284337887825185e-05,
+      "loss": 0.2776,
+      "step": 29083
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928427685401289e-05,
+      "loss": 0.3088,
+      "step": 29084
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928421581769472e-05,
+      "loss": 0.2627,
+      "step": 29085
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9284154778870693e-05,
+      "loss": 0.2431,
+      "step": 29086
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9284093737540835e-05,
+      "loss": 0.3408,
+      "step": 29087
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928403269370516e-05,
+      "loss": 0.2975,
+      "step": 29088
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9283971647363674e-05,
+      "loss": 0.3518,
+      "step": 29089
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9283910598516402e-05,
+      "loss": 0.3016,
+      "step": 29090
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928384954716336e-05,
+      "loss": 0.2721,
+      "step": 29091
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9283788493304562e-05,
+      "loss": 0.3164,
+      "step": 29092
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928372743694003e-05,
+      "loss": 0.3314,
+      "step": 29093
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9283666378069775e-05,
+      "loss": 0.3706,
+      "step": 29094
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928360531669381e-05,
+      "loss": 0.2898,
+      "step": 29095
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928354425281216e-05,
+      "loss": 0.3763,
+      "step": 29096
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9283483186424843e-05,
+      "loss": 0.3231,
+      "step": 29097
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928342211753187e-05,
+      "loss": 0.3618,
+      "step": 29098
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9283361046133252e-05,
+      "loss": 0.3512,
+      "step": 29099
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9283299972229014e-05,
+      "loss": 0.2994,
+      "step": 29100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928323889581917e-05,
+      "loss": 0.2951,
+      "step": 29101
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9283177816903737e-05,
+      "loss": 0.2678,
+      "step": 29102
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928311673548273e-05,
+      "loss": 0.2753,
+      "step": 29103
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9283055651556166e-05,
+      "loss": 0.4027,
+      "step": 29104
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9282994565124064e-05,
+      "loss": 0.3505,
+      "step": 29105
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9282933476186434e-05,
+      "loss": 0.3173,
+      "step": 29106
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.92828723847433e-05,
+      "loss": 0.3164,
+      "step": 29107
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9282811290794676e-05,
+      "loss": 0.2922,
+      "step": 29108
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9282750194340577e-05,
+      "loss": 0.3835,
+      "step": 29109
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928268909538102e-05,
+      "loss": 0.3015,
+      "step": 29110
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928262799391602e-05,
+      "loss": 0.3289,
+      "step": 29111
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.92825668899456e-05,
+      "loss": 0.3176,
+      "step": 29112
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928250578346977e-05,
+      "loss": 0.264,
+      "step": 29113
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928244467448855e-05,
+      "loss": 0.2959,
+      "step": 29114
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928238356300195e-05,
+      "loss": 0.3217,
+      "step": 29115
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928232244900999e-05,
+      "loss": 0.3057,
+      "step": 29116
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9282261332512694e-05,
+      "loss": 0.2886,
+      "step": 29117
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928220021351007e-05,
+      "loss": 0.3182,
+      "step": 29118
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9282139092002137e-05,
+      "loss": 0.3128,
+      "step": 29119
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928207796798891e-05,
+      "loss": 0.2563,
+      "step": 29120
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9282016841470407e-05,
+      "loss": 0.3786,
+      "step": 29121
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9281955712446645e-05,
+      "loss": 0.3199,
+      "step": 29122
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.928189458091764e-05,
+      "loss": 0.3001,
+      "step": 29123
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9281833446883408e-05,
+      "loss": 0.3369,
+      "step": 29124
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9281772310343963e-05,
+      "loss": 0.2909,
+      "step": 29125
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9281711171299327e-05,
+      "loss": 0.3016,
+      "step": 29126
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9281650029749513e-05,
+      "loss": 0.2731,
+      "step": 29127
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.928158888569454e-05,
+      "loss": 0.2998,
+      "step": 29128
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9281527739134423e-05,
+      "loss": 0.3209,
+      "step": 29129
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9281466590069174e-05,
+      "loss": 0.2786,
+      "step": 29130
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9281405438498818e-05,
+      "loss": 0.2614,
+      "step": 29131
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.928134428442337e-05,
+      "loss": 0.3043,
+      "step": 29132
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9281283127842837e-05,
+      "loss": 0.3741,
+      "step": 29133
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9281221968757245e-05,
+      "loss": 0.329,
+      "step": 29134
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9281160807166608e-05,
+      "loss": 0.3496,
+      "step": 29135
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9281099643070945e-05,
+      "loss": 0.2846,
+      "step": 29136
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9281038476470264e-05,
+      "loss": 0.3479,
+      "step": 29137
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.928097730736459e-05,
+      "loss": 0.3398,
+      "step": 29138
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280916135753937e-05,
+      "loss": 0.3416,
+      "step": 29139
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280854961638322e-05,
+      "loss": 0.3248,
+      "step": 29140
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280793785017763e-05,
+      "loss": 0.3087,
+      "step": 29141
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.928073260589227e-05,
+      "loss": 0.3125,
+      "step": 29142
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280671424261867e-05,
+      "loss": 0.3034,
+      "step": 29143
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280610240126563e-05,
+      "loss": 0.327,
+      "step": 29144
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280549053486384e-05,
+      "loss": 0.3419,
+      "step": 29145
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.928048786434134e-05,
+      "loss": 0.3195,
+      "step": 29146
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280426672691446e-05,
+      "loss": 0.3627,
+      "step": 29147
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280365478536725e-05,
+      "loss": 0.3048,
+      "step": 29148
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.928030428187719e-05,
+      "loss": 0.2765,
+      "step": 29149
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280243082712855e-05,
+      "loss": 0.3454,
+      "step": 29150
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.928018188104374e-05,
+      "loss": 0.3401,
+      "step": 29151
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280120676869862e-05,
+      "loss": 0.3134,
+      "step": 29152
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9280059470191235e-05,
+      "loss": 0.329,
+      "step": 29153
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279998261007876e-05,
+      "loss": 0.3096,
+      "step": 29154
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.92799370493198e-05,
+      "loss": 0.3463,
+      "step": 29155
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927987583512703e-05,
+      "loss": 0.3213,
+      "step": 29156
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279814618429576e-05,
+      "loss": 0.2997,
+      "step": 29157
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279753399227458e-05,
+      "loss": 0.2959,
+      "step": 29158
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927969217752069e-05,
+      "loss": 0.3459,
+      "step": 29159
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279630953309286e-05,
+      "loss": 0.2715,
+      "step": 29160
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279569726593274e-05,
+      "loss": 0.3873,
+      "step": 29161
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279508497372657e-05,
+      "loss": 0.3619,
+      "step": 29162
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927944726564746e-05,
+      "loss": 0.2824,
+      "step": 29163
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279386031417698e-05,
+      "loss": 0.3264,
+      "step": 29164
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279324794683383e-05,
+      "loss": 0.2988,
+      "step": 29165
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279263555444534e-05,
+      "loss": 0.3089,
+      "step": 29166
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927920231370117e-05,
+      "loss": 0.2403,
+      "step": 29167
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927914106945331e-05,
+      "loss": 0.3401,
+      "step": 29168
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279079822700963e-05,
+      "loss": 0.271,
+      "step": 29169
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9279018573444147e-05,
+      "loss": 0.3454,
+      "step": 29170
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927895732168288e-05,
+      "loss": 0.2982,
+      "step": 29171
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9278896067417184e-05,
+      "loss": 0.2968,
+      "step": 29172
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9278834810647067e-05,
+      "loss": 0.3149,
+      "step": 29173
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9278773551372548e-05,
+      "loss": 0.3159,
+      "step": 29174
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9278712289593648e-05,
+      "loss": 0.2946,
+      "step": 29175
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927865102531038e-05,
+      "loss": 0.3305,
+      "step": 29176
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927858975852276e-05,
+      "loss": 0.3393,
+      "step": 29177
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9278528489230806e-05,
+      "loss": 0.3327,
+      "step": 29178
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927846721743453e-05,
+      "loss": 0.3481,
+      "step": 29179
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9278405943133957e-05,
+      "loss": 0.403,
+      "step": 29180
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9278344666329095e-05,
+      "loss": 0.3105,
+      "step": 29181
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9278283387019967e-05,
+      "loss": 0.3,
+      "step": 29182
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927822210520659e-05,
+      "loss": 0.2647,
+      "step": 29183
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927816082088897e-05,
+      "loss": 0.3458,
+      "step": 29184
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9278099534067133e-05,
+      "loss": 0.2934,
+      "step": 29185
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9278038244741095e-05,
+      "loss": 0.3819,
+      "step": 29186
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927797695291087e-05,
+      "loss": 0.3605,
+      "step": 29187
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927791565857648e-05,
+      "loss": 0.2814,
+      "step": 29188
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9277854361737934e-05,
+      "loss": 0.2725,
+      "step": 29189
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927779306239525e-05,
+      "loss": 0.3228,
+      "step": 29190
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927773176054845e-05,
+      "loss": 0.2645,
+      "step": 29191
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9277670456197547e-05,
+      "loss": 0.3751,
+      "step": 29192
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9277609149342552e-05,
+      "loss": 0.298,
+      "step": 29193
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9277547839983493e-05,
+      "loss": 0.291,
+      "step": 29194
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927748652812038e-05,
+      "loss": 0.4099,
+      "step": 29195
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9277425213753224e-05,
+      "loss": 0.2877,
+      "step": 29196
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9277363896882054e-05,
+      "loss": 0.3499,
+      "step": 29197
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9277302577506875e-05,
+      "loss": 0.2632,
+      "step": 29198
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927724125562771e-05,
+      "loss": 0.3531,
+      "step": 29199
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9277179931244578e-05,
+      "loss": 0.3404,
+      "step": 29200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927711860435749e-05,
+      "loss": 0.3278,
+      "step": 29201
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9277057274966462e-05,
+      "loss": 0.3631,
+      "step": 29202
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276995943071513e-05,
+      "loss": 0.3626,
+      "step": 29203
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276934608672663e-05,
+      "loss": 0.2972,
+      "step": 29204
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927687327176992e-05,
+      "loss": 0.2773,
+      "step": 29205
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927681193236331e-05,
+      "loss": 0.3652,
+      "step": 29206
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276750590452845e-05,
+      "loss": 0.3267,
+      "step": 29207
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927668924603854e-05,
+      "loss": 0.2731,
+      "step": 29208
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927662789912041e-05,
+      "loss": 0.2957,
+      "step": 29209
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276566549698482e-05,
+      "loss": 0.2721,
+      "step": 29210
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927650519777276e-05,
+      "loss": 0.3308,
+      "step": 29211
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276443843343267e-05,
+      "loss": 0.273,
+      "step": 29212
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276382486410016e-05,
+      "loss": 0.3468,
+      "step": 29213
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276321126973032e-05,
+      "loss": 0.3387,
+      "step": 29214
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276259765032323e-05,
+      "loss": 0.3152,
+      "step": 29215
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276198400587906e-05,
+      "loss": 0.2995,
+      "step": 29216
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276137033639803e-05,
+      "loss": 0.3492,
+      "step": 29217
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9276075664188022e-05,
+      "loss": 0.2925,
+      "step": 29218
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927601429223259e-05,
+      "loss": 0.286,
+      "step": 29219
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9275952917773516e-05,
+      "loss": 0.3224,
+      "step": 29220
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927589154081082e-05,
+      "loss": 0.3503,
+      "step": 29221
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9275830161344514e-05,
+      "loss": 0.2835,
+      "step": 29222
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927576877937462e-05,
+      "loss": 0.3369,
+      "step": 29223
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9275707394901158e-05,
+      "loss": 0.2853,
+      "step": 29224
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9275646007924134e-05,
+      "loss": 0.3254,
+      "step": 29225
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.927558461844357e-05,
+      "loss": 0.2695,
+      "step": 29226
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927552322645948e-05,
+      "loss": 0.3365,
+      "step": 29227
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927546183197189e-05,
+      "loss": 0.2912,
+      "step": 29228
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9275400434980806e-05,
+      "loss": 0.2905,
+      "step": 29229
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9275339035486247e-05,
+      "loss": 0.2878,
+      "step": 29230
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927527763348823e-05,
+      "loss": 0.2921,
+      "step": 29231
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9275216228986774e-05,
+      "loss": 0.3296,
+      "step": 29232
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9275154821981892e-05,
+      "loss": 0.3384,
+      "step": 29233
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9275093412473605e-05,
+      "loss": 0.304,
+      "step": 29234
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9275032000461923e-05,
+      "loss": 0.3269,
+      "step": 29235
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927497058594687e-05,
+      "loss": 0.3148,
+      "step": 29236
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927490916892846e-05,
+      "loss": 0.3282,
+      "step": 29237
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274847749406707e-05,
+      "loss": 0.2821,
+      "step": 29238
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274786327381628e-05,
+      "loss": 0.3376,
+      "step": 29239
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274724902853244e-05,
+      "loss": 0.2919,
+      "step": 29240
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274663475821564e-05,
+      "loss": 0.3111,
+      "step": 29241
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274602046286612e-05,
+      "loss": 0.2771,
+      "step": 29242
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.92745406142484e-05,
+      "loss": 0.3156,
+      "step": 29243
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274479179706946e-05,
+      "loss": 0.3538,
+      "step": 29244
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927441774266227e-05,
+      "loss": 0.3244,
+      "step": 29245
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927435630311438e-05,
+      "loss": 0.3786,
+      "step": 29246
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274294861063302e-05,
+      "loss": 0.2935,
+      "step": 29247
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274233416509047e-05,
+      "loss": 0.2742,
+      "step": 29248
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274171969451634e-05,
+      "loss": 0.2871,
+      "step": 29249
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274110519891076e-05,
+      "loss": 0.2958,
+      "step": 29250
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9274049067827396e-05,
+      "loss": 0.3187,
+      "step": 29251
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9273987613260602e-05,
+      "loss": 0.3037,
+      "step": 29252
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927392615619072e-05,
+      "loss": 0.3071,
+      "step": 29253
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927386469661776e-05,
+      "loss": 0.329,
+      "step": 29254
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9273803234541742e-05,
+      "loss": 0.2991,
+      "step": 29255
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927374176996268e-05,
+      "loss": 0.2965,
+      "step": 29256
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9273680302880595e-05,
+      "loss": 0.3176,
+      "step": 29257
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9273618833295493e-05,
+      "loss": 0.3678,
+      "step": 29258
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9273557361207402e-05,
+      "loss": 0.3372,
+      "step": 29259
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9273495886616336e-05,
+      "loss": 0.2799,
+      "step": 29260
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927343440952231e-05,
+      "loss": 0.3375,
+      "step": 29261
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9273372929925337e-05,
+      "loss": 0.2691,
+      "step": 29262
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927331144782544e-05,
+      "loss": 0.3203,
+      "step": 29263
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927324996322263e-05,
+      "loss": 0.2948,
+      "step": 29264
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927318847611693e-05,
+      "loss": 0.3021,
+      "step": 29265
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927312698650835e-05,
+      "loss": 0.2898,
+      "step": 29266
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9273065494396913e-05,
+      "loss": 0.3149,
+      "step": 29267
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9273003999782632e-05,
+      "loss": 0.2955,
+      "step": 29268
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272942502665524e-05,
+      "loss": 0.3125,
+      "step": 29269
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272881003045606e-05,
+      "loss": 0.2961,
+      "step": 29270
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272819500922893e-05,
+      "loss": 0.3905,
+      "step": 29271
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.92727579962974e-05,
+      "loss": 0.3733,
+      "step": 29272
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927269648916915e-05,
+      "loss": 0.3643,
+      "step": 29273
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272634979538152e-05,
+      "loss": 0.2982,
+      "step": 29274
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927257346740443e-05,
+      "loss": 0.3136,
+      "step": 29275
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272511952767994e-05,
+      "loss": 0.3229,
+      "step": 29276
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272450435628867e-05,
+      "loss": 0.3236,
+      "step": 29277
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272388915987063e-05,
+      "loss": 0.292,
+      "step": 29278
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272327393842594e-05,
+      "loss": 0.2949,
+      "step": 29279
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272265869195482e-05,
+      "loss": 0.3079,
+      "step": 29280
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927220434204574e-05,
+      "loss": 0.2828,
+      "step": 29281
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272142812393392e-05,
+      "loss": 0.279,
+      "step": 29282
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9272081280238444e-05,
+      "loss": 0.3434,
+      "step": 29283
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927201974558092e-05,
+      "loss": 0.3786,
+      "step": 29284
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271958208420835e-05,
+      "loss": 0.322,
+      "step": 29285
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271896668758204e-05,
+      "loss": 0.3239,
+      "step": 29286
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271835126593048e-05,
+      "loss": 0.2917,
+      "step": 29287
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271773581925377e-05,
+      "loss": 0.3353,
+      "step": 29288
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271712034755212e-05,
+      "loss": 0.2893,
+      "step": 29289
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927165048508257e-05,
+      "loss": 0.2893,
+      "step": 29290
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271588932907462e-05,
+      "loss": 0.305,
+      "step": 29291
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271527378229912e-05,
+      "loss": 0.3131,
+      "step": 29292
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271465821049935e-05,
+      "loss": 0.2801,
+      "step": 29293
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271404261367544e-05,
+      "loss": 0.3039,
+      "step": 29294
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271342699182755e-05,
+      "loss": 0.3437,
+      "step": 29295
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271281134495592e-05,
+      "loss": 0.334,
+      "step": 29296
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271219567306064e-05,
+      "loss": 0.3414,
+      "step": 29297
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927115799761419e-05,
+      "loss": 0.3488,
+      "step": 29298
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271096425419992e-05,
+      "loss": 0.297,
+      "step": 29299
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9271034850723475e-05,
+      "loss": 0.3349,
+      "step": 29300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9270973273524667e-05,
+      "loss": 0.3256,
+      "step": 29301
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9270911693823578e-05,
+      "loss": 0.2943,
+      "step": 29302
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927085011162023e-05,
+      "loss": 0.3033,
+      "step": 29303
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927078852691463e-05,
+      "loss": 0.3471,
+      "step": 29304
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9270726939706805e-05,
+      "loss": 0.3173,
+      "step": 29305
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9270665349996768e-05,
+      "loss": 0.3705,
+      "step": 29306
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9270603757784534e-05,
+      "loss": 0.2944,
+      "step": 29307
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927054216307012e-05,
+      "loss": 0.3232,
+      "step": 29308
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9270480565853543e-05,
+      "loss": 0.3446,
+      "step": 29309
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927041896613482e-05,
+      "loss": 0.317,
+      "step": 29310
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.927035736391397e-05,
+      "loss": 0.3272,
+      "step": 29311
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9270295759191003e-05,
+      "loss": 0.3634,
+      "step": 29312
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9270234151965944e-05,
+      "loss": 0.304,
+      "step": 29313
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9270172542238804e-05,
+      "loss": 0.3054,
+      "step": 29314
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.92701109300096e-05,
+      "loss": 0.3361,
+      "step": 29315
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9270049315278352e-05,
+      "loss": 0.3149,
+      "step": 29316
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.926998769804507e-05,
+      "loss": 0.2832,
+      "step": 29317
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.926992607830978e-05,
+      "loss": 0.2978,
+      "step": 29318
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.926986445607249e-05,
+      "loss": 0.2986,
+      "step": 29319
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.926980283133322e-05,
+      "loss": 0.3024,
+      "step": 29320
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.926974120409199e-05,
+      "loss": 0.3267,
+      "step": 29321
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9269679574348815e-05,
+      "loss": 0.2974,
+      "step": 29322
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9269617942103708e-05,
+      "loss": 0.2869,
+      "step": 29323
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9269556307356688e-05,
+      "loss": 0.2726,
+      "step": 29324
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9269494670107767e-05,
+      "loss": 0.3337,
+      "step": 29325
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9269433030356972e-05,
+      "loss": 0.306,
+      "step": 29326
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926937138810431e-05,
+      "loss": 0.3232,
+      "step": 29327
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9269309743349806e-05,
+      "loss": 0.2943,
+      "step": 29328
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9269248096093465e-05,
+      "loss": 0.3172,
+      "step": 29329
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9269186446335314e-05,
+      "loss": 0.277,
+      "step": 29330
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9269124794075367e-05,
+      "loss": 0.3078,
+      "step": 29331
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926906313931364e-05,
+      "loss": 0.3394,
+      "step": 29332
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926900148205015e-05,
+      "loss": 0.3143,
+      "step": 29333
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268939822284912e-05,
+      "loss": 0.3178,
+      "step": 29334
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926887816001794e-05,
+      "loss": 0.3166,
+      "step": 29335
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268816495249263e-05,
+      "loss": 0.3544,
+      "step": 29336
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268754827978885e-05,
+      "loss": 0.3825,
+      "step": 29337
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268693158206823e-05,
+      "loss": 0.3084,
+      "step": 29338
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268631485933104e-05,
+      "loss": 0.3637,
+      "step": 29339
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926856981115773e-05,
+      "loss": 0.2863,
+      "step": 29340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268508133880732e-05,
+      "loss": 0.2824,
+      "step": 29341
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926844645410212e-05,
+      "loss": 0.3155,
+      "step": 29342
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926838477182191e-05,
+      "loss": 0.3024,
+      "step": 29343
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926832308704012e-05,
+      "loss": 0.3337,
+      "step": 29344
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268261399756762e-05,
+      "loss": 0.3384,
+      "step": 29345
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268199709971863e-05,
+      "loss": 0.3019,
+      "step": 29346
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268138017685428e-05,
+      "loss": 0.3155,
+      "step": 29347
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268076322897484e-05,
+      "loss": 0.3182,
+      "step": 29348
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9268014625608043e-05,
+      "loss": 0.2661,
+      "step": 29349
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267952925817116e-05,
+      "loss": 0.2511,
+      "step": 29350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926789122352473e-05,
+      "loss": 0.3089,
+      "step": 29351
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267829518730896e-05,
+      "loss": 0.3319,
+      "step": 29352
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926776781143563e-05,
+      "loss": 0.3226,
+      "step": 29353
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267706101638954e-05,
+      "loss": 0.3369,
+      "step": 29354
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267644389340878e-05,
+      "loss": 0.3243,
+      "step": 29355
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267582674541422e-05,
+      "loss": 0.3222,
+      "step": 29356
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.92675209572406e-05,
+      "loss": 0.3556,
+      "step": 29357
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267459237438432e-05,
+      "loss": 0.3279,
+      "step": 29358
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267397515134935e-05,
+      "loss": 0.3215,
+      "step": 29359
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267335790330122e-05,
+      "loss": 0.3009,
+      "step": 29360
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926727406302401e-05,
+      "loss": 0.3445,
+      "step": 29361
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267212333216626e-05,
+      "loss": 0.2692,
+      "step": 29362
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267150600907972e-05,
+      "loss": 0.2952,
+      "step": 29363
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926708886609807e-05,
+      "loss": 0.2963,
+      "step": 29364
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9267027128786936e-05,
+      "loss": 0.3284,
+      "step": 29365
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266965388974593e-05,
+      "loss": 0.2997,
+      "step": 29366
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266903646661047e-05,
+      "loss": 0.3085,
+      "step": 29367
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266841901846325e-05,
+      "loss": 0.3004,
+      "step": 29368
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266780154530436e-05,
+      "loss": 0.2779,
+      "step": 29369
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266718404713405e-05,
+      "loss": 0.2724,
+      "step": 29370
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266656652395238e-05,
+      "loss": 0.3068,
+      "step": 29371
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926659489757596e-05,
+      "loss": 0.2934,
+      "step": 29372
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266533140255584e-05,
+      "loss": 0.3517,
+      "step": 29373
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266471380434125e-05,
+      "loss": 0.3071,
+      "step": 29374
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266409618111604e-05,
+      "loss": 0.3307,
+      "step": 29375
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266347853288036e-05,
+      "loss": 0.2936,
+      "step": 29376
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266286085963436e-05,
+      "loss": 0.2915,
+      "step": 29377
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266224316137823e-05,
+      "loss": 0.2982,
+      "step": 29378
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266162543811213e-05,
+      "loss": 0.2635,
+      "step": 29379
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9266100768983623e-05,
+      "loss": 0.3412,
+      "step": 29380
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926603899165507e-05,
+      "loss": 0.2973,
+      "step": 29381
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9265977211825568e-05,
+      "loss": 0.2996,
+      "step": 29382
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9265915429495136e-05,
+      "loss": 0.3202,
+      "step": 29383
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926585364466379e-05,
+      "loss": 0.3552,
+      "step": 29384
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9265791857331546e-05,
+      "loss": 0.3148,
+      "step": 29385
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9265730067498422e-05,
+      "loss": 0.3284,
+      "step": 29386
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9265668275164433e-05,
+      "loss": 0.3147,
+      "step": 29387
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9265606480329596e-05,
+      "loss": 0.2814,
+      "step": 29388
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9265544682993932e-05,
+      "loss": 0.3034,
+      "step": 29389
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926548288315745e-05,
+      "loss": 0.2821,
+      "step": 29390
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9265421080820178e-05,
+      "loss": 0.2664,
+      "step": 29391
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926535927598212e-05,
+      "loss": 0.3478,
+      "step": 29392
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9265297468643298e-05,
+      "loss": 0.2473,
+      "step": 29393
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926523565880373e-05,
+      "loss": 0.3504,
+      "step": 29394
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9265173846463432e-05,
+      "loss": 0.2833,
+      "step": 29395
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926511203162242e-05,
+      "loss": 0.2823,
+      "step": 29396
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926505021428071e-05,
+      "loss": 0.3092,
+      "step": 29397
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926498839443832e-05,
+      "loss": 0.2924,
+      "step": 29398
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926492657209527e-05,
+      "loss": 0.3367,
+      "step": 29399
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926486474725157e-05,
+      "loss": 0.3297,
+      "step": 29400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9264802919907238e-05,
+      "loss": 0.3224,
+      "step": 29401
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9264741090062294e-05,
+      "loss": 0.3107,
+      "step": 29402
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926467925771675e-05,
+      "loss": 0.2949,
+      "step": 29403
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926461742287063e-05,
+      "loss": 0.3176,
+      "step": 29404
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9264555585523944e-05,
+      "loss": 0.3237,
+      "step": 29405
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9264493745676716e-05,
+      "loss": 0.266,
+      "step": 29406
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9264431903328952e-05,
+      "loss": 0.3053,
+      "step": 29407
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926437005848068e-05,
+      "loss": 0.2807,
+      "step": 29408
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9264308211131907e-05,
+      "loss": 0.3205,
+      "step": 29409
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9264246361282652e-05,
+      "loss": 0.2693,
+      "step": 29410
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926418450893294e-05,
+      "loss": 0.3577,
+      "step": 29411
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9264122654082776e-05,
+      "loss": 0.2934,
+      "step": 29412
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9264060796732185e-05,
+      "loss": 0.3448,
+      "step": 29413
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926399893688118e-05,
+      "loss": 0.3207,
+      "step": 29414
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9263937074529777e-05,
+      "loss": 0.2855,
+      "step": 29415
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9263875209677994e-05,
+      "loss": 0.3866,
+      "step": 29416
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9263813342325847e-05,
+      "loss": 0.2912,
+      "step": 29417
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9263751472473354e-05,
+      "loss": 0.3218,
+      "step": 29418
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9263689600120536e-05,
+      "loss": 0.2871,
+      "step": 29419
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.92636277252674e-05,
+      "loss": 0.3665,
+      "step": 29420
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.926356584791397e-05,
+      "loss": 0.2716,
+      "step": 29421
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9263503968060258e-05,
+      "loss": 0.2887,
+      "step": 29422
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9263442085706284e-05,
+      "loss": 0.3122,
+      "step": 29423
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9263380200852065e-05,
+      "loss": 0.3385,
+      "step": 29424
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9263318313497615e-05,
+      "loss": 0.3459,
+      "step": 29425
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926325642364295e-05,
+      "loss": 0.3149,
+      "step": 29426
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9263194531288093e-05,
+      "loss": 0.2954,
+      "step": 29427
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926313263643305e-05,
+      "loss": 0.3276,
+      "step": 29428
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9263070739077853e-05,
+      "loss": 0.2826,
+      "step": 29429
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9263008839222504e-05,
+      "loss": 0.322,
+      "step": 29430
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262946936867026e-05,
+      "loss": 0.3831,
+      "step": 29431
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262885032011436e-05,
+      "loss": 0.2802,
+      "step": 29432
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262823124655753e-05,
+      "loss": 0.3448,
+      "step": 29433
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262761214799988e-05,
+      "loss": 0.3342,
+      "step": 29434
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262699302444157e-05,
+      "loss": 0.3631,
+      "step": 29435
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262637387588286e-05,
+      "loss": 0.3145,
+      "step": 29436
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262575470232386e-05,
+      "loss": 0.2948,
+      "step": 29437
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926251355037647e-05,
+      "loss": 0.2569,
+      "step": 29438
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926245162802056e-05,
+      "loss": 0.3058,
+      "step": 29439
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926238970316467e-05,
+      "loss": 0.3878,
+      "step": 29440
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262327775808817e-05,
+      "loss": 0.282,
+      "step": 29441
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262265845953022e-05,
+      "loss": 0.2884,
+      "step": 29442
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262203913597294e-05,
+      "loss": 0.3632,
+      "step": 29443
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262141978741656e-05,
+      "loss": 0.3165,
+      "step": 29444
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9262080041386123e-05,
+      "loss": 0.3075,
+      "step": 29445
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926201810153071e-05,
+      "loss": 0.3064,
+      "step": 29446
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261956159175437e-05,
+      "loss": 0.3598,
+      "step": 29447
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261894214320315e-05,
+      "loss": 0.2997,
+      "step": 29448
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261832266965368e-05,
+      "loss": 0.3235,
+      "step": 29449
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926177031711061e-05,
+      "loss": 0.296,
+      "step": 29450
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261708364756054e-05,
+      "loss": 0.3302,
+      "step": 29451
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926164640990172e-05,
+      "loss": 0.2996,
+      "step": 29452
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261584452547625e-05,
+      "loss": 0.3224,
+      "step": 29453
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261522492693785e-05,
+      "loss": 0.3342,
+      "step": 29454
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261460530340216e-05,
+      "loss": 0.2657,
+      "step": 29455
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261398565486937e-05,
+      "loss": 0.3068,
+      "step": 29456
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261336598133962e-05,
+      "loss": 0.3009,
+      "step": 29457
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926127462828131e-05,
+      "loss": 0.3349,
+      "step": 29458
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261212655928996e-05,
+      "loss": 0.3366,
+      "step": 29459
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926115068107704e-05,
+      "loss": 0.3076,
+      "step": 29460
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926108870372545e-05,
+      "loss": 0.2698,
+      "step": 29461
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9261026723874254e-05,
+      "loss": 0.3255,
+      "step": 29462
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260964741523463e-05,
+      "loss": 0.3011,
+      "step": 29463
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260902756673095e-05,
+      "loss": 0.3041,
+      "step": 29464
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260840769323163e-05,
+      "loss": 0.3592,
+      "step": 29465
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260778779473688e-05,
+      "loss": 0.3355,
+      "step": 29466
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260716787124686e-05,
+      "loss": 0.3466,
+      "step": 29467
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926065479227617e-05,
+      "loss": 0.2669,
+      "step": 29468
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260592794928164e-05,
+      "loss": 0.307,
+      "step": 29469
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260530795080685e-05,
+      "loss": 0.3372,
+      "step": 29470
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260468792733736e-05,
+      "loss": 0.3786,
+      "step": 29471
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926040678788735e-05,
+      "loss": 0.3423,
+      "step": 29472
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260344780541535e-05,
+      "loss": 0.3393,
+      "step": 29473
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260282770696305e-05,
+      "loss": 0.3257,
+      "step": 29474
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260220758351688e-05,
+      "loss": 0.3164,
+      "step": 29475
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926015874350769e-05,
+      "loss": 0.2819,
+      "step": 29476
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260096726164335e-05,
+      "loss": 0.3056,
+      "step": 29477
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9260034706321633e-05,
+      "loss": 0.2868,
+      "step": 29478
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259972683979605e-05,
+      "loss": 0.3215,
+      "step": 29479
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259910659138274e-05,
+      "loss": 0.3166,
+      "step": 29480
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259848631797643e-05,
+      "loss": 0.3516,
+      "step": 29481
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259786601957737e-05,
+      "loss": 0.3094,
+      "step": 29482
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925972456961857e-05,
+      "loss": 0.3047,
+      "step": 29483
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925966253478016e-05,
+      "loss": 0.338,
+      "step": 29484
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259600497442525e-05,
+      "loss": 0.3193,
+      "step": 29485
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259538457605682e-05,
+      "loss": 0.2945,
+      "step": 29486
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259476415269644e-05,
+      "loss": 0.3097,
+      "step": 29487
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259414370434432e-05,
+      "loss": 0.3164,
+      "step": 29488
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925935232310006e-05,
+      "loss": 0.3245,
+      "step": 29489
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259290273266544e-05,
+      "loss": 0.3302,
+      "step": 29490
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259228220933904e-05,
+      "loss": 0.3507,
+      "step": 29491
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259166166102155e-05,
+      "loss": 0.3016,
+      "step": 29492
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925910410877131e-05,
+      "loss": 0.3063,
+      "step": 29493
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9259042048941396e-05,
+      "loss": 0.3216,
+      "step": 29494
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925897998661242e-05,
+      "loss": 0.3338,
+      "step": 29495
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.92589179217844e-05,
+      "loss": 0.3914,
+      "step": 29496
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258855854457357e-05,
+      "loss": 0.317,
+      "step": 29497
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258793784631305e-05,
+      "loss": 0.3235,
+      "step": 29498
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258731712306266e-05,
+      "loss": 0.3153,
+      "step": 29499
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258669637482245e-05,
+      "loss": 0.2931,
+      "step": 29500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258607560159268e-05,
+      "loss": 0.338,
+      "step": 29501
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925854548033735e-05,
+      "loss": 0.376,
+      "step": 29502
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258483398016508e-05,
+      "loss": 0.3157,
+      "step": 29503
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258421313196757e-05,
+      "loss": 0.3371,
+      "step": 29504
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258359225878113e-05,
+      "loss": 0.3052,
+      "step": 29505
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.92582971360606e-05,
+      "loss": 0.3113,
+      "step": 29506
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258235043744224e-05,
+      "loss": 0.3376,
+      "step": 29507
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258172948929006e-05,
+      "loss": 0.2792,
+      "step": 29508
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925811085161497e-05,
+      "loss": 0.3552,
+      "step": 29509
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9258048751802123e-05,
+      "loss": 0.2899,
+      "step": 29510
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9257986649490487e-05,
+      "loss": 0.3391,
+      "step": 29511
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9257924544680078e-05,
+      "loss": 0.3722,
+      "step": 29512
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9257862437370907e-05,
+      "loss": 0.3155,
+      "step": 29513
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9257800327563e-05,
+      "loss": 0.3316,
+      "step": 29514
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925773821525637e-05,
+      "loss": 0.3228,
+      "step": 29515
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925767610045103e-05,
+      "loss": 0.3597,
+      "step": 29516
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9257613983147e-05,
+      "loss": 0.345,
+      "step": 29517
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.92575518633443e-05,
+      "loss": 0.3402,
+      "step": 29518
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925748974104294e-05,
+      "loss": 0.2854,
+      "step": 29519
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.925742761624294e-05,
+      "loss": 0.3051,
+      "step": 29520
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9257365488944318e-05,
+      "loss": 0.3822,
+      "step": 29521
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9257303359147093e-05,
+      "loss": 0.274,
+      "step": 29522
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9257241226851274e-05,
+      "loss": 0.314,
+      "step": 29523
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9257179092056884e-05,
+      "loss": 0.3585,
+      "step": 29524
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925711695476394e-05,
+      "loss": 0.3635,
+      "step": 29525
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9257054814972455e-05,
+      "loss": 0.3285,
+      "step": 29526
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9256992672682447e-05,
+      "loss": 0.2794,
+      "step": 29527
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9256930527893932e-05,
+      "loss": 0.3519,
+      "step": 29528
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925686838060693e-05,
+      "loss": 0.3101,
+      "step": 29529
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925680623082146e-05,
+      "loss": 0.3241,
+      "step": 29530
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925674407853753e-05,
+      "loss": 0.3484,
+      "step": 29531
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925668192375516e-05,
+      "loss": 0.2712,
+      "step": 29532
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9256619766474374e-05,
+      "loss": 0.3538,
+      "step": 29533
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9256557606695182e-05,
+      "loss": 0.3415,
+      "step": 29534
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.92564954444176e-05,
+      "loss": 0.2881,
+      "step": 29535
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9256433279641644e-05,
+      "loss": 0.3496,
+      "step": 29536
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9256371112367335e-05,
+      "loss": 0.3025,
+      "step": 29537
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9256308942594692e-05,
+      "loss": 0.2881,
+      "step": 29538
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9256246770323722e-05,
+      "loss": 0.3673,
+      "step": 29539
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925618459555445e-05,
+      "loss": 0.3345,
+      "step": 29540
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9256122418286893e-05,
+      "loss": 0.3181,
+      "step": 29541
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9256060238521063e-05,
+      "loss": 0.3253,
+      "step": 29542
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925599805625698e-05,
+      "loss": 0.2703,
+      "step": 29543
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255935871494658e-05,
+      "loss": 0.3692,
+      "step": 29544
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255873684234115e-05,
+      "loss": 0.3061,
+      "step": 29545
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255811494475374e-05,
+      "loss": 0.3339,
+      "step": 29546
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255749302218442e-05,
+      "loss": 0.2726,
+      "step": 29547
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925568710746334e-05,
+      "loss": 0.389,
+      "step": 29548
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255624910210087e-05,
+      "loss": 0.3112,
+      "step": 29549
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255562710458695e-05,
+      "loss": 0.2999,
+      "step": 29550
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255500508209183e-05,
+      "loss": 0.3303,
+      "step": 29551
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925543830346157e-05,
+      "loss": 0.3178,
+      "step": 29552
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925537609621587e-05,
+      "loss": 0.3058,
+      "step": 29553
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255313886472103e-05,
+      "loss": 0.325,
+      "step": 29554
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925525167423028e-05,
+      "loss": 0.358,
+      "step": 29555
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255189459490422e-05,
+      "loss": 0.3034,
+      "step": 29556
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255127242252546e-05,
+      "loss": 0.3182,
+      "step": 29557
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925506502251667e-05,
+      "loss": 0.3409,
+      "step": 29558
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9255002800282807e-05,
+      "loss": 0.3176,
+      "step": 29559
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254940575550973e-05,
+      "loss": 0.3083,
+      "step": 29560
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254878348321192e-05,
+      "loss": 0.3098,
+      "step": 29561
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254816118593474e-05,
+      "loss": 0.3519,
+      "step": 29562
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254753886367835e-05,
+      "loss": 0.3264,
+      "step": 29563
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254691651644296e-05,
+      "loss": 0.3212,
+      "step": 29564
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254629414422873e-05,
+      "loss": 0.3618,
+      "step": 29565
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254567174703585e-05,
+      "loss": 0.2908,
+      "step": 29566
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254504932486443e-05,
+      "loss": 0.2923,
+      "step": 29567
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925444268777147e-05,
+      "loss": 0.3177,
+      "step": 29568
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254380440558677e-05,
+      "loss": 0.3289,
+      "step": 29569
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254318190848082e-05,
+      "loss": 0.3239,
+      "step": 29570
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254255938639706e-05,
+      "loss": 0.2876,
+      "step": 29571
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254193683933562e-05,
+      "loss": 0.1731,
+      "step": 29572
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254131426729667e-05,
+      "loss": 0.1478,
+      "step": 29573
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925406916702804e-05,
+      "loss": 0.1513,
+      "step": 29574
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9254006904828694e-05,
+      "loss": 0.1507,
+      "step": 29575
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925394464013165e-05,
+      "loss": 0.1697,
+      "step": 29576
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9253882372936927e-05,
+      "loss": 0.1484,
+      "step": 29577
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9253820103244534e-05,
+      "loss": 0.1414,
+      "step": 29578
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925375783105449e-05,
+      "loss": 0.1176,
+      "step": 29579
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9253695556366816e-05,
+      "loss": 0.1581,
+      "step": 29580
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9253633279181526e-05,
+      "loss": 0.1364,
+      "step": 29581
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9253570999498637e-05,
+      "loss": 0.1408,
+      "step": 29582
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9253508717318166e-05,
+      "loss": 0.1646,
+      "step": 29583
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925344643264013e-05,
+      "loss": 0.1418,
+      "step": 29584
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9253384145464546e-05,
+      "loss": 0.1261,
+      "step": 29585
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925332185579143e-05,
+      "loss": 0.1475,
+      "step": 29586
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9253259563620797e-05,
+      "loss": 0.1201,
+      "step": 29587
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925319726895267e-05,
+      "loss": 0.151,
+      "step": 29588
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925313497178706e-05,
+      "loss": 0.1412,
+      "step": 29589
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925307267212398e-05,
+      "loss": 0.1644,
+      "step": 29590
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925301036996346e-05,
+      "loss": 0.1364,
+      "step": 29591
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9252948065305506e-05,
+      "loss": 0.1498,
+      "step": 29592
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925288575815014e-05,
+      "loss": 0.1217,
+      "step": 29593
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9252823448497375e-05,
+      "loss": 0.129,
+      "step": 29594
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9252761136347232e-05,
+      "loss": 0.1295,
+      "step": 29595
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9252698821699724e-05,
+      "loss": 0.1445,
+      "step": 29596
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925263650455487e-05,
+      "loss": 0.1166,
+      "step": 29597
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9252574184912688e-05,
+      "loss": 0.135,
+      "step": 29598
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925251186277319e-05,
+      "loss": 0.1365,
+      "step": 29599
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9252449538136396e-05,
+      "loss": 0.1232,
+      "step": 29600
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925238721100232e-05,
+      "loss": 0.1298,
+      "step": 29601
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9252324881370986e-05,
+      "loss": 0.1604,
+      "step": 29602
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9252262549242405e-05,
+      "loss": 0.1564,
+      "step": 29603
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9252200214616592e-05,
+      "loss": 0.1273,
+      "step": 29604
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925213787749357e-05,
+      "loss": 0.1308,
+      "step": 29605
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9252075537873355e-05,
+      "loss": 0.1253,
+      "step": 29606
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925201319575596e-05,
+      "loss": 0.1395,
+      "step": 29607
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.92519508511414e-05,
+      "loss": 0.1189,
+      "step": 29608
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9251888504029698e-05,
+      "loss": 0.1763,
+      "step": 29609
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9251826154420868e-05,
+      "loss": 0.1321,
+      "step": 29610
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9251763802314927e-05,
+      "loss": 0.1148,
+      "step": 29611
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925170144771189e-05,
+      "loss": 0.1451,
+      "step": 29612
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9251639090611777e-05,
+      "loss": 0.1463,
+      "step": 29613
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9251576731014602e-05,
+      "loss": 0.1545,
+      "step": 29614
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9251514368920384e-05,
+      "loss": 0.1397,
+      "step": 29615
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.925145200432914e-05,
+      "loss": 0.1318,
+      "step": 29616
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9251389637240884e-05,
+      "loss": 0.1521,
+      "step": 29617
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9251327267655635e-05,
+      "loss": 0.1404,
+      "step": 29618
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9251264895573412e-05,
+      "loss": 0.1347,
+      "step": 29619
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9251202520994226e-05,
+      "loss": 0.1086,
+      "step": 29620
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.92511401439181e-05,
+      "loss": 0.1427,
+      "step": 29621
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.925107776434505e-05,
+      "loss": 0.1142,
+      "step": 29622
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9251015382275083e-05,
+      "loss": 0.1632,
+      "step": 29623
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.925095299770823e-05,
+      "loss": 0.131,
+      "step": 29624
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250890610644498e-05,
+      "loss": 0.1427,
+      "step": 29625
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250828221083912e-05,
+      "loss": 0.0996,
+      "step": 29626
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250765829026483e-05,
+      "loss": 0.1621,
+      "step": 29627
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250703434472228e-05,
+      "loss": 0.1452,
+      "step": 29628
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250641037421164e-05,
+      "loss": 0.1121,
+      "step": 29629
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.925057863787331e-05,
+      "loss": 0.1538,
+      "step": 29630
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.925051623582868e-05,
+      "loss": 0.1456,
+      "step": 29631
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250453831287292e-05,
+      "loss": 0.0998,
+      "step": 29632
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250391424249164e-05,
+      "loss": 0.1152,
+      "step": 29633
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250329014714314e-05,
+      "loss": 0.155,
+      "step": 29634
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250266602682758e-05,
+      "loss": 0.1344,
+      "step": 29635
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250204188154507e-05,
+      "loss": 0.1432,
+      "step": 29636
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250141771129586e-05,
+      "loss": 0.1407,
+      "step": 29637
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250079351608008e-05,
+      "loss": 0.1142,
+      "step": 29638
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9250016929589787e-05,
+      "loss": 0.1443,
+      "step": 29639
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924995450507495e-05,
+      "loss": 0.1214,
+      "step": 29640
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.92498920780635e-05,
+      "loss": 0.1391,
+      "step": 29641
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249829648555468e-05,
+      "loss": 0.1319,
+      "step": 29642
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249767216550857e-05,
+      "loss": 0.1226,
+      "step": 29643
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249704782049695e-05,
+      "loss": 0.1336,
+      "step": 29644
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924964234505199e-05,
+      "loss": 0.1306,
+      "step": 29645
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249579905557766e-05,
+      "loss": 0.1006,
+      "step": 29646
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249517463567036e-05,
+      "loss": 0.141,
+      "step": 29647
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924945501907982e-05,
+      "loss": 0.1473,
+      "step": 29648
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249392572096136e-05,
+      "loss": 0.1335,
+      "step": 29649
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249330122615992e-05,
+      "loss": 0.1246,
+      "step": 29650
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249267670639412e-05,
+      "loss": 0.1259,
+      "step": 29651
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249205216166412e-05,
+      "loss": 0.1135,
+      "step": 29652
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924914275919701e-05,
+      "loss": 0.139,
+      "step": 29653
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249080299731218e-05,
+      "loss": 0.1338,
+      "step": 29654
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9249017837769058e-05,
+      "loss": 0.1257,
+      "step": 29655
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9248955373310542e-05,
+      "loss": 0.1592,
+      "step": 29656
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9248892906355695e-05,
+      "loss": 0.1267,
+      "step": 29657
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924883043690452e-05,
+      "loss": 0.1049,
+      "step": 29658
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924876796495705e-05,
+      "loss": 0.1381,
+      "step": 29659
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9248705490513295e-05,
+      "loss": 0.1407,
+      "step": 29660
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9248643013573268e-05,
+      "loss": 0.1304,
+      "step": 29661
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924858053413699e-05,
+      "loss": 0.1467,
+      "step": 29662
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9248518052204473e-05,
+      "loss": 0.1244,
+      "step": 29663
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9248455567775743e-05,
+      "loss": 0.1454,
+      "step": 29664
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924839308085081e-05,
+      "loss": 0.1324,
+      "step": 29665
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924833059142969e-05,
+      "loss": 0.1406,
+      "step": 29666
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9248268099512405e-05,
+      "loss": 0.1322,
+      "step": 29667
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924820560509897e-05,
+      "loss": 0.1337,
+      "step": 29668
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9248143108189398e-05,
+      "loss": 0.1323,
+      "step": 29669
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924808060878371e-05,
+      "loss": 0.1353,
+      "step": 29670
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9248018106881923e-05,
+      "loss": 0.1224,
+      "step": 29671
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9247955602484052e-05,
+      "loss": 0.1236,
+      "step": 29672
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9247893095590114e-05,
+      "loss": 0.1308,
+      "step": 29673
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9247830586200126e-05,
+      "loss": 0.1213,
+      "step": 29674
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924776807431411e-05,
+      "loss": 0.1357,
+      "step": 29675
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9247705559932074e-05,
+      "loss": 0.1407,
+      "step": 29676
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9247643043054037e-05,
+      "loss": 0.1176,
+      "step": 29677
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9247580523680024e-05,
+      "loss": 0.1473,
+      "step": 29678
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924751800181004e-05,
+      "loss": 0.1155,
+      "step": 29679
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924745547744411e-05,
+      "loss": 0.1362,
+      "step": 29680
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924739295058225e-05,
+      "loss": 0.1323,
+      "step": 29681
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9247330421224474e-05,
+      "loss": 0.1211,
+      "step": 29682
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.92472678893708e-05,
+      "loss": 0.1147,
+      "step": 29683
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9247205355021244e-05,
+      "loss": 0.1286,
+      "step": 29684
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9247142818175825e-05,
+      "loss": 0.1559,
+      "step": 29685
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924708027883456e-05,
+      "loss": 0.1051,
+      "step": 29686
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9247017736997466e-05,
+      "loss": 0.1433,
+      "step": 29687
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246955192664556e-05,
+      "loss": 0.1412,
+      "step": 29688
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246892645835848e-05,
+      "loss": 0.1286,
+      "step": 29689
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246830096511363e-05,
+      "loss": 0.1243,
+      "step": 29690
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246767544691116e-05,
+      "loss": 0.1294,
+      "step": 29691
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924670499037512e-05,
+      "loss": 0.139,
+      "step": 29692
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.92466424335634e-05,
+      "loss": 0.1249,
+      "step": 29693
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246579874255965e-05,
+      "loss": 0.1228,
+      "step": 29694
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246517312452835e-05,
+      "loss": 0.112,
+      "step": 29695
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246454748154028e-05,
+      "loss": 0.1447,
+      "step": 29696
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246392181359557e-05,
+      "loss": 0.1052,
+      "step": 29697
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924632961206944e-05,
+      "loss": 0.1538,
+      "step": 29698
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246267040283703e-05,
+      "loss": 0.129,
+      "step": 29699
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246204466002352e-05,
+      "loss": 0.1565,
+      "step": 29700
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246141889225404e-05,
+      "loss": 0.1182,
+      "step": 29701
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9246079309952884e-05,
+      "loss": 0.1509,
+      "step": 29702
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.92460167281848e-05,
+      "loss": 0.1223,
+      "step": 29703
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245954143921173e-05,
+      "loss": 0.1079,
+      "step": 29704
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245891557162023e-05,
+      "loss": 0.1249,
+      "step": 29705
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245828967907362e-05,
+      "loss": 0.1424,
+      "step": 29706
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245766376157206e-05,
+      "loss": 0.1377,
+      "step": 29707
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245703781911576e-05,
+      "loss": 0.1238,
+      "step": 29708
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.924564118517049e-05,
+      "loss": 0.1579,
+      "step": 29709
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245578585933958e-05,
+      "loss": 0.1442,
+      "step": 29710
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245515984202003e-05,
+      "loss": 0.1465,
+      "step": 29711
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245453379974642e-05,
+      "loss": 0.1672,
+      "step": 29712
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245390773251888e-05,
+      "loss": 0.1129,
+      "step": 29713
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245328164033758e-05,
+      "loss": 0.12,
+      "step": 29714
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245265552320272e-05,
+      "loss": 0.1461,
+      "step": 29715
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245202938111448e-05,
+      "loss": 0.1336,
+      "step": 29716
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245140321407298e-05,
+      "loss": 0.1325,
+      "step": 29717
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9245077702207844e-05,
+      "loss": 0.1172,
+      "step": 29718
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9245015080513098e-05,
+      "loss": 0.1194,
+      "step": 29719
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244952456323078e-05,
+      "loss": 0.139,
+      "step": 29720
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244889829637804e-05,
+      "loss": 0.1194,
+      "step": 29721
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244827200457292e-05,
+      "loss": 0.1129,
+      "step": 29722
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244764568781557e-05,
+      "loss": 0.1251,
+      "step": 29723
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244701934610616e-05,
+      "loss": 0.1242,
+      "step": 29724
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924463929794449e-05,
+      "loss": 0.1472,
+      "step": 29725
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244576658783188e-05,
+      "loss": 0.1559,
+      "step": 29726
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244514017126732e-05,
+      "loss": 0.1144,
+      "step": 29727
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924445137297514e-05,
+      "loss": 0.128,
+      "step": 29728
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924438872632843e-05,
+      "loss": 0.1116,
+      "step": 29729
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244326077186612e-05,
+      "loss": 0.122,
+      "step": 29730
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924426342554971e-05,
+      "loss": 0.1616,
+      "step": 29731
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244200771417734e-05,
+      "loss": 0.1174,
+      "step": 29732
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244138114790706e-05,
+      "loss": 0.1666,
+      "step": 29733
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244075455668648e-05,
+      "loss": 0.1196,
+      "step": 29734
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9244012794051565e-05,
+      "loss": 0.1339,
+      "step": 29735
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9243950129939483e-05,
+      "loss": 0.1463,
+      "step": 29736
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924388746333241e-05,
+      "loss": 0.1358,
+      "step": 29737
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9243824794230376e-05,
+      "loss": 0.1357,
+      "step": 29738
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9243762122633386e-05,
+      "loss": 0.1048,
+      "step": 29739
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9243699448541463e-05,
+      "loss": 0.1378,
+      "step": 29740
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9243636771954622e-05,
+      "loss": 0.127,
+      "step": 29741
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924357409287288e-05,
+      "loss": 0.143,
+      "step": 29742
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924351141129625e-05,
+      "loss": 0.1273,
+      "step": 29743
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924344872722476e-05,
+      "loss": 0.1313,
+      "step": 29744
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9243386040658416e-05,
+      "loss": 0.149,
+      "step": 29745
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924332335159724e-05,
+      "loss": 0.1306,
+      "step": 29746
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924326066004125e-05,
+      "loss": 0.1388,
+      "step": 29747
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9243197965990455e-05,
+      "loss": 0.134,
+      "step": 29748
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9243135269444885e-05,
+      "loss": 0.1516,
+      "step": 29749
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9243072570404545e-05,
+      "loss": 0.1378,
+      "step": 29750
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9243009868869456e-05,
+      "loss": 0.12,
+      "step": 29751
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242947164839635e-05,
+      "loss": 0.1182,
+      "step": 29752
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242884458315102e-05,
+      "loss": 0.1485,
+      "step": 29753
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924282174929587e-05,
+      "loss": 0.1213,
+      "step": 29754
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242759037781958e-05,
+      "loss": 0.1323,
+      "step": 29755
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242696323773384e-05,
+      "loss": 0.1191,
+      "step": 29756
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924263360727016e-05,
+      "loss": 0.1478,
+      "step": 29757
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242570888272304e-05,
+      "loss": 0.1702,
+      "step": 29758
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242508166779838e-05,
+      "loss": 0.1209,
+      "step": 29759
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242445442792775e-05,
+      "loss": 0.1272,
+      "step": 29760
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242382716311137e-05,
+      "loss": 0.1403,
+      "step": 29761
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924231998733493e-05,
+      "loss": 0.1453,
+      "step": 29762
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242257255864184e-05,
+      "loss": 0.1352,
+      "step": 29763
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242194521898904e-05,
+      "loss": 0.1454,
+      "step": 29764
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9242131785439115e-05,
+      "loss": 0.1406,
+      "step": 29765
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924206904648483e-05,
+      "loss": 0.1341,
+      "step": 29766
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924200630503607e-05,
+      "loss": 0.1294,
+      "step": 29767
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924194356109285e-05,
+      "loss": 0.1099,
+      "step": 29768
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9241880814655182e-05,
+      "loss": 0.1727,
+      "step": 29769
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924181806572309e-05,
+      "loss": 0.1063,
+      "step": 29770
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9241755314296586e-05,
+      "loss": 0.1459,
+      "step": 29771
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924169256037569e-05,
+      "loss": 0.1264,
+      "step": 29772
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924162980396042e-05,
+      "loss": 0.1278,
+      "step": 29773
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9241567045050788e-05,
+      "loss": 0.1109,
+      "step": 29774
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9241504283646814e-05,
+      "loss": 0.1411,
+      "step": 29775
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9241441519748518e-05,
+      "loss": 0.1547,
+      "step": 29776
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924137875335591e-05,
+      "loss": 0.141,
+      "step": 29777
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9241315984469013e-05,
+      "loss": 0.1158,
+      "step": 29778
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924125321308784e-05,
+      "loss": 0.1297,
+      "step": 29779
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924119043921241e-05,
+      "loss": 0.145,
+      "step": 29780
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924112766284274e-05,
+      "loss": 0.1181,
+      "step": 29781
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9241064883978847e-05,
+      "loss": 0.1232,
+      "step": 29782
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924100210262075e-05,
+      "loss": 0.1369,
+      "step": 29783
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9240939318768457e-05,
+      "loss": 0.1373,
+      "step": 29784
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9240876532421995e-05,
+      "loss": 0.096,
+      "step": 29785
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9240813743581377e-05,
+      "loss": 0.1358,
+      "step": 29786
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924075095224662e-05,
+      "loss": 0.1219,
+      "step": 29787
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924068815841774e-05,
+      "loss": 0.1367,
+      "step": 29788
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9240625362094755e-05,
+      "loss": 0.1204,
+      "step": 29789
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9240562563277682e-05,
+      "loss": 0.1638,
+      "step": 29790
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924049976196654e-05,
+      "loss": 0.1486,
+      "step": 29791
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9240436958161344e-05,
+      "loss": 0.1259,
+      "step": 29792
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924037415186211e-05,
+      "loss": 0.1233,
+      "step": 29793
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9240311343068855e-05,
+      "loss": 0.1418,
+      "step": 29794
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9240248531781596e-05,
+      "loss": 0.1421,
+      "step": 29795
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9240185718000354e-05,
+      "loss": 0.1253,
+      "step": 29796
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.924012290172514e-05,
+      "loss": 0.1465,
+      "step": 29797
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9240060082955977e-05,
+      "loss": 0.1126,
+      "step": 29798
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239997261692873e-05,
+      "loss": 0.1558,
+      "step": 29799
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239934437935856e-05,
+      "loss": 0.129,
+      "step": 29800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239871611684934e-05,
+      "loss": 0.1275,
+      "step": 29801
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.923980878294013e-05,
+      "loss": 0.1487,
+      "step": 29802
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239745951701457e-05,
+      "loss": 0.135,
+      "step": 29803
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.923968311796893e-05,
+      "loss": 0.1502,
+      "step": 29804
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.923962028174257e-05,
+      "loss": 0.1338,
+      "step": 29805
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.92395574430224e-05,
+      "loss": 0.1079,
+      "step": 29806
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239494601808426e-05,
+      "loss": 0.1534,
+      "step": 29807
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.923943175810067e-05,
+      "loss": 0.1387,
+      "step": 29808
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239368911899147e-05,
+      "loss": 0.1281,
+      "step": 29809
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239306063203874e-05,
+      "loss": 0.1421,
+      "step": 29810
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239243212014872e-05,
+      "loss": 0.1003,
+      "step": 29811
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239180358332154e-05,
+      "loss": 0.1866,
+      "step": 29812
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239117502155738e-05,
+      "loss": 0.164,
+      "step": 29813
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9239054643485642e-05,
+      "loss": 0.1276,
+      "step": 29814
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.923899178232188e-05,
+      "loss": 0.1454,
+      "step": 29815
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.923892891866447e-05,
+      "loss": 0.1335,
+      "step": 29816
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9238866052513434e-05,
+      "loss": 0.1176,
+      "step": 29817
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923880318386878e-05,
+      "loss": 0.1294,
+      "step": 29818
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9238740312730536e-05,
+      "loss": 0.1407,
+      "step": 29819
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9238677439098708e-05,
+      "loss": 0.1305,
+      "step": 29820
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923861456297332e-05,
+      "loss": 0.15,
+      "step": 29821
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9238551684354383e-05,
+      "loss": 0.1174,
+      "step": 29822
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9238488803241923e-05,
+      "loss": 0.1373,
+      "step": 29823
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9238425919635946e-05,
+      "loss": 0.1358,
+      "step": 29824
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923836303353648e-05,
+      "loss": 0.1218,
+      "step": 29825
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9238300144943534e-05,
+      "loss": 0.1349,
+      "step": 29826
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923823725385713e-05,
+      "loss": 0.1218,
+      "step": 29827
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923817436027728e-05,
+      "loss": 0.1289,
+      "step": 29828
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9238111464204004e-05,
+      "loss": 0.1258,
+      "step": 29829
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923804856563732e-05,
+      "loss": 0.1697,
+      "step": 29830
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9237985664577243e-05,
+      "loss": 0.1227,
+      "step": 29831
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9237922761023786e-05,
+      "loss": 0.1471,
+      "step": 29832
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9237859854976973e-05,
+      "loss": 0.1291,
+      "step": 29833
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923779694643682e-05,
+      "loss": 0.1471,
+      "step": 29834
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9237734035403345e-05,
+      "loss": 0.1676,
+      "step": 29835
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9237671121876557e-05,
+      "loss": 0.1429,
+      "step": 29836
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9237608205856483e-05,
+      "loss": 0.1397,
+      "step": 29837
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923754528734313e-05,
+      "loss": 0.1397,
+      "step": 29838
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9237482366336526e-05,
+      "loss": 0.1551,
+      "step": 29839
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923741944283668e-05,
+      "loss": 0.1702,
+      "step": 29840
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923735651684361e-05,
+      "loss": 0.1131,
+      "step": 29841
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9237293588357336e-05,
+      "loss": 0.1228,
+      "step": 29842
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923723065737787e-05,
+      "loss": 0.1292,
+      "step": 29843
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923716772390524e-05,
+      "loss": 0.1562,
+      "step": 29844
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923710478793945e-05,
+      "loss": 0.1497,
+      "step": 29845
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923704184948052e-05,
+      "loss": 0.1574,
+      "step": 29846
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9236978908528473e-05,
+      "loss": 0.1317,
+      "step": 29847
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923691596508332e-05,
+      "loss": 0.1064,
+      "step": 29848
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923685301914508e-05,
+      "loss": 0.1353,
+      "step": 29849
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9236790070713774e-05,
+      "loss": 0.1282,
+      "step": 29850
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923672711978941e-05,
+      "loss": 0.1354,
+      "step": 29851
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9236664166372012e-05,
+      "loss": 0.1305,
+      "step": 29852
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.92366012104616e-05,
+      "loss": 0.1357,
+      "step": 29853
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923653825205818e-05,
+      "loss": 0.1216,
+      "step": 29854
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9236475291161775e-05,
+      "loss": 0.1454,
+      "step": 29855
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9236412327772406e-05,
+      "loss": 0.1134,
+      "step": 29856
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9236349361890085e-05,
+      "loss": 0.1238,
+      "step": 29857
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923628639351483e-05,
+      "loss": 0.1454,
+      "step": 29858
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9236223422646656e-05,
+      "loss": 0.1351,
+      "step": 29859
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9236160449285582e-05,
+      "loss": 0.1158,
+      "step": 29860
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923609747343163e-05,
+      "loss": 0.124,
+      "step": 29861
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9236034495084808e-05,
+      "loss": 0.1499,
+      "step": 29862
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923597151424514e-05,
+      "loss": 0.1558,
+      "step": 29863
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9235908530912637e-05,
+      "loss": 0.1698,
+      "step": 29864
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923584554508732e-05,
+      "loss": 0.1271,
+      "step": 29865
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9235782556769207e-05,
+      "loss": 0.1202,
+      "step": 29866
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923571956595831e-05,
+      "loss": 0.1269,
+      "step": 29867
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923565657265465e-05,
+      "loss": 0.1511,
+      "step": 29868
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9235593576858246e-05,
+      "loss": 0.1225,
+      "step": 29869
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9235530578569112e-05,
+      "loss": 0.1336,
+      "step": 29870
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923546757778726e-05,
+      "loss": 0.1239,
+      "step": 29871
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9235404574512717e-05,
+      "loss": 0.1171,
+      "step": 29872
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923534156874549e-05,
+      "loss": 0.1741,
+      "step": 29873
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9235278560485606e-05,
+      "loss": 0.1151,
+      "step": 29874
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9235215549733076e-05,
+      "loss": 0.1313,
+      "step": 29875
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9235152536487918e-05,
+      "loss": 0.1291,
+      "step": 29876
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923508952075015e-05,
+      "loss": 0.142,
+      "step": 29877
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9235026502519788e-05,
+      "loss": 0.1481,
+      "step": 29878
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234963481796847e-05,
+      "loss": 0.1088,
+      "step": 29879
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923490045858135e-05,
+      "loss": 0.1332,
+      "step": 29880
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234837432873307e-05,
+      "loss": 0.1295,
+      "step": 29881
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234774404672735e-05,
+      "loss": 0.1607,
+      "step": 29882
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923471137397966e-05,
+      "loss": 0.1215,
+      "step": 29883
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234648340794093e-05,
+      "loss": 0.1319,
+      "step": 29884
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234585305116046e-05,
+      "loss": 0.1653,
+      "step": 29885
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234522266945544e-05,
+      "loss": 0.136,
+      "step": 29886
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234459226282604e-05,
+      "loss": 0.149,
+      "step": 29887
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234396183127238e-05,
+      "loss": 0.1644,
+      "step": 29888
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234333137479465e-05,
+      "loss": 0.1512,
+      "step": 29889
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234270089339304e-05,
+      "loss": 0.1462,
+      "step": 29890
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234207038706772e-05,
+      "loss": 0.1223,
+      "step": 29891
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9234143985581883e-05,
+      "loss": 0.1427,
+      "step": 29892
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923408092996465e-05,
+      "loss": 0.1319,
+      "step": 29893
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.92340178718551e-05,
+      "loss": 0.1449,
+      "step": 29894
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233954811253246e-05,
+      "loss": 0.1375,
+      "step": 29895
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233891748159104e-05,
+      "loss": 0.1438,
+      "step": 29896
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923382868257269e-05,
+      "loss": 0.1429,
+      "step": 29897
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233765614494024e-05,
+      "loss": 0.1452,
+      "step": 29898
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923370254392312e-05,
+      "loss": 0.1184,
+      "step": 29899
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233639470859996e-05,
+      "loss": 0.1275,
+      "step": 29900
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233576395304672e-05,
+      "loss": 0.118,
+      "step": 29901
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233513317257163e-05,
+      "loss": 0.1527,
+      "step": 29902
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233450236717484e-05,
+      "loss": 0.1296,
+      "step": 29903
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233387153685654e-05,
+      "loss": 0.1282,
+      "step": 29904
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233324068161687e-05,
+      "loss": 0.1187,
+      "step": 29905
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233260980145606e-05,
+      "loss": 0.1211,
+      "step": 29906
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233197889637423e-05,
+      "loss": 0.1308,
+      "step": 29907
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233134796637157e-05,
+      "loss": 0.1233,
+      "step": 29908
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233071701144825e-05,
+      "loss": 0.1222,
+      "step": 29909
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9233008603160444e-05,
+      "loss": 0.1381,
+      "step": 29910
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9232945502684032e-05,
+      "loss": 0.1531,
+      "step": 29911
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9232882399715604e-05,
+      "loss": 0.1447,
+      "step": 29912
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923281929425518e-05,
+      "loss": 0.1332,
+      "step": 29913
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.923275618630277e-05,
+      "loss": 0.1237,
+      "step": 29914
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9232693075858398e-05,
+      "loss": 0.1523,
+      "step": 29915
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923262996292208e-05,
+      "loss": 0.1267,
+      "step": 29916
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923256684749383e-05,
+      "loss": 0.1275,
+      "step": 29917
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923250372957367e-05,
+      "loss": 0.1432,
+      "step": 29918
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923244060916161e-05,
+      "loss": 0.141,
+      "step": 29919
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9232377486257677e-05,
+      "loss": 0.1093,
+      "step": 29920
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9232314360861877e-05,
+      "loss": 0.1491,
+      "step": 29921
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9232251232974235e-05,
+      "loss": 0.1197,
+      "step": 29922
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9232188102594764e-05,
+      "loss": 0.1126,
+      "step": 29923
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923212496972348e-05,
+      "loss": 0.132,
+      "step": 29924
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9232061834360408e-05,
+      "loss": 0.181,
+      "step": 29925
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9231998696505553e-05,
+      "loss": 0.1213,
+      "step": 29926
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9231935556158943e-05,
+      "loss": 0.1341,
+      "step": 29927
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923187241332059e-05,
+      "loss": 0.1717,
+      "step": 29928
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923180926799051e-05,
+      "loss": 0.1733,
+      "step": 29929
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923174612016872e-05,
+      "loss": 0.1204,
+      "step": 29930
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9231682969855244e-05,
+      "loss": 0.107,
+      "step": 29931
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9231619817050088e-05,
+      "loss": 0.1126,
+      "step": 29932
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9231556661753276e-05,
+      "loss": 0.1147,
+      "step": 29933
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9231493503964825e-05,
+      "loss": 0.1345,
+      "step": 29934
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923143034368475e-05,
+      "loss": 0.1169,
+      "step": 29935
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923136718091307e-05,
+      "loss": 0.1491,
+      "step": 29936
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9231304015649802e-05,
+      "loss": 0.1373,
+      "step": 29937
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9231240847894955e-05,
+      "loss": 0.1615,
+      "step": 29938
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923117767764856e-05,
+      "loss": 0.1155,
+      "step": 29939
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9231114504910625e-05,
+      "loss": 0.1143,
+      "step": 29940
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9231051329681166e-05,
+      "loss": 0.1253,
+      "step": 29941
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230988151960208e-05,
+      "loss": 0.1195,
+      "step": 29942
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230924971747762e-05,
+      "loss": 0.1426,
+      "step": 29943
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230861789043844e-05,
+      "loss": 0.1558,
+      "step": 29944
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230798603848474e-05,
+      "loss": 0.1158,
+      "step": 29945
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923073541616167e-05,
+      "loss": 0.1392,
+      "step": 29946
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230672225983446e-05,
+      "loss": 0.1047,
+      "step": 29947
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923060903331382e-05,
+      "loss": 0.14,
+      "step": 29948
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923054583815281e-05,
+      "loss": 0.149,
+      "step": 29949
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230482640500435e-05,
+      "loss": 0.1488,
+      "step": 29950
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.923041944035671e-05,
+      "loss": 0.1201,
+      "step": 29951
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230356237721644e-05,
+      "loss": 0.1282,
+      "step": 29952
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230293032595266e-05,
+      "loss": 0.1361,
+      "step": 29953
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230229824977592e-05,
+      "loss": 0.1382,
+      "step": 29954
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230166614868634e-05,
+      "loss": 0.1344,
+      "step": 29955
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230103402268407e-05,
+      "loss": 0.1438,
+      "step": 29956
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9230040187176934e-05,
+      "loss": 0.1488,
+      "step": 29957
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229976969594232e-05,
+      "loss": 0.1037,
+      "step": 29958
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229913749520315e-05,
+      "loss": 0.1576,
+      "step": 29959
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229850526955202e-05,
+      "loss": 0.1256,
+      "step": 29960
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229787301898908e-05,
+      "loss": 0.1264,
+      "step": 29961
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229724074351453e-05,
+      "loss": 0.1486,
+      "step": 29962
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.922966084431285e-05,
+      "loss": 0.1092,
+      "step": 29963
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229597611783117e-05,
+      "loss": 0.1376,
+      "step": 29964
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229534376762277e-05,
+      "loss": 0.1122,
+      "step": 29965
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229471139250343e-05,
+      "loss": 0.1139,
+      "step": 29966
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229407899247326e-05,
+      "loss": 0.1073,
+      "step": 29967
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229344656753253e-05,
+      "loss": 0.1407,
+      "step": 29968
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229281411768134e-05,
+      "loss": 0.1248,
+      "step": 29969
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.922921816429199e-05,
+      "loss": 0.1359,
+      "step": 29970
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229154914324836e-05,
+      "loss": 0.1284,
+      "step": 29971
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229091661866694e-05,
+      "loss": 0.1322,
+      "step": 29972
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9229028406917574e-05,
+      "loss": 0.1238,
+      "step": 29973
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228965149477496e-05,
+      "loss": 0.1134,
+      "step": 29974
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228901889546474e-05,
+      "loss": 0.1517,
+      "step": 29975
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228838627124534e-05,
+      "loss": 0.1114,
+      "step": 29976
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.922877536221168e-05,
+      "loss": 0.1222,
+      "step": 29977
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228712094807944e-05,
+      "loss": 0.1356,
+      "step": 29978
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.922864882491333e-05,
+      "loss": 0.1533,
+      "step": 29979
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.922858555252787e-05,
+      "loss": 0.1257,
+      "step": 29980
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.922852227765156e-05,
+      "loss": 0.1105,
+      "step": 29981
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228459000284435e-05,
+      "loss": 0.1067,
+      "step": 29982
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228395720426506e-05,
+      "loss": 0.1354,
+      "step": 29983
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228332438077788e-05,
+      "loss": 0.1161,
+      "step": 29984
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228269153238298e-05,
+      "loss": 0.1604,
+      "step": 29985
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228205865908057e-05,
+      "loss": 0.1484,
+      "step": 29986
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228142576087082e-05,
+      "loss": 0.1425,
+      "step": 29987
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228079283775386e-05,
+      "loss": 0.1293,
+      "step": 29988
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9228015988972986e-05,
+      "loss": 0.1217,
+      "step": 29989
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9227952691679907e-05,
+      "loss": 0.1643,
+      "step": 29990
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9227889391896157e-05,
+      "loss": 0.1283,
+      "step": 29991
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9227826089621758e-05,
+      "loss": 0.1366,
+      "step": 29992
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9227762784856723e-05,
+      "loss": 0.1478,
+      "step": 29993
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9227699477601072e-05,
+      "loss": 0.1341,
+      "step": 29994
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9227636167854826e-05,
+      "loss": 0.1332,
+      "step": 29995
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9227572855617994e-05,
+      "loss": 0.1314,
+      "step": 29996
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9227509540890598e-05,
+      "loss": 0.1193,
+      "step": 29997
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9227446223672654e-05,
+      "loss": 0.1147,
+      "step": 29998
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9227382903964175e-05,
+      "loss": 0.1387,
+      "step": 29999
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.922731958176519e-05,
+      "loss": 0.1172,
+      "step": 30000
+    }
+  ],
+  "max_steps": 197140,
+  "num_train_epochs": 20,
+  "total_flos": 1.176095624085373e+19,
+  "trial_name": null,
+  "trial_params": null
+}